1

Far-red sun-induced chlorophyll fluorescence shows ecosystem-specific relationships to 1

gross primary production An assessment based on observational and modeling 2

approaches 3

4

1Damm A 2Guanter L 3Paul-Limoges E 4van der Tol C 1Hueni A 3Buchmann N5

3Eugster W 5Ammann C 1Schaepman ME 6

1 Remote Sensing Laboratories University of Zurich Winterthurerstrasse 190 8057 Zurich 7

Switzerland 8

2 German Research Centre for Geosciences (GFZ) Remote Sensing Section Telegrafenberg 9

14473 Potsdam Germany 10

3 Institute of Agricultural Sciences ETH Zurich Universitaetsstrasse 2 8092 Zurich 11

Switzerland 12

4 University of Twente Faculty of Geo-Information Science and Earth Observation (ITC) 13

PO Box 217 7500 AE Enschede The Netherlands 14

5 Research Station Agroscope Reckenholz-Taumlnikon ART Reckenholzstrasse 191 8046 15

Zurich Switzerland 16

17

Corresponding author Tel +41 44 635 5251 E-mail alexanderdammgeouzhch 18

19

Abstract 20

Sun-induced chlorophyll fluorescence (SIF) is a radiation flux emitted from 21

chlorophyll molecules and is considered an indicator of the actual functional state of plant 22

photosynthesis The remote measurement of SIF opens a new perspective to assess actual 23

2

photosynthesis at ecologically relevant larger scales and provides an alternative approach to 24

study the terrestrial carbon cycle Recent studies demonstrated the reliability of measured SIF 25

signals and showed significant relationships between SIF and gross primary production (GPP) 26

at ecosystem and global scales Despite these encouraging results understanding the complex 27

mechanisms between SIF and GPP remains challenging before SIF can be finally utilized to 28

constrain estimates of GPP In this study we present a comprehensive assessment of the 29

relationship between far red SIF retrieved at 760 nm (SIF760) and GPP and its transferability 30

across three structurally and physiologically contrasting ecosystems perennial grassland 31

cropland and mixed temperate forest We use multi-temporal imaging spectroscopy (IS) data 32

acquired with the Airborne Prism Experiment (APEX) sensor as well as eddy covariance (EC) 33

flux tower data to evaluate the relationship between SIF760 and GPPEC We use simulations 34

performed with the coupled photosynthesis-fluorescence model SCOPE to prove trends 35

obtained from our observational data and assess apparent confounding factors such as 36

physiological and structural interferences or temporal scaling effects Observed relationships 37

between SIF760 and GPPEC were asymptotic and ecosystem-specific ie perennial grassland 38

(R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and mixed temperate 39

forest (R2 = 048 rRMSE = 1588) We demonstrate that asymptotic leaf level relationships 40

between SIF760-GPPEC became more linear at canopy level and scaled with temporal 41

aggregation We conclude that remote sensing of SIF provides a new observational approach 42

to decrease uncertainties in estimating GPP across ecosystems but requires dedicated 43

strategies to compensate for the various confounding factors impacting SIF-GPP 44

relationships Our findings help in bridging the gap between mechanistic understanding at leaf 45

level and ecosystem-specific observations of the relationships between SIF and GPP 46

47

Keywords Sun-induced chlorophyll fluorescence gross primary production airborne-based 48

spectroscopy APEX Fraunhofer line depth (FLD) eddy-covariance SCOPE 49

3

1 Introduction 50

Plant photosynthesis is a key process in terrestrial ecosystems mediating gas and 51

energy exchanges in the atmosphere-biosphere system (Baldocchi et al 2001 Ozanne et al 52

2003) Products of photosynthesis provide a wealth of ecosystem services that are essential 53

for human well-being including food fiber energy and oxygen (Imhoff et al 2004 54

Krausmann et al 2013 Schroter et al 2005) Photosynthesis as the underlying process for 55

plant growth is a particularly interesting indicator of crop efficiency and agricultural 56

management practices (Falloon and Betts 2010 Guanter et al 2014 Trnka et al 2004) both 57

of which having important implications for yield forecasts and for the implementation of 58

climate change adaptation strategies (IPCC 2013) 59

Observing the highly dynamic process of photosynthesis beyond the individual leaf or 60

plant levels in-situ is based on measuring the carbon dioxide (CO2) exchange between 61

vegetation and atmosphere with eddy-covariance (EC) flux towers and partitioning it into 62

gross primary production (GPP) and ecosystem respiration (Baldocchi et al 2001) 63

Measurements of plant - light interactions using spectrometers installed on for example EC 64

towers allow deriving information about the pigment status and provide an alternative 65

approach to estimate GPP (Balzarolo et al 2011 Gamon et al 2010 Hilker et al 2011) At 66

landscape scale photosynthesis can be assessed using process-based models (Sitch et al 67

2003) greenness-based satellite observations (Running et al 2004) or hybrid approaches 68

combining in-situ observations and statistical modeling (Jung et al 2011) All these 69

approaches provide important insights to study photosynthesis but usually do not allow 70

assessing photosynthesis at larger scales while preserving the high spatial variability present 71

in ecosystems EC flux tower measurements represent only smaller areas in preselected 72

ecosystems (Drolet et al 2008 Turner et al 2005) are not spatially distributed according to 73

carbon stocks (Schimel et al 2014) and do not allow spatial differentiations within the 74

4

measured footprint (Barcza et al 2009 Kljun et al 2002) Combined large scale modeling 75

and observational approaches based on vegetation greenness are spatially contiguous but face 76

the complexity of naturally varying systems including diverse interactions and complex 77

feedbacks which limit their predictive capabilities (Beer et al 2010 Goetz and Prince 1999 78

Turner et al 2005) 79

Over the last decade significant progress has been made in measuring plant-light 80

interactions and the process of photosynthesis Remote measurements of sun-induced 81

chlorophyll fluorescence (SIF) in particular open a new perspective to assess photosynthesis 82

at ecosystem scale SIF is a radiation flux emitted from plant chlorophyll molecules a few 83

nanoseconds after light absorption in the wavelength range from 600-800 nm and is 84

considered an indicator for the functional status of actual plant photosynthesis (Baker 2008) 85

Various studies demonstrated the possibility to measure SIF at certain wavelengths on ground 86

(Guanter et al 2013 Rascher et al 2009) from airborne platforms (Damm et al 2014 87

Guanter et al 2007 Zarco-Tejada et al 2012) and from satellites (Frankenberg et al 2011 88

Guanter et al 2014 Joiner et al 2013) Recent research demonstrated SIF being sensitive to 89

changes in photosynthesis showing strong links to GPP at the level of leaves (Meroni et al 90

2008 Middleton et al 2002) plants (Damm et al 2010 Rossini et al 2010) canopies (Zarco-91

Tejada et al 2013) and ecosystems (Frankenberg et al 2011 Guanter et al 2012) 92

Observed relationships between SIF and GPP are conceptually explained using an 93

approximation of GPP based on Monteithrsquos light use efficiency concept (Monteith 1972) 94

∙ (1) 95

where APAR is the absorbed photosynthetically active radiation expressed in radiance units 96

and LUEp is the efficiency of light utilization for photosynthesis and allows converting 97

measured radiances into the number of fixed CO2 molecules SIF is expressed by expanding 98

the GPP notation in Eq (1) following Guanter et al (2014) 99

13 ∙ ∙ (2) 100

5

where LUEf is the light use efficiency of SIF (fluorescence yield) and fesc accounts for a 101

structural interference determining the fraction of SIF photons escaping the canopy 102

Relationships between SIF and GPP are mostly driven by the common APAR term In 103

addition a covariance between both light use efficiencies LUEp and LUEf is expected to 104

occur in absence of the confounding impact of other protective mechanisms (Damm et al 105

2010 Guanter et al 2014) 106

The above outlined concept relating SIF and GPP simplifies a complex set of 107

underlying mechanisms and violation of any assumptions made will directly confound the 108

SIF-GPP relationship In particular the competition of three processes for de-exciting 109

absorbed light energy ie photochemistry radiative energy loss (SIF) and non-radiative 110

energy dissipation (commonly approximated as non-photochemical quenching NPQ) causes 111

complex and changing sensitivities of emitted SIF to actual rates of photosynthesis (Porcar-112

Castell et al 2014 van der Tol et al 2014 van der Tol et al 2009a) This directly implies that 113

the functional link between SIF and GPP depends on the rate of NPQ and consequently on 114

ambient stress levels At canopy scale the three-dimensional structure causes gradients in 115

light interception and light quality within canopies (Nobel et al 1993 Stewart et al 2003) 116

additionally altering the rate of NPQ (Demmig-Adams 1998 Niinemets et al 2003) and thus 117

impacting the SIF-GPP relationship Canopy structure also increases the probability for the 118

emitted SIF photons either to be re-absorbed by chlorophyll or to escape the canopy (Fournier 119

et al 2012 Knyazikhin et al 2013) to some extent violating the assumption of fesc to be 120

constant In addition to these structural and physiological effects variations in SIF signals 121

caused by for example instrumental (Damm et al 2011) or atmospheric effects (Damm et al 122

2014 Guanter et al 2010) and retrieval uncertainties related to the estimation of surface 123

irradiance (Damm et al 2015) can potentially affect the apparent relationship between SIF 124

and GPP Proper understanding of confounding factors remains crucial to use SIF to constrain 125

6

estimates for GPP at ecosystem or continental scales (Garbulsky et al 2014 Guanter et al 126

2012 Parazoo et al 2014) 127

Considering the above listed mechanisms several aspects need to be addressed to 128

further exploit SIF as a robust constraint for estimating GPP We therefore aim at 129

investigating the functional information content of SIF and its link to GPP considering three 130

structurally and physiologically contrasting ecosystems ie cropland perennial grassland 131

and mixed temperate forest We use an innovative combination of multi-temporal imaging 132

spectroscopy (IS) data EC flux tower observations and modeling approaches at the leaf and 133

canopy levels i) to assess the relationship between far red SIF retrieved in the O2-A band at 134

760 nm (SIF760) and GPP across ecosystems and ii) to investigate the impact of confounding 135

factors on the SIF760-GPP relationship ie temporal scaling effects and structural and 136

physiological interferences using a photosynthesis model Our findings contribute to a better 137

understanding of the information inherent in remotely measured SIF760 and its functional 138

relationship to GPP Aspects discussed will help bridging the gap between small scale studies 139

and observational attempts to estimate GPP globally 140

141

2 Methods 142

21 Study sites 143

We investigated three contrasting ecosystems located on the Central Swiss Plateau in 144

terms of structure heterogeneity species composition and annual productivity Two of these 145

ecosystems were collocated in the agricultural area near the town of Oensingen (47deg1711rdquo N 146

7deg4401rdquo E 452 masl Figure 1A) This area is characterized by relatively small agricultural 147

parcels with grassland clover fallow cropping bean maize rapeseed pea sugar beet winter 148

barley and winter wheat rotating as dominant crops Two grassland fields differently 149

managed in terms of species composition fertilization and harvesting activities (Ammann et 150

7

al 2007) were investigated as representatives of the ecosystem type perennial grassland as 151

well as a cropland (one field with arable crop rotation) for the ecosystem type cropland The 152

forest area (47deg2842rdquo N 8deg2152rdquo E Figure 1B) is located on the south-facing slope of the 153

Laegeren mountain northwest of the city of Zurich The temperate mixed forest is 154

characterized by a relatively high species diversity and a complex canopy structure with 155

beech ash sycamore and spruce being the dominant species (Eugster et al 2007 Schneider 156

et al 2014) 157

158

lt Figure 1 gt 159

160

Both test sites are well instrumented (ie with eddy-covariance flux towers 161

micrometeorological stations) and were extensively sampled during several airborne 162

campaigns between 2009 and 2013 (Table 1) 163

164

lt Table 1 gt 165

166

22 Airborne spectroscopy 167

All three study sites were measured with the Airborne Prism EXperiment (APEX) 168

sensor on several days between 2009 and 2013 (Table 1) APEX is an airborne dispersive 169

pushbroom imaging spectrometer covering the 400-2500 nm spectral region in 313 narrow 170

continuous spectral bands APEX allows capturing the O2-A atmospheric absorption feature 171

and thus retrieving SIF760 with a spectral sampling interval of 45 nm a full width at half 172

maximum (FWHM) of 57 nm and a signal to noise ratio of approximately 450 in and 800 173

outside of the O2-A feature for a 50 reflective target and a sun zenith angle of 30deg (Jehle et 174

al 2010 Schaepman et al 2015) 175

8

Radiance calibration of APEX images is performed using the approach of Hueni et al 176

(2009) Multitemporal comparison of images is secured by using consistent calibration and 177

processing versions Precise retrievals of SIF typically require data measured in sub-178

nanometer spectral resolution However the feasibility of using APEX like instruments (ie 179

FWHM ~5nm) for SIF760 retrievals was theoretically assessed in Damm et al (2011) and 180

experimentally demonstrated in recent work (Zarco-Tejada et al 2009 Zarco-Tejada et al 181

2013) The use of calibrated APEX radiance data for the fluorescence retrieval is described in 182

Section 24 and Appendix-B Vegetation indices (see Section 25) were calculated using 183

Hemispherical-Conical-Reflectance-Factor data obtained after atmospheric correction using 184

ATCOR-4 (Richter and Schlaumlpfer 2002) Resulting SIF760 and vegetation indices were 185

projected in the same map projection using parametric geocoding (Schlaumlpfer and Richter 186

2002) Details about APEX data calibration are listed in Appendix-A 187

188

23 Field spectroscopy 189

ASD (PANalytics Boulder US) field spectrometer data were acquired for validation 190

purposes within one hour of the APEX overflight during four campaigns In total 23 191

agricultural fields covering various crops including carrot maize pea rye sugar beet wheat 192

white clover were systematically sampled by measuring reflected and emitted radiances in 193

four homogeneous 3 m2 plots per field (Table 2) Incident light was quantified before and 194

after measuring each plot using a Spectralon reference panel 195

196

lt Table 2 gt 197

198

24 SIF760 retrieval 199

Emitted SIF was analytically separated from the reflected radiance flux measured with 200

ground and airborne-based spectrometers (see Appendix-B) We applied the Fraunhofer Line 201

9

Depth (FLD) approach (Plascyk 1975) which serves as a standard for SIF retrievals using 202

medium resolution instruments (Meroni et al 2009) The FLD method uses atmospheric 203

absorption bands characterized by lower incident light compared to wavelength regions 204

outside of these bands This configuration allows evaluating the in-filling of the absorption 205

bands by SIF (Damm et al 2014) In this study we used the broad O2-A oxygen absorption 206

band around 760 nm for the SIF760 retrieval For the ground data surface irradiance was 207

estimated from white reference measurements while perturbing atmospheric absorption and 208

scattering effects were assumed to be negligible on the 1 m pathway between canopy and 209

sensor (cf Damm et al (2014) for details on the set-up for SIF760 retrieval on the ground) 210

For the airborne measurements we compensated atmospheric absorption and scattering using 211

the atmospheric radiative transfer model MODTRAN5 (Berk et al 2005) Details of the 212

SIF760 retrievals using airborne data are provided in Appendix-B 213

214

25 Derivation of additional spectral indices 215

The relationship between two vegetation indices and GPP was investigated in parallel 216

to judge the performance of the new remote observation of SIF compared to commonly used 217

remote sensing approaches The photochemical reflectance index (PRI) (Gamon et al 1992) 218

was found to be sensitive to pigment changes related to the xanthophyll cycle and is 219

frequently used to quantify LUEp and GPP (Cheng et al 2013 Drolet et al 2008 Hilker et al 220

2009) The PRI was calculated as 221

(3) 222

while Rλ accounts for the surface reflectance at the wavelength λ 223

We further tested the soil adjusted Modified Chlorophyll Absorption Ratio Index 2 224

(MCARI2) (Haboudane et al 2004) that was found to be strongly related to green LAI thus 225

potential photosynthesis and is calculated as 226

10

2 $amp($)$+-amp0 (4) 227

228

26 Eddy-covariance measurements 229

Continuous turbulent fluxes of CO2 were measured at the agricultural and forest sites 230

between 2009 and 2013 using the EC technique (Baldocchi 2003) At each site the EC 231

instrumentation consisted of an open-path infrared gas analyzer (IRGA) (model LI-7500 LI-232

COR Inc Lincoln NE USA) and a three-dimensional sonic anemometer-thermometer 233

(models HS R3 and R2 Gill Instruments Lymington UK) EC measurements were made at a 234

frequency of 20 Hz and processed to half-hourly averages using the eth-flux EC software for 235

the forest and cropland (Mauder et al 2008) or a comparable custom-made EC software for 236

the grassland (Ammann et al 2007) Post-processing included corrections for damping losses 237

(Eugster and Senn 1995) air density fluctuations (Webb et al 1980) data screening for 238

optical sensor contamination stationary or low turbulence conditions (Foken and Wichura 239

1996) and statistical outliers Standardized gap filling and partitioning of CO2 fluxes was 240

performed using the methodology from Lasslop et al (2010) for the forest and cropland and 241

using the methodology from Falge et al (2001) for the grassland Measurements of 242

photosynthetic photon flux density (PPFD) quantified in micromol m-2 s-1 were converted to 243

radiance units (mW m-2 sr-1 nm-1) representing the photosynthetic active radiation (PAR) to 244

be consistent with Monteithrsquos LUE concept and the SIF radiance units 245

246

27 Leaf area index measurements 247

The Leaf Area Index (LAI) of the grassland and cropland was measured at the time of 248

APEX flights using an optical non-destructive approach based on LAI-2000 leaf area meter 249

observations (Li-Cor Lincoln USA) The LAI of the forest was estimated using digital 250

11

hemispherical photographs at two subplots following the VALERIE sampling scheme while 251

subsequent analysis were applied to obtain the LAI (Schneider et al 2014) 252

253

28 Statistical analysis of SIF760 - GPP relationship 254

SIF760 values were manually extracted from APEX-SIF760 maps for areas representing 255

the footprint of eddy-flux measurements (Figure 1) For the perennial grassland and cropland 256

we selected all pixels with SIF760 values ge 0 mW m-2 sr-1 nm-1 in respective fields although the 257

EC tower footprint can be slightly smaller than the entire field The increased number of 258

resulting pixels allows reducing data noise and covering small spatial variations within the 259

footprint Using the pixels of the entire field is justified by the large homogeneity of the 260

underlying fields (Figure 1A) For the forest site Laegeren all pixels representing an area that 261

covered 70 of all footprint occurrences were selected irrespective of spatial footprint 262

changes due to changing wind conditions Using the same area for all years reduces variations 263

in SIF due to changes in biomass and canopy structure but also reduces the representativeness 264

of selected pixels for actual spatial EC footprints 265

We used a hyperbolic model to relate GPP to SIF760 and to both vegetation indices 266

Although the hyperbolic model only empirically approximates the data behavior it is 267

supported by theoretical arguments for the SIF760-GPP relationship that are outlined hereafter 268

The photosynthetic efficiency LUEp is often described as a hyperbolic function of APAR 269

following the Michaelis-Menten theory (Michaelis and Menten 1913) 270

∙123467899 (5) 271

where c is a coefficient in radiance units and LUEpmax c represents GPPmax Following Eqs 272

(1) and (5) GPP rises asymptotically with APAR to GPPmax This equation is used in many 273

models for partitioning GPP eg Falge et al (2001) Substituting LUEp from Eq (5) and 274

APAR = SIF760 (LUEf fesc) from Eq (2) into Eq (1) results in 275

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

2

photosynthesis at ecologically relevant larger scales and provides an alternative approach to 24

study the terrestrial carbon cycle Recent studies demonstrated the reliability of measured SIF 25

signals and showed significant relationships between SIF and gross primary production (GPP) 26

at ecosystem and global scales Despite these encouraging results understanding the complex 27

mechanisms between SIF and GPP remains challenging before SIF can be finally utilized to 28

constrain estimates of GPP In this study we present a comprehensive assessment of the 29

relationship between far red SIF retrieved at 760 nm (SIF760) and GPP and its transferability 30

across three structurally and physiologically contrasting ecosystems perennial grassland 31

cropland and mixed temperate forest We use multi-temporal imaging spectroscopy (IS) data 32

acquired with the Airborne Prism Experiment (APEX) sensor as well as eddy covariance (EC) 33

flux tower data to evaluate the relationship between SIF760 and GPPEC We use simulations 34

performed with the coupled photosynthesis-fluorescence model SCOPE to prove trends 35

obtained from our observational data and assess apparent confounding factors such as 36

physiological and structural interferences or temporal scaling effects Observed relationships 37

between SIF760 and GPPEC were asymptotic and ecosystem-specific ie perennial grassland 38

(R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and mixed temperate 39

forest (R2 = 048 rRMSE = 1588) We demonstrate that asymptotic leaf level relationships 40

between SIF760-GPPEC became more linear at canopy level and scaled with temporal 41

aggregation We conclude that remote sensing of SIF provides a new observational approach 42

to decrease uncertainties in estimating GPP across ecosystems but requires dedicated 43

strategies to compensate for the various confounding factors impacting SIF-GPP 44

relationships Our findings help in bridging the gap between mechanistic understanding at leaf 45

level and ecosystem-specific observations of the relationships between SIF and GPP 46

47

Keywords Sun-induced chlorophyll fluorescence gross primary production airborne-based 48

spectroscopy APEX Fraunhofer line depth (FLD) eddy-covariance SCOPE 49

3

1 Introduction 50

Plant photosynthesis is a key process in terrestrial ecosystems mediating gas and 51

energy exchanges in the atmosphere-biosphere system (Baldocchi et al 2001 Ozanne et al 52

2003) Products of photosynthesis provide a wealth of ecosystem services that are essential 53

for human well-being including food fiber energy and oxygen (Imhoff et al 2004 54

Krausmann et al 2013 Schroter et al 2005) Photosynthesis as the underlying process for 55

plant growth is a particularly interesting indicator of crop efficiency and agricultural 56

management practices (Falloon and Betts 2010 Guanter et al 2014 Trnka et al 2004) both 57

of which having important implications for yield forecasts and for the implementation of 58

climate change adaptation strategies (IPCC 2013) 59

Observing the highly dynamic process of photosynthesis beyond the individual leaf or 60

plant levels in-situ is based on measuring the carbon dioxide (CO2) exchange between 61

vegetation and atmosphere with eddy-covariance (EC) flux towers and partitioning it into 62

gross primary production (GPP) and ecosystem respiration (Baldocchi et al 2001) 63

Measurements of plant - light interactions using spectrometers installed on for example EC 64

towers allow deriving information about the pigment status and provide an alternative 65

approach to estimate GPP (Balzarolo et al 2011 Gamon et al 2010 Hilker et al 2011) At 66

landscape scale photosynthesis can be assessed using process-based models (Sitch et al 67

2003) greenness-based satellite observations (Running et al 2004) or hybrid approaches 68

combining in-situ observations and statistical modeling (Jung et al 2011) All these 69

approaches provide important insights to study photosynthesis but usually do not allow 70

assessing photosynthesis at larger scales while preserving the high spatial variability present 71

in ecosystems EC flux tower measurements represent only smaller areas in preselected 72

ecosystems (Drolet et al 2008 Turner et al 2005) are not spatially distributed according to 73

carbon stocks (Schimel et al 2014) and do not allow spatial differentiations within the 74

4

measured footprint (Barcza et al 2009 Kljun et al 2002) Combined large scale modeling 75

and observational approaches based on vegetation greenness are spatially contiguous but face 76

the complexity of naturally varying systems including diverse interactions and complex 77

feedbacks which limit their predictive capabilities (Beer et al 2010 Goetz and Prince 1999 78

Turner et al 2005) 79

Over the last decade significant progress has been made in measuring plant-light 80

interactions and the process of photosynthesis Remote measurements of sun-induced 81

chlorophyll fluorescence (SIF) in particular open a new perspective to assess photosynthesis 82

at ecosystem scale SIF is a radiation flux emitted from plant chlorophyll molecules a few 83

nanoseconds after light absorption in the wavelength range from 600-800 nm and is 84

considered an indicator for the functional status of actual plant photosynthesis (Baker 2008) 85

Various studies demonstrated the possibility to measure SIF at certain wavelengths on ground 86

(Guanter et al 2013 Rascher et al 2009) from airborne platforms (Damm et al 2014 87

Guanter et al 2007 Zarco-Tejada et al 2012) and from satellites (Frankenberg et al 2011 88

Guanter et al 2014 Joiner et al 2013) Recent research demonstrated SIF being sensitive to 89

changes in photosynthesis showing strong links to GPP at the level of leaves (Meroni et al 90

2008 Middleton et al 2002) plants (Damm et al 2010 Rossini et al 2010) canopies (Zarco-91

Tejada et al 2013) and ecosystems (Frankenberg et al 2011 Guanter et al 2012) 92

Observed relationships between SIF and GPP are conceptually explained using an 93

approximation of GPP based on Monteithrsquos light use efficiency concept (Monteith 1972) 94

∙ (1) 95

where APAR is the absorbed photosynthetically active radiation expressed in radiance units 96

and LUEp is the efficiency of light utilization for photosynthesis and allows converting 97

measured radiances into the number of fixed CO2 molecules SIF is expressed by expanding 98

the GPP notation in Eq (1) following Guanter et al (2014) 99

13 ∙ ∙ (2) 100

5

where LUEf is the light use efficiency of SIF (fluorescence yield) and fesc accounts for a 101

structural interference determining the fraction of SIF photons escaping the canopy 102

Relationships between SIF and GPP are mostly driven by the common APAR term In 103

addition a covariance between both light use efficiencies LUEp and LUEf is expected to 104

occur in absence of the confounding impact of other protective mechanisms (Damm et al 105

2010 Guanter et al 2014) 106

The above outlined concept relating SIF and GPP simplifies a complex set of 107

underlying mechanisms and violation of any assumptions made will directly confound the 108

SIF-GPP relationship In particular the competition of three processes for de-exciting 109

absorbed light energy ie photochemistry radiative energy loss (SIF) and non-radiative 110

energy dissipation (commonly approximated as non-photochemical quenching NPQ) causes 111

complex and changing sensitivities of emitted SIF to actual rates of photosynthesis (Porcar-112

Castell et al 2014 van der Tol et al 2014 van der Tol et al 2009a) This directly implies that 113

the functional link between SIF and GPP depends on the rate of NPQ and consequently on 114

ambient stress levels At canopy scale the three-dimensional structure causes gradients in 115

light interception and light quality within canopies (Nobel et al 1993 Stewart et al 2003) 116

additionally altering the rate of NPQ (Demmig-Adams 1998 Niinemets et al 2003) and thus 117

impacting the SIF-GPP relationship Canopy structure also increases the probability for the 118

emitted SIF photons either to be re-absorbed by chlorophyll or to escape the canopy (Fournier 119

et al 2012 Knyazikhin et al 2013) to some extent violating the assumption of fesc to be 120

constant In addition to these structural and physiological effects variations in SIF signals 121

caused by for example instrumental (Damm et al 2011) or atmospheric effects (Damm et al 122

2014 Guanter et al 2010) and retrieval uncertainties related to the estimation of surface 123

irradiance (Damm et al 2015) can potentially affect the apparent relationship between SIF 124

and GPP Proper understanding of confounding factors remains crucial to use SIF to constrain 125

6

estimates for GPP at ecosystem or continental scales (Garbulsky et al 2014 Guanter et al 126

2012 Parazoo et al 2014) 127

Considering the above listed mechanisms several aspects need to be addressed to 128

further exploit SIF as a robust constraint for estimating GPP We therefore aim at 129

investigating the functional information content of SIF and its link to GPP considering three 130

structurally and physiologically contrasting ecosystems ie cropland perennial grassland 131

and mixed temperate forest We use an innovative combination of multi-temporal imaging 132

spectroscopy (IS) data EC flux tower observations and modeling approaches at the leaf and 133

canopy levels i) to assess the relationship between far red SIF retrieved in the O2-A band at 134

760 nm (SIF760) and GPP across ecosystems and ii) to investigate the impact of confounding 135

factors on the SIF760-GPP relationship ie temporal scaling effects and structural and 136

physiological interferences using a photosynthesis model Our findings contribute to a better 137

understanding of the information inherent in remotely measured SIF760 and its functional 138

relationship to GPP Aspects discussed will help bridging the gap between small scale studies 139

and observational attempts to estimate GPP globally 140

141

2 Methods 142

21 Study sites 143

We investigated three contrasting ecosystems located on the Central Swiss Plateau in 144

terms of structure heterogeneity species composition and annual productivity Two of these 145

ecosystems were collocated in the agricultural area near the town of Oensingen (47deg1711rdquo N 146

7deg4401rdquo E 452 masl Figure 1A) This area is characterized by relatively small agricultural 147

parcels with grassland clover fallow cropping bean maize rapeseed pea sugar beet winter 148

barley and winter wheat rotating as dominant crops Two grassland fields differently 149

managed in terms of species composition fertilization and harvesting activities (Ammann et 150

7

al 2007) were investigated as representatives of the ecosystem type perennial grassland as 151

well as a cropland (one field with arable crop rotation) for the ecosystem type cropland The 152

forest area (47deg2842rdquo N 8deg2152rdquo E Figure 1B) is located on the south-facing slope of the 153

Laegeren mountain northwest of the city of Zurich The temperate mixed forest is 154

characterized by a relatively high species diversity and a complex canopy structure with 155

beech ash sycamore and spruce being the dominant species (Eugster et al 2007 Schneider 156

et al 2014) 157

158

lt Figure 1 gt 159

160

Both test sites are well instrumented (ie with eddy-covariance flux towers 161

micrometeorological stations) and were extensively sampled during several airborne 162

campaigns between 2009 and 2013 (Table 1) 163

164

lt Table 1 gt 165

166

22 Airborne spectroscopy 167

All three study sites were measured with the Airborne Prism EXperiment (APEX) 168

sensor on several days between 2009 and 2013 (Table 1) APEX is an airborne dispersive 169

pushbroom imaging spectrometer covering the 400-2500 nm spectral region in 313 narrow 170

continuous spectral bands APEX allows capturing the O2-A atmospheric absorption feature 171

and thus retrieving SIF760 with a spectral sampling interval of 45 nm a full width at half 172

maximum (FWHM) of 57 nm and a signal to noise ratio of approximately 450 in and 800 173

outside of the O2-A feature for a 50 reflective target and a sun zenith angle of 30deg (Jehle et 174

al 2010 Schaepman et al 2015) 175

8

Radiance calibration of APEX images is performed using the approach of Hueni et al 176

(2009) Multitemporal comparison of images is secured by using consistent calibration and 177

processing versions Precise retrievals of SIF typically require data measured in sub-178

nanometer spectral resolution However the feasibility of using APEX like instruments (ie 179

FWHM ~5nm) for SIF760 retrievals was theoretically assessed in Damm et al (2011) and 180

experimentally demonstrated in recent work (Zarco-Tejada et al 2009 Zarco-Tejada et al 181

2013) The use of calibrated APEX radiance data for the fluorescence retrieval is described in 182

Section 24 and Appendix-B Vegetation indices (see Section 25) were calculated using 183

Hemispherical-Conical-Reflectance-Factor data obtained after atmospheric correction using 184

ATCOR-4 (Richter and Schlaumlpfer 2002) Resulting SIF760 and vegetation indices were 185

projected in the same map projection using parametric geocoding (Schlaumlpfer and Richter 186

2002) Details about APEX data calibration are listed in Appendix-A 187

188

23 Field spectroscopy 189

ASD (PANalytics Boulder US) field spectrometer data were acquired for validation 190

purposes within one hour of the APEX overflight during four campaigns In total 23 191

agricultural fields covering various crops including carrot maize pea rye sugar beet wheat 192

white clover were systematically sampled by measuring reflected and emitted radiances in 193

four homogeneous 3 m2 plots per field (Table 2) Incident light was quantified before and 194

after measuring each plot using a Spectralon reference panel 195

196

lt Table 2 gt 197

198

24 SIF760 retrieval 199

Emitted SIF was analytically separated from the reflected radiance flux measured with 200

ground and airborne-based spectrometers (see Appendix-B) We applied the Fraunhofer Line 201

9

Depth (FLD) approach (Plascyk 1975) which serves as a standard for SIF retrievals using 202

medium resolution instruments (Meroni et al 2009) The FLD method uses atmospheric 203

absorption bands characterized by lower incident light compared to wavelength regions 204

outside of these bands This configuration allows evaluating the in-filling of the absorption 205

bands by SIF (Damm et al 2014) In this study we used the broad O2-A oxygen absorption 206

band around 760 nm for the SIF760 retrieval For the ground data surface irradiance was 207

estimated from white reference measurements while perturbing atmospheric absorption and 208

scattering effects were assumed to be negligible on the 1 m pathway between canopy and 209

sensor (cf Damm et al (2014) for details on the set-up for SIF760 retrieval on the ground) 210

For the airborne measurements we compensated atmospheric absorption and scattering using 211

the atmospheric radiative transfer model MODTRAN5 (Berk et al 2005) Details of the 212

SIF760 retrievals using airborne data are provided in Appendix-B 213

214

25 Derivation of additional spectral indices 215

The relationship between two vegetation indices and GPP was investigated in parallel 216

to judge the performance of the new remote observation of SIF compared to commonly used 217

remote sensing approaches The photochemical reflectance index (PRI) (Gamon et al 1992) 218

was found to be sensitive to pigment changes related to the xanthophyll cycle and is 219

frequently used to quantify LUEp and GPP (Cheng et al 2013 Drolet et al 2008 Hilker et al 220

2009) The PRI was calculated as 221

(3) 222

while Rλ accounts for the surface reflectance at the wavelength λ 223

We further tested the soil adjusted Modified Chlorophyll Absorption Ratio Index 2 224

(MCARI2) (Haboudane et al 2004) that was found to be strongly related to green LAI thus 225

potential photosynthesis and is calculated as 226

10

2 $amp($)$+-amp0 (4) 227

228

26 Eddy-covariance measurements 229

Continuous turbulent fluxes of CO2 were measured at the agricultural and forest sites 230

between 2009 and 2013 using the EC technique (Baldocchi 2003) At each site the EC 231

instrumentation consisted of an open-path infrared gas analyzer (IRGA) (model LI-7500 LI-232

COR Inc Lincoln NE USA) and a three-dimensional sonic anemometer-thermometer 233

(models HS R3 and R2 Gill Instruments Lymington UK) EC measurements were made at a 234

frequency of 20 Hz and processed to half-hourly averages using the eth-flux EC software for 235

the forest and cropland (Mauder et al 2008) or a comparable custom-made EC software for 236

the grassland (Ammann et al 2007) Post-processing included corrections for damping losses 237

(Eugster and Senn 1995) air density fluctuations (Webb et al 1980) data screening for 238

optical sensor contamination stationary or low turbulence conditions (Foken and Wichura 239

1996) and statistical outliers Standardized gap filling and partitioning of CO2 fluxes was 240

performed using the methodology from Lasslop et al (2010) for the forest and cropland and 241

using the methodology from Falge et al (2001) for the grassland Measurements of 242

photosynthetic photon flux density (PPFD) quantified in micromol m-2 s-1 were converted to 243

radiance units (mW m-2 sr-1 nm-1) representing the photosynthetic active radiation (PAR) to 244

be consistent with Monteithrsquos LUE concept and the SIF radiance units 245

246

27 Leaf area index measurements 247

The Leaf Area Index (LAI) of the grassland and cropland was measured at the time of 248

APEX flights using an optical non-destructive approach based on LAI-2000 leaf area meter 249

observations (Li-Cor Lincoln USA) The LAI of the forest was estimated using digital 250

11

hemispherical photographs at two subplots following the VALERIE sampling scheme while 251

subsequent analysis were applied to obtain the LAI (Schneider et al 2014) 252

253

28 Statistical analysis of SIF760 - GPP relationship 254

SIF760 values were manually extracted from APEX-SIF760 maps for areas representing 255

the footprint of eddy-flux measurements (Figure 1) For the perennial grassland and cropland 256

we selected all pixels with SIF760 values ge 0 mW m-2 sr-1 nm-1 in respective fields although the 257

EC tower footprint can be slightly smaller than the entire field The increased number of 258

resulting pixels allows reducing data noise and covering small spatial variations within the 259

footprint Using the pixels of the entire field is justified by the large homogeneity of the 260

underlying fields (Figure 1A) For the forest site Laegeren all pixels representing an area that 261

covered 70 of all footprint occurrences were selected irrespective of spatial footprint 262

changes due to changing wind conditions Using the same area for all years reduces variations 263

in SIF due to changes in biomass and canopy structure but also reduces the representativeness 264

of selected pixels for actual spatial EC footprints 265

We used a hyperbolic model to relate GPP to SIF760 and to both vegetation indices 266

Although the hyperbolic model only empirically approximates the data behavior it is 267

supported by theoretical arguments for the SIF760-GPP relationship that are outlined hereafter 268

The photosynthetic efficiency LUEp is often described as a hyperbolic function of APAR 269

following the Michaelis-Menten theory (Michaelis and Menten 1913) 270

∙123467899 (5) 271

where c is a coefficient in radiance units and LUEpmax c represents GPPmax Following Eqs 272

(1) and (5) GPP rises asymptotically with APAR to GPPmax This equation is used in many 273

models for partitioning GPP eg Falge et al (2001) Substituting LUEp from Eq (5) and 274

APAR = SIF760 (LUEf fesc) from Eq (2) into Eq (1) results in 275

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

3

1 Introduction 50

Plant photosynthesis is a key process in terrestrial ecosystems mediating gas and 51

energy exchanges in the atmosphere-biosphere system (Baldocchi et al 2001 Ozanne et al 52

2003) Products of photosynthesis provide a wealth of ecosystem services that are essential 53

for human well-being including food fiber energy and oxygen (Imhoff et al 2004 54

Krausmann et al 2013 Schroter et al 2005) Photosynthesis as the underlying process for 55

plant growth is a particularly interesting indicator of crop efficiency and agricultural 56

management practices (Falloon and Betts 2010 Guanter et al 2014 Trnka et al 2004) both 57

of which having important implications for yield forecasts and for the implementation of 58

climate change adaptation strategies (IPCC 2013) 59

Observing the highly dynamic process of photosynthesis beyond the individual leaf or 60

plant levels in-situ is based on measuring the carbon dioxide (CO2) exchange between 61

vegetation and atmosphere with eddy-covariance (EC) flux towers and partitioning it into 62

gross primary production (GPP) and ecosystem respiration (Baldocchi et al 2001) 63

Measurements of plant - light interactions using spectrometers installed on for example EC 64

towers allow deriving information about the pigment status and provide an alternative 65

approach to estimate GPP (Balzarolo et al 2011 Gamon et al 2010 Hilker et al 2011) At 66

landscape scale photosynthesis can be assessed using process-based models (Sitch et al 67

2003) greenness-based satellite observations (Running et al 2004) or hybrid approaches 68

combining in-situ observations and statistical modeling (Jung et al 2011) All these 69

approaches provide important insights to study photosynthesis but usually do not allow 70

assessing photosynthesis at larger scales while preserving the high spatial variability present 71

in ecosystems EC flux tower measurements represent only smaller areas in preselected 72

ecosystems (Drolet et al 2008 Turner et al 2005) are not spatially distributed according to 73

carbon stocks (Schimel et al 2014) and do not allow spatial differentiations within the 74

4

measured footprint (Barcza et al 2009 Kljun et al 2002) Combined large scale modeling 75

and observational approaches based on vegetation greenness are spatially contiguous but face 76

the complexity of naturally varying systems including diverse interactions and complex 77

feedbacks which limit their predictive capabilities (Beer et al 2010 Goetz and Prince 1999 78

Turner et al 2005) 79

Over the last decade significant progress has been made in measuring plant-light 80

interactions and the process of photosynthesis Remote measurements of sun-induced 81

chlorophyll fluorescence (SIF) in particular open a new perspective to assess photosynthesis 82

at ecosystem scale SIF is a radiation flux emitted from plant chlorophyll molecules a few 83

nanoseconds after light absorption in the wavelength range from 600-800 nm and is 84

considered an indicator for the functional status of actual plant photosynthesis (Baker 2008) 85

Various studies demonstrated the possibility to measure SIF at certain wavelengths on ground 86

(Guanter et al 2013 Rascher et al 2009) from airborne platforms (Damm et al 2014 87

Guanter et al 2007 Zarco-Tejada et al 2012) and from satellites (Frankenberg et al 2011 88

Guanter et al 2014 Joiner et al 2013) Recent research demonstrated SIF being sensitive to 89

changes in photosynthesis showing strong links to GPP at the level of leaves (Meroni et al 90

2008 Middleton et al 2002) plants (Damm et al 2010 Rossini et al 2010) canopies (Zarco-91

Tejada et al 2013) and ecosystems (Frankenberg et al 2011 Guanter et al 2012) 92

Observed relationships between SIF and GPP are conceptually explained using an 93

approximation of GPP based on Monteithrsquos light use efficiency concept (Monteith 1972) 94

∙ (1) 95

where APAR is the absorbed photosynthetically active radiation expressed in radiance units 96

and LUEp is the efficiency of light utilization for photosynthesis and allows converting 97

measured radiances into the number of fixed CO2 molecules SIF is expressed by expanding 98

the GPP notation in Eq (1) following Guanter et al (2014) 99

13 ∙ ∙ (2) 100

5

where LUEf is the light use efficiency of SIF (fluorescence yield) and fesc accounts for a 101

structural interference determining the fraction of SIF photons escaping the canopy 102

Relationships between SIF and GPP are mostly driven by the common APAR term In 103

addition a covariance between both light use efficiencies LUEp and LUEf is expected to 104

occur in absence of the confounding impact of other protective mechanisms (Damm et al 105

2010 Guanter et al 2014) 106

The above outlined concept relating SIF and GPP simplifies a complex set of 107

underlying mechanisms and violation of any assumptions made will directly confound the 108

SIF-GPP relationship In particular the competition of three processes for de-exciting 109

absorbed light energy ie photochemistry radiative energy loss (SIF) and non-radiative 110

energy dissipation (commonly approximated as non-photochemical quenching NPQ) causes 111

complex and changing sensitivities of emitted SIF to actual rates of photosynthesis (Porcar-112

Castell et al 2014 van der Tol et al 2014 van der Tol et al 2009a) This directly implies that 113

the functional link between SIF and GPP depends on the rate of NPQ and consequently on 114

ambient stress levels At canopy scale the three-dimensional structure causes gradients in 115

light interception and light quality within canopies (Nobel et al 1993 Stewart et al 2003) 116

additionally altering the rate of NPQ (Demmig-Adams 1998 Niinemets et al 2003) and thus 117

impacting the SIF-GPP relationship Canopy structure also increases the probability for the 118

emitted SIF photons either to be re-absorbed by chlorophyll or to escape the canopy (Fournier 119

et al 2012 Knyazikhin et al 2013) to some extent violating the assumption of fesc to be 120

constant In addition to these structural and physiological effects variations in SIF signals 121

caused by for example instrumental (Damm et al 2011) or atmospheric effects (Damm et al 122

2014 Guanter et al 2010) and retrieval uncertainties related to the estimation of surface 123

irradiance (Damm et al 2015) can potentially affect the apparent relationship between SIF 124

and GPP Proper understanding of confounding factors remains crucial to use SIF to constrain 125

6

estimates for GPP at ecosystem or continental scales (Garbulsky et al 2014 Guanter et al 126

2012 Parazoo et al 2014) 127

Considering the above listed mechanisms several aspects need to be addressed to 128

further exploit SIF as a robust constraint for estimating GPP We therefore aim at 129

investigating the functional information content of SIF and its link to GPP considering three 130

structurally and physiologically contrasting ecosystems ie cropland perennial grassland 131

and mixed temperate forest We use an innovative combination of multi-temporal imaging 132

spectroscopy (IS) data EC flux tower observations and modeling approaches at the leaf and 133

canopy levels i) to assess the relationship between far red SIF retrieved in the O2-A band at 134

760 nm (SIF760) and GPP across ecosystems and ii) to investigate the impact of confounding 135

factors on the SIF760-GPP relationship ie temporal scaling effects and structural and 136

physiological interferences using a photosynthesis model Our findings contribute to a better 137

understanding of the information inherent in remotely measured SIF760 and its functional 138

relationship to GPP Aspects discussed will help bridging the gap between small scale studies 139

and observational attempts to estimate GPP globally 140

141

2 Methods 142

21 Study sites 143

We investigated three contrasting ecosystems located on the Central Swiss Plateau in 144

terms of structure heterogeneity species composition and annual productivity Two of these 145

ecosystems were collocated in the agricultural area near the town of Oensingen (47deg1711rdquo N 146

7deg4401rdquo E 452 masl Figure 1A) This area is characterized by relatively small agricultural 147

parcels with grassland clover fallow cropping bean maize rapeseed pea sugar beet winter 148

barley and winter wheat rotating as dominant crops Two grassland fields differently 149

managed in terms of species composition fertilization and harvesting activities (Ammann et 150

7

al 2007) were investigated as representatives of the ecosystem type perennial grassland as 151

well as a cropland (one field with arable crop rotation) for the ecosystem type cropland The 152

forest area (47deg2842rdquo N 8deg2152rdquo E Figure 1B) is located on the south-facing slope of the 153

Laegeren mountain northwest of the city of Zurich The temperate mixed forest is 154

characterized by a relatively high species diversity and a complex canopy structure with 155

beech ash sycamore and spruce being the dominant species (Eugster et al 2007 Schneider 156

et al 2014) 157

158

lt Figure 1 gt 159

160

Both test sites are well instrumented (ie with eddy-covariance flux towers 161

micrometeorological stations) and were extensively sampled during several airborne 162

campaigns between 2009 and 2013 (Table 1) 163

164

lt Table 1 gt 165

166

22 Airborne spectroscopy 167

All three study sites were measured with the Airborne Prism EXperiment (APEX) 168

sensor on several days between 2009 and 2013 (Table 1) APEX is an airborne dispersive 169

pushbroom imaging spectrometer covering the 400-2500 nm spectral region in 313 narrow 170

continuous spectral bands APEX allows capturing the O2-A atmospheric absorption feature 171

and thus retrieving SIF760 with a spectral sampling interval of 45 nm a full width at half 172

maximum (FWHM) of 57 nm and a signal to noise ratio of approximately 450 in and 800 173

outside of the O2-A feature for a 50 reflective target and a sun zenith angle of 30deg (Jehle et 174

al 2010 Schaepman et al 2015) 175

8

Radiance calibration of APEX images is performed using the approach of Hueni et al 176

(2009) Multitemporal comparison of images is secured by using consistent calibration and 177

processing versions Precise retrievals of SIF typically require data measured in sub-178

nanometer spectral resolution However the feasibility of using APEX like instruments (ie 179

FWHM ~5nm) for SIF760 retrievals was theoretically assessed in Damm et al (2011) and 180

experimentally demonstrated in recent work (Zarco-Tejada et al 2009 Zarco-Tejada et al 181

2013) The use of calibrated APEX radiance data for the fluorescence retrieval is described in 182

Section 24 and Appendix-B Vegetation indices (see Section 25) were calculated using 183

Hemispherical-Conical-Reflectance-Factor data obtained after atmospheric correction using 184

ATCOR-4 (Richter and Schlaumlpfer 2002) Resulting SIF760 and vegetation indices were 185

projected in the same map projection using parametric geocoding (Schlaumlpfer and Richter 186

2002) Details about APEX data calibration are listed in Appendix-A 187

188

23 Field spectroscopy 189

ASD (PANalytics Boulder US) field spectrometer data were acquired for validation 190

purposes within one hour of the APEX overflight during four campaigns In total 23 191

agricultural fields covering various crops including carrot maize pea rye sugar beet wheat 192

white clover were systematically sampled by measuring reflected and emitted radiances in 193

four homogeneous 3 m2 plots per field (Table 2) Incident light was quantified before and 194

after measuring each plot using a Spectralon reference panel 195

196

lt Table 2 gt 197

198

24 SIF760 retrieval 199

Emitted SIF was analytically separated from the reflected radiance flux measured with 200

ground and airborne-based spectrometers (see Appendix-B) We applied the Fraunhofer Line 201

9

Depth (FLD) approach (Plascyk 1975) which serves as a standard for SIF retrievals using 202

medium resolution instruments (Meroni et al 2009) The FLD method uses atmospheric 203

absorption bands characterized by lower incident light compared to wavelength regions 204

outside of these bands This configuration allows evaluating the in-filling of the absorption 205

bands by SIF (Damm et al 2014) In this study we used the broad O2-A oxygen absorption 206

band around 760 nm for the SIF760 retrieval For the ground data surface irradiance was 207

estimated from white reference measurements while perturbing atmospheric absorption and 208

scattering effects were assumed to be negligible on the 1 m pathway between canopy and 209

sensor (cf Damm et al (2014) for details on the set-up for SIF760 retrieval on the ground) 210

For the airborne measurements we compensated atmospheric absorption and scattering using 211

the atmospheric radiative transfer model MODTRAN5 (Berk et al 2005) Details of the 212

SIF760 retrievals using airborne data are provided in Appendix-B 213

214

25 Derivation of additional spectral indices 215

The relationship between two vegetation indices and GPP was investigated in parallel 216

to judge the performance of the new remote observation of SIF compared to commonly used 217

remote sensing approaches The photochemical reflectance index (PRI) (Gamon et al 1992) 218

was found to be sensitive to pigment changes related to the xanthophyll cycle and is 219

frequently used to quantify LUEp and GPP (Cheng et al 2013 Drolet et al 2008 Hilker et al 220

2009) The PRI was calculated as 221

(3) 222

while Rλ accounts for the surface reflectance at the wavelength λ 223

We further tested the soil adjusted Modified Chlorophyll Absorption Ratio Index 2 224

(MCARI2) (Haboudane et al 2004) that was found to be strongly related to green LAI thus 225

potential photosynthesis and is calculated as 226

10

2 $amp($)$+-amp0 (4) 227

228

26 Eddy-covariance measurements 229

Continuous turbulent fluxes of CO2 were measured at the agricultural and forest sites 230

between 2009 and 2013 using the EC technique (Baldocchi 2003) At each site the EC 231

instrumentation consisted of an open-path infrared gas analyzer (IRGA) (model LI-7500 LI-232

COR Inc Lincoln NE USA) and a three-dimensional sonic anemometer-thermometer 233

(models HS R3 and R2 Gill Instruments Lymington UK) EC measurements were made at a 234

frequency of 20 Hz and processed to half-hourly averages using the eth-flux EC software for 235

the forest and cropland (Mauder et al 2008) or a comparable custom-made EC software for 236

the grassland (Ammann et al 2007) Post-processing included corrections for damping losses 237

(Eugster and Senn 1995) air density fluctuations (Webb et al 1980) data screening for 238

optical sensor contamination stationary or low turbulence conditions (Foken and Wichura 239

1996) and statistical outliers Standardized gap filling and partitioning of CO2 fluxes was 240

performed using the methodology from Lasslop et al (2010) for the forest and cropland and 241

using the methodology from Falge et al (2001) for the grassland Measurements of 242

photosynthetic photon flux density (PPFD) quantified in micromol m-2 s-1 were converted to 243

radiance units (mW m-2 sr-1 nm-1) representing the photosynthetic active radiation (PAR) to 244

be consistent with Monteithrsquos LUE concept and the SIF radiance units 245

246

27 Leaf area index measurements 247

The Leaf Area Index (LAI) of the grassland and cropland was measured at the time of 248

APEX flights using an optical non-destructive approach based on LAI-2000 leaf area meter 249

observations (Li-Cor Lincoln USA) The LAI of the forest was estimated using digital 250

11

hemispherical photographs at two subplots following the VALERIE sampling scheme while 251

subsequent analysis were applied to obtain the LAI (Schneider et al 2014) 252

253

28 Statistical analysis of SIF760 - GPP relationship 254

SIF760 values were manually extracted from APEX-SIF760 maps for areas representing 255

the footprint of eddy-flux measurements (Figure 1) For the perennial grassland and cropland 256

we selected all pixels with SIF760 values ge 0 mW m-2 sr-1 nm-1 in respective fields although the 257

EC tower footprint can be slightly smaller than the entire field The increased number of 258

resulting pixels allows reducing data noise and covering small spatial variations within the 259

footprint Using the pixels of the entire field is justified by the large homogeneity of the 260

underlying fields (Figure 1A) For the forest site Laegeren all pixels representing an area that 261

covered 70 of all footprint occurrences were selected irrespective of spatial footprint 262

changes due to changing wind conditions Using the same area for all years reduces variations 263

in SIF due to changes in biomass and canopy structure but also reduces the representativeness 264

of selected pixels for actual spatial EC footprints 265

We used a hyperbolic model to relate GPP to SIF760 and to both vegetation indices 266

Although the hyperbolic model only empirically approximates the data behavior it is 267

supported by theoretical arguments for the SIF760-GPP relationship that are outlined hereafter 268

The photosynthetic efficiency LUEp is often described as a hyperbolic function of APAR 269

following the Michaelis-Menten theory (Michaelis and Menten 1913) 270

∙123467899 (5) 271

where c is a coefficient in radiance units and LUEpmax c represents GPPmax Following Eqs 272

(1) and (5) GPP rises asymptotically with APAR to GPPmax This equation is used in many 273

models for partitioning GPP eg Falge et al (2001) Substituting LUEp from Eq (5) and 274

APAR = SIF760 (LUEf fesc) from Eq (2) into Eq (1) results in 275

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

4

measured footprint (Barcza et al 2009 Kljun et al 2002) Combined large scale modeling 75

and observational approaches based on vegetation greenness are spatially contiguous but face 76

the complexity of naturally varying systems including diverse interactions and complex 77

feedbacks which limit their predictive capabilities (Beer et al 2010 Goetz and Prince 1999 78

Turner et al 2005) 79

Over the last decade significant progress has been made in measuring plant-light 80

interactions and the process of photosynthesis Remote measurements of sun-induced 81

chlorophyll fluorescence (SIF) in particular open a new perspective to assess photosynthesis 82

at ecosystem scale SIF is a radiation flux emitted from plant chlorophyll molecules a few 83

nanoseconds after light absorption in the wavelength range from 600-800 nm and is 84

considered an indicator for the functional status of actual plant photosynthesis (Baker 2008) 85

Various studies demonstrated the possibility to measure SIF at certain wavelengths on ground 86

(Guanter et al 2013 Rascher et al 2009) from airborne platforms (Damm et al 2014 87

Guanter et al 2007 Zarco-Tejada et al 2012) and from satellites (Frankenberg et al 2011 88

Guanter et al 2014 Joiner et al 2013) Recent research demonstrated SIF being sensitive to 89

changes in photosynthesis showing strong links to GPP at the level of leaves (Meroni et al 90

2008 Middleton et al 2002) plants (Damm et al 2010 Rossini et al 2010) canopies (Zarco-91

Tejada et al 2013) and ecosystems (Frankenberg et al 2011 Guanter et al 2012) 92

Observed relationships between SIF and GPP are conceptually explained using an 93

approximation of GPP based on Monteithrsquos light use efficiency concept (Monteith 1972) 94

∙ (1) 95

where APAR is the absorbed photosynthetically active radiation expressed in radiance units 96

and LUEp is the efficiency of light utilization for photosynthesis and allows converting 97

measured radiances into the number of fixed CO2 molecules SIF is expressed by expanding 98

the GPP notation in Eq (1) following Guanter et al (2014) 99

13 ∙ ∙ (2) 100

5

where LUEf is the light use efficiency of SIF (fluorescence yield) and fesc accounts for a 101

structural interference determining the fraction of SIF photons escaping the canopy 102

Relationships between SIF and GPP are mostly driven by the common APAR term In 103

addition a covariance between both light use efficiencies LUEp and LUEf is expected to 104

occur in absence of the confounding impact of other protective mechanisms (Damm et al 105

2010 Guanter et al 2014) 106

The above outlined concept relating SIF and GPP simplifies a complex set of 107

underlying mechanisms and violation of any assumptions made will directly confound the 108

SIF-GPP relationship In particular the competition of three processes for de-exciting 109

absorbed light energy ie photochemistry radiative energy loss (SIF) and non-radiative 110

energy dissipation (commonly approximated as non-photochemical quenching NPQ) causes 111

complex and changing sensitivities of emitted SIF to actual rates of photosynthesis (Porcar-112

Castell et al 2014 van der Tol et al 2014 van der Tol et al 2009a) This directly implies that 113

the functional link between SIF and GPP depends on the rate of NPQ and consequently on 114

ambient stress levels At canopy scale the three-dimensional structure causes gradients in 115

light interception and light quality within canopies (Nobel et al 1993 Stewart et al 2003) 116

additionally altering the rate of NPQ (Demmig-Adams 1998 Niinemets et al 2003) and thus 117

impacting the SIF-GPP relationship Canopy structure also increases the probability for the 118

emitted SIF photons either to be re-absorbed by chlorophyll or to escape the canopy (Fournier 119

et al 2012 Knyazikhin et al 2013) to some extent violating the assumption of fesc to be 120

constant In addition to these structural and physiological effects variations in SIF signals 121

caused by for example instrumental (Damm et al 2011) or atmospheric effects (Damm et al 122

2014 Guanter et al 2010) and retrieval uncertainties related to the estimation of surface 123

irradiance (Damm et al 2015) can potentially affect the apparent relationship between SIF 124

and GPP Proper understanding of confounding factors remains crucial to use SIF to constrain 125

6

estimates for GPP at ecosystem or continental scales (Garbulsky et al 2014 Guanter et al 126

2012 Parazoo et al 2014) 127

Considering the above listed mechanisms several aspects need to be addressed to 128

further exploit SIF as a robust constraint for estimating GPP We therefore aim at 129

investigating the functional information content of SIF and its link to GPP considering three 130

structurally and physiologically contrasting ecosystems ie cropland perennial grassland 131

and mixed temperate forest We use an innovative combination of multi-temporal imaging 132

spectroscopy (IS) data EC flux tower observations and modeling approaches at the leaf and 133

canopy levels i) to assess the relationship between far red SIF retrieved in the O2-A band at 134

760 nm (SIF760) and GPP across ecosystems and ii) to investigate the impact of confounding 135

factors on the SIF760-GPP relationship ie temporal scaling effects and structural and 136

physiological interferences using a photosynthesis model Our findings contribute to a better 137

understanding of the information inherent in remotely measured SIF760 and its functional 138

relationship to GPP Aspects discussed will help bridging the gap between small scale studies 139

and observational attempts to estimate GPP globally 140

141

2 Methods 142

21 Study sites 143

We investigated three contrasting ecosystems located on the Central Swiss Plateau in 144

terms of structure heterogeneity species composition and annual productivity Two of these 145

ecosystems were collocated in the agricultural area near the town of Oensingen (47deg1711rdquo N 146

7deg4401rdquo E 452 masl Figure 1A) This area is characterized by relatively small agricultural 147

parcels with grassland clover fallow cropping bean maize rapeseed pea sugar beet winter 148

barley and winter wheat rotating as dominant crops Two grassland fields differently 149

managed in terms of species composition fertilization and harvesting activities (Ammann et 150

7

al 2007) were investigated as representatives of the ecosystem type perennial grassland as 151

well as a cropland (one field with arable crop rotation) for the ecosystem type cropland The 152

forest area (47deg2842rdquo N 8deg2152rdquo E Figure 1B) is located on the south-facing slope of the 153

Laegeren mountain northwest of the city of Zurich The temperate mixed forest is 154

characterized by a relatively high species diversity and a complex canopy structure with 155

beech ash sycamore and spruce being the dominant species (Eugster et al 2007 Schneider 156

et al 2014) 157

158

lt Figure 1 gt 159

160

Both test sites are well instrumented (ie with eddy-covariance flux towers 161

micrometeorological stations) and were extensively sampled during several airborne 162

campaigns between 2009 and 2013 (Table 1) 163

164

lt Table 1 gt 165

166

22 Airborne spectroscopy 167

All three study sites were measured with the Airborne Prism EXperiment (APEX) 168

sensor on several days between 2009 and 2013 (Table 1) APEX is an airborne dispersive 169

pushbroom imaging spectrometer covering the 400-2500 nm spectral region in 313 narrow 170

continuous spectral bands APEX allows capturing the O2-A atmospheric absorption feature 171

and thus retrieving SIF760 with a spectral sampling interval of 45 nm a full width at half 172

maximum (FWHM) of 57 nm and a signal to noise ratio of approximately 450 in and 800 173

outside of the O2-A feature for a 50 reflective target and a sun zenith angle of 30deg (Jehle et 174

al 2010 Schaepman et al 2015) 175

8

Radiance calibration of APEX images is performed using the approach of Hueni et al 176

(2009) Multitemporal comparison of images is secured by using consistent calibration and 177

processing versions Precise retrievals of SIF typically require data measured in sub-178

nanometer spectral resolution However the feasibility of using APEX like instruments (ie 179

FWHM ~5nm) for SIF760 retrievals was theoretically assessed in Damm et al (2011) and 180

experimentally demonstrated in recent work (Zarco-Tejada et al 2009 Zarco-Tejada et al 181

2013) The use of calibrated APEX radiance data for the fluorescence retrieval is described in 182

Section 24 and Appendix-B Vegetation indices (see Section 25) were calculated using 183

Hemispherical-Conical-Reflectance-Factor data obtained after atmospheric correction using 184

ATCOR-4 (Richter and Schlaumlpfer 2002) Resulting SIF760 and vegetation indices were 185

projected in the same map projection using parametric geocoding (Schlaumlpfer and Richter 186

2002) Details about APEX data calibration are listed in Appendix-A 187

188

23 Field spectroscopy 189

ASD (PANalytics Boulder US) field spectrometer data were acquired for validation 190

purposes within one hour of the APEX overflight during four campaigns In total 23 191

agricultural fields covering various crops including carrot maize pea rye sugar beet wheat 192

white clover were systematically sampled by measuring reflected and emitted radiances in 193

four homogeneous 3 m2 plots per field (Table 2) Incident light was quantified before and 194

after measuring each plot using a Spectralon reference panel 195

196

lt Table 2 gt 197

198

24 SIF760 retrieval 199

Emitted SIF was analytically separated from the reflected radiance flux measured with 200

ground and airborne-based spectrometers (see Appendix-B) We applied the Fraunhofer Line 201

9

Depth (FLD) approach (Plascyk 1975) which serves as a standard for SIF retrievals using 202

medium resolution instruments (Meroni et al 2009) The FLD method uses atmospheric 203

absorption bands characterized by lower incident light compared to wavelength regions 204

outside of these bands This configuration allows evaluating the in-filling of the absorption 205

bands by SIF (Damm et al 2014) In this study we used the broad O2-A oxygen absorption 206

band around 760 nm for the SIF760 retrieval For the ground data surface irradiance was 207

estimated from white reference measurements while perturbing atmospheric absorption and 208

scattering effects were assumed to be negligible on the 1 m pathway between canopy and 209

sensor (cf Damm et al (2014) for details on the set-up for SIF760 retrieval on the ground) 210

For the airborne measurements we compensated atmospheric absorption and scattering using 211

the atmospheric radiative transfer model MODTRAN5 (Berk et al 2005) Details of the 212

SIF760 retrievals using airborne data are provided in Appendix-B 213

214

25 Derivation of additional spectral indices 215

The relationship between two vegetation indices and GPP was investigated in parallel 216

to judge the performance of the new remote observation of SIF compared to commonly used 217

remote sensing approaches The photochemical reflectance index (PRI) (Gamon et al 1992) 218

was found to be sensitive to pigment changes related to the xanthophyll cycle and is 219

frequently used to quantify LUEp and GPP (Cheng et al 2013 Drolet et al 2008 Hilker et al 220

2009) The PRI was calculated as 221

(3) 222

while Rλ accounts for the surface reflectance at the wavelength λ 223

We further tested the soil adjusted Modified Chlorophyll Absorption Ratio Index 2 224

(MCARI2) (Haboudane et al 2004) that was found to be strongly related to green LAI thus 225

potential photosynthesis and is calculated as 226

10

2 $amp($)$+-amp0 (4) 227

228

26 Eddy-covariance measurements 229

Continuous turbulent fluxes of CO2 were measured at the agricultural and forest sites 230

between 2009 and 2013 using the EC technique (Baldocchi 2003) At each site the EC 231

instrumentation consisted of an open-path infrared gas analyzer (IRGA) (model LI-7500 LI-232

COR Inc Lincoln NE USA) and a three-dimensional sonic anemometer-thermometer 233

(models HS R3 and R2 Gill Instruments Lymington UK) EC measurements were made at a 234

frequency of 20 Hz and processed to half-hourly averages using the eth-flux EC software for 235

the forest and cropland (Mauder et al 2008) or a comparable custom-made EC software for 236

the grassland (Ammann et al 2007) Post-processing included corrections for damping losses 237

(Eugster and Senn 1995) air density fluctuations (Webb et al 1980) data screening for 238

optical sensor contamination stationary or low turbulence conditions (Foken and Wichura 239

1996) and statistical outliers Standardized gap filling and partitioning of CO2 fluxes was 240

performed using the methodology from Lasslop et al (2010) for the forest and cropland and 241

using the methodology from Falge et al (2001) for the grassland Measurements of 242

photosynthetic photon flux density (PPFD) quantified in micromol m-2 s-1 were converted to 243

radiance units (mW m-2 sr-1 nm-1) representing the photosynthetic active radiation (PAR) to 244

be consistent with Monteithrsquos LUE concept and the SIF radiance units 245

246

27 Leaf area index measurements 247

The Leaf Area Index (LAI) of the grassland and cropland was measured at the time of 248

APEX flights using an optical non-destructive approach based on LAI-2000 leaf area meter 249

observations (Li-Cor Lincoln USA) The LAI of the forest was estimated using digital 250

11

hemispherical photographs at two subplots following the VALERIE sampling scheme while 251

subsequent analysis were applied to obtain the LAI (Schneider et al 2014) 252

253

28 Statistical analysis of SIF760 - GPP relationship 254

SIF760 values were manually extracted from APEX-SIF760 maps for areas representing 255

the footprint of eddy-flux measurements (Figure 1) For the perennial grassland and cropland 256

we selected all pixels with SIF760 values ge 0 mW m-2 sr-1 nm-1 in respective fields although the 257

EC tower footprint can be slightly smaller than the entire field The increased number of 258

resulting pixels allows reducing data noise and covering small spatial variations within the 259

footprint Using the pixels of the entire field is justified by the large homogeneity of the 260

underlying fields (Figure 1A) For the forest site Laegeren all pixels representing an area that 261

covered 70 of all footprint occurrences were selected irrespective of spatial footprint 262

changes due to changing wind conditions Using the same area for all years reduces variations 263

in SIF due to changes in biomass and canopy structure but also reduces the representativeness 264

of selected pixels for actual spatial EC footprints 265

We used a hyperbolic model to relate GPP to SIF760 and to both vegetation indices 266

Although the hyperbolic model only empirically approximates the data behavior it is 267

supported by theoretical arguments for the SIF760-GPP relationship that are outlined hereafter 268

The photosynthetic efficiency LUEp is often described as a hyperbolic function of APAR 269

following the Michaelis-Menten theory (Michaelis and Menten 1913) 270

∙123467899 (5) 271

where c is a coefficient in radiance units and LUEpmax c represents GPPmax Following Eqs 272

(1) and (5) GPP rises asymptotically with APAR to GPPmax This equation is used in many 273

models for partitioning GPP eg Falge et al (2001) Substituting LUEp from Eq (5) and 274

APAR = SIF760 (LUEf fesc) from Eq (2) into Eq (1) results in 275

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

5

where LUEf is the light use efficiency of SIF (fluorescence yield) and fesc accounts for a 101

structural interference determining the fraction of SIF photons escaping the canopy 102

Relationships between SIF and GPP are mostly driven by the common APAR term In 103

addition a covariance between both light use efficiencies LUEp and LUEf is expected to 104

occur in absence of the confounding impact of other protective mechanisms (Damm et al 105

2010 Guanter et al 2014) 106

The above outlined concept relating SIF and GPP simplifies a complex set of 107

underlying mechanisms and violation of any assumptions made will directly confound the 108

SIF-GPP relationship In particular the competition of three processes for de-exciting 109

absorbed light energy ie photochemistry radiative energy loss (SIF) and non-radiative 110

energy dissipation (commonly approximated as non-photochemical quenching NPQ) causes 111

complex and changing sensitivities of emitted SIF to actual rates of photosynthesis (Porcar-112

Castell et al 2014 van der Tol et al 2014 van der Tol et al 2009a) This directly implies that 113

the functional link between SIF and GPP depends on the rate of NPQ and consequently on 114

ambient stress levels At canopy scale the three-dimensional structure causes gradients in 115

light interception and light quality within canopies (Nobel et al 1993 Stewart et al 2003) 116

additionally altering the rate of NPQ (Demmig-Adams 1998 Niinemets et al 2003) and thus 117

impacting the SIF-GPP relationship Canopy structure also increases the probability for the 118

emitted SIF photons either to be re-absorbed by chlorophyll or to escape the canopy (Fournier 119

et al 2012 Knyazikhin et al 2013) to some extent violating the assumption of fesc to be 120

constant In addition to these structural and physiological effects variations in SIF signals 121

caused by for example instrumental (Damm et al 2011) or atmospheric effects (Damm et al 122

2014 Guanter et al 2010) and retrieval uncertainties related to the estimation of surface 123

irradiance (Damm et al 2015) can potentially affect the apparent relationship between SIF 124

and GPP Proper understanding of confounding factors remains crucial to use SIF to constrain 125

6

estimates for GPP at ecosystem or continental scales (Garbulsky et al 2014 Guanter et al 126

2012 Parazoo et al 2014) 127

Considering the above listed mechanisms several aspects need to be addressed to 128

further exploit SIF as a robust constraint for estimating GPP We therefore aim at 129

investigating the functional information content of SIF and its link to GPP considering three 130

structurally and physiologically contrasting ecosystems ie cropland perennial grassland 131

and mixed temperate forest We use an innovative combination of multi-temporal imaging 132

spectroscopy (IS) data EC flux tower observations and modeling approaches at the leaf and 133

canopy levels i) to assess the relationship between far red SIF retrieved in the O2-A band at 134

760 nm (SIF760) and GPP across ecosystems and ii) to investigate the impact of confounding 135

factors on the SIF760-GPP relationship ie temporal scaling effects and structural and 136

physiological interferences using a photosynthesis model Our findings contribute to a better 137

understanding of the information inherent in remotely measured SIF760 and its functional 138

relationship to GPP Aspects discussed will help bridging the gap between small scale studies 139

and observational attempts to estimate GPP globally 140

141

2 Methods 142

21 Study sites 143

We investigated three contrasting ecosystems located on the Central Swiss Plateau in 144

terms of structure heterogeneity species composition and annual productivity Two of these 145

ecosystems were collocated in the agricultural area near the town of Oensingen (47deg1711rdquo N 146

7deg4401rdquo E 452 masl Figure 1A) This area is characterized by relatively small agricultural 147

parcels with grassland clover fallow cropping bean maize rapeseed pea sugar beet winter 148

barley and winter wheat rotating as dominant crops Two grassland fields differently 149

managed in terms of species composition fertilization and harvesting activities (Ammann et 150

7

al 2007) were investigated as representatives of the ecosystem type perennial grassland as 151

well as a cropland (one field with arable crop rotation) for the ecosystem type cropland The 152

forest area (47deg2842rdquo N 8deg2152rdquo E Figure 1B) is located on the south-facing slope of the 153

Laegeren mountain northwest of the city of Zurich The temperate mixed forest is 154

characterized by a relatively high species diversity and a complex canopy structure with 155

beech ash sycamore and spruce being the dominant species (Eugster et al 2007 Schneider 156

et al 2014) 157

158

lt Figure 1 gt 159

160

Both test sites are well instrumented (ie with eddy-covariance flux towers 161

micrometeorological stations) and were extensively sampled during several airborne 162

campaigns between 2009 and 2013 (Table 1) 163

164

lt Table 1 gt 165

166

22 Airborne spectroscopy 167

All three study sites were measured with the Airborne Prism EXperiment (APEX) 168

sensor on several days between 2009 and 2013 (Table 1) APEX is an airborne dispersive 169

pushbroom imaging spectrometer covering the 400-2500 nm spectral region in 313 narrow 170

continuous spectral bands APEX allows capturing the O2-A atmospheric absorption feature 171

and thus retrieving SIF760 with a spectral sampling interval of 45 nm a full width at half 172

maximum (FWHM) of 57 nm and a signal to noise ratio of approximately 450 in and 800 173

outside of the O2-A feature for a 50 reflective target and a sun zenith angle of 30deg (Jehle et 174

al 2010 Schaepman et al 2015) 175

8

Radiance calibration of APEX images is performed using the approach of Hueni et al 176

(2009) Multitemporal comparison of images is secured by using consistent calibration and 177

processing versions Precise retrievals of SIF typically require data measured in sub-178

nanometer spectral resolution However the feasibility of using APEX like instruments (ie 179

FWHM ~5nm) for SIF760 retrievals was theoretically assessed in Damm et al (2011) and 180

experimentally demonstrated in recent work (Zarco-Tejada et al 2009 Zarco-Tejada et al 181

2013) The use of calibrated APEX radiance data for the fluorescence retrieval is described in 182

Section 24 and Appendix-B Vegetation indices (see Section 25) were calculated using 183

Hemispherical-Conical-Reflectance-Factor data obtained after atmospheric correction using 184

ATCOR-4 (Richter and Schlaumlpfer 2002) Resulting SIF760 and vegetation indices were 185

projected in the same map projection using parametric geocoding (Schlaumlpfer and Richter 186

2002) Details about APEX data calibration are listed in Appendix-A 187

188

23 Field spectroscopy 189

ASD (PANalytics Boulder US) field spectrometer data were acquired for validation 190

purposes within one hour of the APEX overflight during four campaigns In total 23 191

agricultural fields covering various crops including carrot maize pea rye sugar beet wheat 192

white clover were systematically sampled by measuring reflected and emitted radiances in 193

four homogeneous 3 m2 plots per field (Table 2) Incident light was quantified before and 194

after measuring each plot using a Spectralon reference panel 195

196

lt Table 2 gt 197

198

24 SIF760 retrieval 199

Emitted SIF was analytically separated from the reflected radiance flux measured with 200

ground and airborne-based spectrometers (see Appendix-B) We applied the Fraunhofer Line 201

9

Depth (FLD) approach (Plascyk 1975) which serves as a standard for SIF retrievals using 202

medium resolution instruments (Meroni et al 2009) The FLD method uses atmospheric 203

absorption bands characterized by lower incident light compared to wavelength regions 204

outside of these bands This configuration allows evaluating the in-filling of the absorption 205

bands by SIF (Damm et al 2014) In this study we used the broad O2-A oxygen absorption 206

band around 760 nm for the SIF760 retrieval For the ground data surface irradiance was 207

estimated from white reference measurements while perturbing atmospheric absorption and 208

scattering effects were assumed to be negligible on the 1 m pathway between canopy and 209

sensor (cf Damm et al (2014) for details on the set-up for SIF760 retrieval on the ground) 210

For the airborne measurements we compensated atmospheric absorption and scattering using 211

the atmospheric radiative transfer model MODTRAN5 (Berk et al 2005) Details of the 212

SIF760 retrievals using airborne data are provided in Appendix-B 213

214

25 Derivation of additional spectral indices 215

The relationship between two vegetation indices and GPP was investigated in parallel 216

to judge the performance of the new remote observation of SIF compared to commonly used 217

remote sensing approaches The photochemical reflectance index (PRI) (Gamon et al 1992) 218

was found to be sensitive to pigment changes related to the xanthophyll cycle and is 219

frequently used to quantify LUEp and GPP (Cheng et al 2013 Drolet et al 2008 Hilker et al 220

2009) The PRI was calculated as 221

(3) 222

while Rλ accounts for the surface reflectance at the wavelength λ 223

We further tested the soil adjusted Modified Chlorophyll Absorption Ratio Index 2 224

(MCARI2) (Haboudane et al 2004) that was found to be strongly related to green LAI thus 225

potential photosynthesis and is calculated as 226

10

2 $amp($)$+-amp0 (4) 227

228

26 Eddy-covariance measurements 229

Continuous turbulent fluxes of CO2 were measured at the agricultural and forest sites 230

between 2009 and 2013 using the EC technique (Baldocchi 2003) At each site the EC 231

instrumentation consisted of an open-path infrared gas analyzer (IRGA) (model LI-7500 LI-232

COR Inc Lincoln NE USA) and a three-dimensional sonic anemometer-thermometer 233

(models HS R3 and R2 Gill Instruments Lymington UK) EC measurements were made at a 234

frequency of 20 Hz and processed to half-hourly averages using the eth-flux EC software for 235

the forest and cropland (Mauder et al 2008) or a comparable custom-made EC software for 236

the grassland (Ammann et al 2007) Post-processing included corrections for damping losses 237

(Eugster and Senn 1995) air density fluctuations (Webb et al 1980) data screening for 238

optical sensor contamination stationary or low turbulence conditions (Foken and Wichura 239

1996) and statistical outliers Standardized gap filling and partitioning of CO2 fluxes was 240

performed using the methodology from Lasslop et al (2010) for the forest and cropland and 241

using the methodology from Falge et al (2001) for the grassland Measurements of 242

photosynthetic photon flux density (PPFD) quantified in micromol m-2 s-1 were converted to 243

radiance units (mW m-2 sr-1 nm-1) representing the photosynthetic active radiation (PAR) to 244

be consistent with Monteithrsquos LUE concept and the SIF radiance units 245

246

27 Leaf area index measurements 247

The Leaf Area Index (LAI) of the grassland and cropland was measured at the time of 248

APEX flights using an optical non-destructive approach based on LAI-2000 leaf area meter 249

observations (Li-Cor Lincoln USA) The LAI of the forest was estimated using digital 250

11

hemispherical photographs at two subplots following the VALERIE sampling scheme while 251

subsequent analysis were applied to obtain the LAI (Schneider et al 2014) 252

253

28 Statistical analysis of SIF760 - GPP relationship 254

SIF760 values were manually extracted from APEX-SIF760 maps for areas representing 255

the footprint of eddy-flux measurements (Figure 1) For the perennial grassland and cropland 256

we selected all pixels with SIF760 values ge 0 mW m-2 sr-1 nm-1 in respective fields although the 257

EC tower footprint can be slightly smaller than the entire field The increased number of 258

resulting pixels allows reducing data noise and covering small spatial variations within the 259

footprint Using the pixels of the entire field is justified by the large homogeneity of the 260

underlying fields (Figure 1A) For the forest site Laegeren all pixels representing an area that 261

covered 70 of all footprint occurrences were selected irrespective of spatial footprint 262

changes due to changing wind conditions Using the same area for all years reduces variations 263

in SIF due to changes in biomass and canopy structure but also reduces the representativeness 264

of selected pixels for actual spatial EC footprints 265

We used a hyperbolic model to relate GPP to SIF760 and to both vegetation indices 266

Although the hyperbolic model only empirically approximates the data behavior it is 267

supported by theoretical arguments for the SIF760-GPP relationship that are outlined hereafter 268

The photosynthetic efficiency LUEp is often described as a hyperbolic function of APAR 269

following the Michaelis-Menten theory (Michaelis and Menten 1913) 270

∙123467899 (5) 271

where c is a coefficient in radiance units and LUEpmax c represents GPPmax Following Eqs 272

(1) and (5) GPP rises asymptotically with APAR to GPPmax This equation is used in many 273

models for partitioning GPP eg Falge et al (2001) Substituting LUEp from Eq (5) and 274

APAR = SIF760 (LUEf fesc) from Eq (2) into Eq (1) results in 275

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

6

estimates for GPP at ecosystem or continental scales (Garbulsky et al 2014 Guanter et al 126

2012 Parazoo et al 2014) 127

Considering the above listed mechanisms several aspects need to be addressed to 128

further exploit SIF as a robust constraint for estimating GPP We therefore aim at 129

investigating the functional information content of SIF and its link to GPP considering three 130

structurally and physiologically contrasting ecosystems ie cropland perennial grassland 131

and mixed temperate forest We use an innovative combination of multi-temporal imaging 132

spectroscopy (IS) data EC flux tower observations and modeling approaches at the leaf and 133

canopy levels i) to assess the relationship between far red SIF retrieved in the O2-A band at 134

760 nm (SIF760) and GPP across ecosystems and ii) to investigate the impact of confounding 135

factors on the SIF760-GPP relationship ie temporal scaling effects and structural and 136

physiological interferences using a photosynthesis model Our findings contribute to a better 137

understanding of the information inherent in remotely measured SIF760 and its functional 138

relationship to GPP Aspects discussed will help bridging the gap between small scale studies 139

and observational attempts to estimate GPP globally 140

141

2 Methods 142

21 Study sites 143

We investigated three contrasting ecosystems located on the Central Swiss Plateau in 144

terms of structure heterogeneity species composition and annual productivity Two of these 145

ecosystems were collocated in the agricultural area near the town of Oensingen (47deg1711rdquo N 146

7deg4401rdquo E 452 masl Figure 1A) This area is characterized by relatively small agricultural 147

parcels with grassland clover fallow cropping bean maize rapeseed pea sugar beet winter 148

barley and winter wheat rotating as dominant crops Two grassland fields differently 149

managed in terms of species composition fertilization and harvesting activities (Ammann et 150

7

al 2007) were investigated as representatives of the ecosystem type perennial grassland as 151

well as a cropland (one field with arable crop rotation) for the ecosystem type cropland The 152

forest area (47deg2842rdquo N 8deg2152rdquo E Figure 1B) is located on the south-facing slope of the 153

Laegeren mountain northwest of the city of Zurich The temperate mixed forest is 154

characterized by a relatively high species diversity and a complex canopy structure with 155

beech ash sycamore and spruce being the dominant species (Eugster et al 2007 Schneider 156

et al 2014) 157

158

lt Figure 1 gt 159

160

Both test sites are well instrumented (ie with eddy-covariance flux towers 161

micrometeorological stations) and were extensively sampled during several airborne 162

campaigns between 2009 and 2013 (Table 1) 163

164

lt Table 1 gt 165

166

22 Airborne spectroscopy 167

All three study sites were measured with the Airborne Prism EXperiment (APEX) 168

sensor on several days between 2009 and 2013 (Table 1) APEX is an airborne dispersive 169

pushbroom imaging spectrometer covering the 400-2500 nm spectral region in 313 narrow 170

continuous spectral bands APEX allows capturing the O2-A atmospheric absorption feature 171

and thus retrieving SIF760 with a spectral sampling interval of 45 nm a full width at half 172

maximum (FWHM) of 57 nm and a signal to noise ratio of approximately 450 in and 800 173

outside of the O2-A feature for a 50 reflective target and a sun zenith angle of 30deg (Jehle et 174

al 2010 Schaepman et al 2015) 175

8

Radiance calibration of APEX images is performed using the approach of Hueni et al 176

(2009) Multitemporal comparison of images is secured by using consistent calibration and 177

processing versions Precise retrievals of SIF typically require data measured in sub-178

nanometer spectral resolution However the feasibility of using APEX like instruments (ie 179

FWHM ~5nm) for SIF760 retrievals was theoretically assessed in Damm et al (2011) and 180

experimentally demonstrated in recent work (Zarco-Tejada et al 2009 Zarco-Tejada et al 181

2013) The use of calibrated APEX radiance data for the fluorescence retrieval is described in 182

Section 24 and Appendix-B Vegetation indices (see Section 25) were calculated using 183

Hemispherical-Conical-Reflectance-Factor data obtained after atmospheric correction using 184

ATCOR-4 (Richter and Schlaumlpfer 2002) Resulting SIF760 and vegetation indices were 185

projected in the same map projection using parametric geocoding (Schlaumlpfer and Richter 186

2002) Details about APEX data calibration are listed in Appendix-A 187

188

23 Field spectroscopy 189

ASD (PANalytics Boulder US) field spectrometer data were acquired for validation 190

purposes within one hour of the APEX overflight during four campaigns In total 23 191

agricultural fields covering various crops including carrot maize pea rye sugar beet wheat 192

white clover were systematically sampled by measuring reflected and emitted radiances in 193

four homogeneous 3 m2 plots per field (Table 2) Incident light was quantified before and 194

after measuring each plot using a Spectralon reference panel 195

196

lt Table 2 gt 197

198

24 SIF760 retrieval 199

Emitted SIF was analytically separated from the reflected radiance flux measured with 200

ground and airborne-based spectrometers (see Appendix-B) We applied the Fraunhofer Line 201

9

Depth (FLD) approach (Plascyk 1975) which serves as a standard for SIF retrievals using 202

medium resolution instruments (Meroni et al 2009) The FLD method uses atmospheric 203

absorption bands characterized by lower incident light compared to wavelength regions 204

outside of these bands This configuration allows evaluating the in-filling of the absorption 205

bands by SIF (Damm et al 2014) In this study we used the broad O2-A oxygen absorption 206

band around 760 nm for the SIF760 retrieval For the ground data surface irradiance was 207

estimated from white reference measurements while perturbing atmospheric absorption and 208

scattering effects were assumed to be negligible on the 1 m pathway between canopy and 209

sensor (cf Damm et al (2014) for details on the set-up for SIF760 retrieval on the ground) 210

For the airborne measurements we compensated atmospheric absorption and scattering using 211

the atmospheric radiative transfer model MODTRAN5 (Berk et al 2005) Details of the 212

SIF760 retrievals using airborne data are provided in Appendix-B 213

214

25 Derivation of additional spectral indices 215

The relationship between two vegetation indices and GPP was investigated in parallel 216

to judge the performance of the new remote observation of SIF compared to commonly used 217

remote sensing approaches The photochemical reflectance index (PRI) (Gamon et al 1992) 218

was found to be sensitive to pigment changes related to the xanthophyll cycle and is 219

frequently used to quantify LUEp and GPP (Cheng et al 2013 Drolet et al 2008 Hilker et al 220

2009) The PRI was calculated as 221

(3) 222

while Rλ accounts for the surface reflectance at the wavelength λ 223

We further tested the soil adjusted Modified Chlorophyll Absorption Ratio Index 2 224

(MCARI2) (Haboudane et al 2004) that was found to be strongly related to green LAI thus 225

potential photosynthesis and is calculated as 226

10

2 $amp($)$+-amp0 (4) 227

228

26 Eddy-covariance measurements 229

Continuous turbulent fluxes of CO2 were measured at the agricultural and forest sites 230

between 2009 and 2013 using the EC technique (Baldocchi 2003) At each site the EC 231

instrumentation consisted of an open-path infrared gas analyzer (IRGA) (model LI-7500 LI-232

COR Inc Lincoln NE USA) and a three-dimensional sonic anemometer-thermometer 233

(models HS R3 and R2 Gill Instruments Lymington UK) EC measurements were made at a 234

frequency of 20 Hz and processed to half-hourly averages using the eth-flux EC software for 235

the forest and cropland (Mauder et al 2008) or a comparable custom-made EC software for 236

the grassland (Ammann et al 2007) Post-processing included corrections for damping losses 237

(Eugster and Senn 1995) air density fluctuations (Webb et al 1980) data screening for 238

optical sensor contamination stationary or low turbulence conditions (Foken and Wichura 239

1996) and statistical outliers Standardized gap filling and partitioning of CO2 fluxes was 240

performed using the methodology from Lasslop et al (2010) for the forest and cropland and 241

using the methodology from Falge et al (2001) for the grassland Measurements of 242

photosynthetic photon flux density (PPFD) quantified in micromol m-2 s-1 were converted to 243

radiance units (mW m-2 sr-1 nm-1) representing the photosynthetic active radiation (PAR) to 244

be consistent with Monteithrsquos LUE concept and the SIF radiance units 245

246

27 Leaf area index measurements 247

The Leaf Area Index (LAI) of the grassland and cropland was measured at the time of 248

APEX flights using an optical non-destructive approach based on LAI-2000 leaf area meter 249

observations (Li-Cor Lincoln USA) The LAI of the forest was estimated using digital 250

11

hemispherical photographs at two subplots following the VALERIE sampling scheme while 251

subsequent analysis were applied to obtain the LAI (Schneider et al 2014) 252

253

28 Statistical analysis of SIF760 - GPP relationship 254

SIF760 values were manually extracted from APEX-SIF760 maps for areas representing 255

the footprint of eddy-flux measurements (Figure 1) For the perennial grassland and cropland 256

we selected all pixels with SIF760 values ge 0 mW m-2 sr-1 nm-1 in respective fields although the 257

EC tower footprint can be slightly smaller than the entire field The increased number of 258

resulting pixels allows reducing data noise and covering small spatial variations within the 259

footprint Using the pixels of the entire field is justified by the large homogeneity of the 260

underlying fields (Figure 1A) For the forest site Laegeren all pixels representing an area that 261

covered 70 of all footprint occurrences were selected irrespective of spatial footprint 262

changes due to changing wind conditions Using the same area for all years reduces variations 263

in SIF due to changes in biomass and canopy structure but also reduces the representativeness 264

of selected pixels for actual spatial EC footprints 265

We used a hyperbolic model to relate GPP to SIF760 and to both vegetation indices 266

Although the hyperbolic model only empirically approximates the data behavior it is 267

supported by theoretical arguments for the SIF760-GPP relationship that are outlined hereafter 268

The photosynthetic efficiency LUEp is often described as a hyperbolic function of APAR 269

following the Michaelis-Menten theory (Michaelis and Menten 1913) 270

∙123467899 (5) 271

where c is a coefficient in radiance units and LUEpmax c represents GPPmax Following Eqs 272

(1) and (5) GPP rises asymptotically with APAR to GPPmax This equation is used in many 273

models for partitioning GPP eg Falge et al (2001) Substituting LUEp from Eq (5) and 274

APAR = SIF760 (LUEf fesc) from Eq (2) into Eq (1) results in 275

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

7

al 2007) were investigated as representatives of the ecosystem type perennial grassland as 151

well as a cropland (one field with arable crop rotation) for the ecosystem type cropland The 152

forest area (47deg2842rdquo N 8deg2152rdquo E Figure 1B) is located on the south-facing slope of the 153

Laegeren mountain northwest of the city of Zurich The temperate mixed forest is 154

characterized by a relatively high species diversity and a complex canopy structure with 155

beech ash sycamore and spruce being the dominant species (Eugster et al 2007 Schneider 156

et al 2014) 157

158

lt Figure 1 gt 159

160

Both test sites are well instrumented (ie with eddy-covariance flux towers 161

micrometeorological stations) and were extensively sampled during several airborne 162

campaigns between 2009 and 2013 (Table 1) 163

164

lt Table 1 gt 165

166

22 Airborne spectroscopy 167

All three study sites were measured with the Airborne Prism EXperiment (APEX) 168

sensor on several days between 2009 and 2013 (Table 1) APEX is an airborne dispersive 169

pushbroom imaging spectrometer covering the 400-2500 nm spectral region in 313 narrow 170

continuous spectral bands APEX allows capturing the O2-A atmospheric absorption feature 171

and thus retrieving SIF760 with a spectral sampling interval of 45 nm a full width at half 172

maximum (FWHM) of 57 nm and a signal to noise ratio of approximately 450 in and 800 173

outside of the O2-A feature for a 50 reflective target and a sun zenith angle of 30deg (Jehle et 174

al 2010 Schaepman et al 2015) 175

8

Radiance calibration of APEX images is performed using the approach of Hueni et al 176

(2009) Multitemporal comparison of images is secured by using consistent calibration and 177

processing versions Precise retrievals of SIF typically require data measured in sub-178

nanometer spectral resolution However the feasibility of using APEX like instruments (ie 179

FWHM ~5nm) for SIF760 retrievals was theoretically assessed in Damm et al (2011) and 180

experimentally demonstrated in recent work (Zarco-Tejada et al 2009 Zarco-Tejada et al 181

2013) The use of calibrated APEX radiance data for the fluorescence retrieval is described in 182

Section 24 and Appendix-B Vegetation indices (see Section 25) were calculated using 183

Hemispherical-Conical-Reflectance-Factor data obtained after atmospheric correction using 184

ATCOR-4 (Richter and Schlaumlpfer 2002) Resulting SIF760 and vegetation indices were 185

projected in the same map projection using parametric geocoding (Schlaumlpfer and Richter 186

2002) Details about APEX data calibration are listed in Appendix-A 187

188

23 Field spectroscopy 189

ASD (PANalytics Boulder US) field spectrometer data were acquired for validation 190

purposes within one hour of the APEX overflight during four campaigns In total 23 191

agricultural fields covering various crops including carrot maize pea rye sugar beet wheat 192

white clover were systematically sampled by measuring reflected and emitted radiances in 193

four homogeneous 3 m2 plots per field (Table 2) Incident light was quantified before and 194

after measuring each plot using a Spectralon reference panel 195

196

lt Table 2 gt 197

198

24 SIF760 retrieval 199

Emitted SIF was analytically separated from the reflected radiance flux measured with 200

ground and airborne-based spectrometers (see Appendix-B) We applied the Fraunhofer Line 201

9

Depth (FLD) approach (Plascyk 1975) which serves as a standard for SIF retrievals using 202

medium resolution instruments (Meroni et al 2009) The FLD method uses atmospheric 203

absorption bands characterized by lower incident light compared to wavelength regions 204

outside of these bands This configuration allows evaluating the in-filling of the absorption 205

bands by SIF (Damm et al 2014) In this study we used the broad O2-A oxygen absorption 206

band around 760 nm for the SIF760 retrieval For the ground data surface irradiance was 207

estimated from white reference measurements while perturbing atmospheric absorption and 208

scattering effects were assumed to be negligible on the 1 m pathway between canopy and 209

sensor (cf Damm et al (2014) for details on the set-up for SIF760 retrieval on the ground) 210

For the airborne measurements we compensated atmospheric absorption and scattering using 211

the atmospheric radiative transfer model MODTRAN5 (Berk et al 2005) Details of the 212

SIF760 retrievals using airborne data are provided in Appendix-B 213

214

25 Derivation of additional spectral indices 215

The relationship between two vegetation indices and GPP was investigated in parallel 216

to judge the performance of the new remote observation of SIF compared to commonly used 217

remote sensing approaches The photochemical reflectance index (PRI) (Gamon et al 1992) 218

was found to be sensitive to pigment changes related to the xanthophyll cycle and is 219

frequently used to quantify LUEp and GPP (Cheng et al 2013 Drolet et al 2008 Hilker et al 220

2009) The PRI was calculated as 221

(3) 222

while Rλ accounts for the surface reflectance at the wavelength λ 223

We further tested the soil adjusted Modified Chlorophyll Absorption Ratio Index 2 224

(MCARI2) (Haboudane et al 2004) that was found to be strongly related to green LAI thus 225

potential photosynthesis and is calculated as 226

10

2 $amp($)$+-amp0 (4) 227

228

26 Eddy-covariance measurements 229

Continuous turbulent fluxes of CO2 were measured at the agricultural and forest sites 230

between 2009 and 2013 using the EC technique (Baldocchi 2003) At each site the EC 231

instrumentation consisted of an open-path infrared gas analyzer (IRGA) (model LI-7500 LI-232

COR Inc Lincoln NE USA) and a three-dimensional sonic anemometer-thermometer 233

(models HS R3 and R2 Gill Instruments Lymington UK) EC measurements were made at a 234

frequency of 20 Hz and processed to half-hourly averages using the eth-flux EC software for 235

the forest and cropland (Mauder et al 2008) or a comparable custom-made EC software for 236

the grassland (Ammann et al 2007) Post-processing included corrections for damping losses 237

(Eugster and Senn 1995) air density fluctuations (Webb et al 1980) data screening for 238

optical sensor contamination stationary or low turbulence conditions (Foken and Wichura 239

1996) and statistical outliers Standardized gap filling and partitioning of CO2 fluxes was 240

performed using the methodology from Lasslop et al (2010) for the forest and cropland and 241

using the methodology from Falge et al (2001) for the grassland Measurements of 242

photosynthetic photon flux density (PPFD) quantified in micromol m-2 s-1 were converted to 243

radiance units (mW m-2 sr-1 nm-1) representing the photosynthetic active radiation (PAR) to 244

be consistent with Monteithrsquos LUE concept and the SIF radiance units 245

246

27 Leaf area index measurements 247

The Leaf Area Index (LAI) of the grassland and cropland was measured at the time of 248

APEX flights using an optical non-destructive approach based on LAI-2000 leaf area meter 249

observations (Li-Cor Lincoln USA) The LAI of the forest was estimated using digital 250

11

hemispherical photographs at two subplots following the VALERIE sampling scheme while 251

subsequent analysis were applied to obtain the LAI (Schneider et al 2014) 252

253

28 Statistical analysis of SIF760 - GPP relationship 254

SIF760 values were manually extracted from APEX-SIF760 maps for areas representing 255

the footprint of eddy-flux measurements (Figure 1) For the perennial grassland and cropland 256

we selected all pixels with SIF760 values ge 0 mW m-2 sr-1 nm-1 in respective fields although the 257

EC tower footprint can be slightly smaller than the entire field The increased number of 258

resulting pixels allows reducing data noise and covering small spatial variations within the 259

footprint Using the pixels of the entire field is justified by the large homogeneity of the 260

underlying fields (Figure 1A) For the forest site Laegeren all pixels representing an area that 261

covered 70 of all footprint occurrences were selected irrespective of spatial footprint 262

changes due to changing wind conditions Using the same area for all years reduces variations 263

in SIF due to changes in biomass and canopy structure but also reduces the representativeness 264

of selected pixels for actual spatial EC footprints 265

We used a hyperbolic model to relate GPP to SIF760 and to both vegetation indices 266

Although the hyperbolic model only empirically approximates the data behavior it is 267

supported by theoretical arguments for the SIF760-GPP relationship that are outlined hereafter 268

The photosynthetic efficiency LUEp is often described as a hyperbolic function of APAR 269

following the Michaelis-Menten theory (Michaelis and Menten 1913) 270

∙123467899 (5) 271

where c is a coefficient in radiance units and LUEpmax c represents GPPmax Following Eqs 272

(1) and (5) GPP rises asymptotically with APAR to GPPmax This equation is used in many 273

models for partitioning GPP eg Falge et al (2001) Substituting LUEp from Eq (5) and 274

APAR = SIF760 (LUEf fesc) from Eq (2) into Eq (1) results in 275

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

8

Radiance calibration of APEX images is performed using the approach of Hueni et al 176

(2009) Multitemporal comparison of images is secured by using consistent calibration and 177

processing versions Precise retrievals of SIF typically require data measured in sub-178

nanometer spectral resolution However the feasibility of using APEX like instruments (ie 179

FWHM ~5nm) for SIF760 retrievals was theoretically assessed in Damm et al (2011) and 180

experimentally demonstrated in recent work (Zarco-Tejada et al 2009 Zarco-Tejada et al 181

2013) The use of calibrated APEX radiance data for the fluorescence retrieval is described in 182

Section 24 and Appendix-B Vegetation indices (see Section 25) were calculated using 183

Hemispherical-Conical-Reflectance-Factor data obtained after atmospheric correction using 184

ATCOR-4 (Richter and Schlaumlpfer 2002) Resulting SIF760 and vegetation indices were 185

projected in the same map projection using parametric geocoding (Schlaumlpfer and Richter 186

2002) Details about APEX data calibration are listed in Appendix-A 187

188

23 Field spectroscopy 189

ASD (PANalytics Boulder US) field spectrometer data were acquired for validation 190

purposes within one hour of the APEX overflight during four campaigns In total 23 191

agricultural fields covering various crops including carrot maize pea rye sugar beet wheat 192

white clover were systematically sampled by measuring reflected and emitted radiances in 193

four homogeneous 3 m2 plots per field (Table 2) Incident light was quantified before and 194

after measuring each plot using a Spectralon reference panel 195

196

lt Table 2 gt 197

198

24 SIF760 retrieval 199

Emitted SIF was analytically separated from the reflected radiance flux measured with 200

ground and airborne-based spectrometers (see Appendix-B) We applied the Fraunhofer Line 201

9

Depth (FLD) approach (Plascyk 1975) which serves as a standard for SIF retrievals using 202

medium resolution instruments (Meroni et al 2009) The FLD method uses atmospheric 203

absorption bands characterized by lower incident light compared to wavelength regions 204

outside of these bands This configuration allows evaluating the in-filling of the absorption 205

bands by SIF (Damm et al 2014) In this study we used the broad O2-A oxygen absorption 206

band around 760 nm for the SIF760 retrieval For the ground data surface irradiance was 207

estimated from white reference measurements while perturbing atmospheric absorption and 208

scattering effects were assumed to be negligible on the 1 m pathway between canopy and 209

sensor (cf Damm et al (2014) for details on the set-up for SIF760 retrieval on the ground) 210

For the airborne measurements we compensated atmospheric absorption and scattering using 211

the atmospheric radiative transfer model MODTRAN5 (Berk et al 2005) Details of the 212

SIF760 retrievals using airborne data are provided in Appendix-B 213

214

25 Derivation of additional spectral indices 215

The relationship between two vegetation indices and GPP was investigated in parallel 216

to judge the performance of the new remote observation of SIF compared to commonly used 217

remote sensing approaches The photochemical reflectance index (PRI) (Gamon et al 1992) 218

was found to be sensitive to pigment changes related to the xanthophyll cycle and is 219

frequently used to quantify LUEp and GPP (Cheng et al 2013 Drolet et al 2008 Hilker et al 220

2009) The PRI was calculated as 221

(3) 222

while Rλ accounts for the surface reflectance at the wavelength λ 223

We further tested the soil adjusted Modified Chlorophyll Absorption Ratio Index 2 224

(MCARI2) (Haboudane et al 2004) that was found to be strongly related to green LAI thus 225

potential photosynthesis and is calculated as 226

10

2 $amp($)$+-amp0 (4) 227

228

26 Eddy-covariance measurements 229

Continuous turbulent fluxes of CO2 were measured at the agricultural and forest sites 230

between 2009 and 2013 using the EC technique (Baldocchi 2003) At each site the EC 231

instrumentation consisted of an open-path infrared gas analyzer (IRGA) (model LI-7500 LI-232

COR Inc Lincoln NE USA) and a three-dimensional sonic anemometer-thermometer 233

(models HS R3 and R2 Gill Instruments Lymington UK) EC measurements were made at a 234

frequency of 20 Hz and processed to half-hourly averages using the eth-flux EC software for 235

the forest and cropland (Mauder et al 2008) or a comparable custom-made EC software for 236

the grassland (Ammann et al 2007) Post-processing included corrections for damping losses 237

(Eugster and Senn 1995) air density fluctuations (Webb et al 1980) data screening for 238

optical sensor contamination stationary or low turbulence conditions (Foken and Wichura 239

1996) and statistical outliers Standardized gap filling and partitioning of CO2 fluxes was 240

performed using the methodology from Lasslop et al (2010) for the forest and cropland and 241

using the methodology from Falge et al (2001) for the grassland Measurements of 242

photosynthetic photon flux density (PPFD) quantified in micromol m-2 s-1 were converted to 243

radiance units (mW m-2 sr-1 nm-1) representing the photosynthetic active radiation (PAR) to 244

be consistent with Monteithrsquos LUE concept and the SIF radiance units 245

246

27 Leaf area index measurements 247

The Leaf Area Index (LAI) of the grassland and cropland was measured at the time of 248

APEX flights using an optical non-destructive approach based on LAI-2000 leaf area meter 249

observations (Li-Cor Lincoln USA) The LAI of the forest was estimated using digital 250

11

hemispherical photographs at two subplots following the VALERIE sampling scheme while 251

subsequent analysis were applied to obtain the LAI (Schneider et al 2014) 252

253

28 Statistical analysis of SIF760 - GPP relationship 254

SIF760 values were manually extracted from APEX-SIF760 maps for areas representing 255

the footprint of eddy-flux measurements (Figure 1) For the perennial grassland and cropland 256

we selected all pixels with SIF760 values ge 0 mW m-2 sr-1 nm-1 in respective fields although the 257

EC tower footprint can be slightly smaller than the entire field The increased number of 258

resulting pixels allows reducing data noise and covering small spatial variations within the 259

footprint Using the pixels of the entire field is justified by the large homogeneity of the 260

underlying fields (Figure 1A) For the forest site Laegeren all pixels representing an area that 261

covered 70 of all footprint occurrences were selected irrespective of spatial footprint 262

changes due to changing wind conditions Using the same area for all years reduces variations 263

in SIF due to changes in biomass and canopy structure but also reduces the representativeness 264

of selected pixels for actual spatial EC footprints 265

We used a hyperbolic model to relate GPP to SIF760 and to both vegetation indices 266

Although the hyperbolic model only empirically approximates the data behavior it is 267

supported by theoretical arguments for the SIF760-GPP relationship that are outlined hereafter 268

The photosynthetic efficiency LUEp is often described as a hyperbolic function of APAR 269

following the Michaelis-Menten theory (Michaelis and Menten 1913) 270

∙123467899 (5) 271

where c is a coefficient in radiance units and LUEpmax c represents GPPmax Following Eqs 272

(1) and (5) GPP rises asymptotically with APAR to GPPmax This equation is used in many 273

models for partitioning GPP eg Falge et al (2001) Substituting LUEp from Eq (5) and 274

APAR = SIF760 (LUEf fesc) from Eq (2) into Eq (1) results in 275

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

9

Depth (FLD) approach (Plascyk 1975) which serves as a standard for SIF retrievals using 202

medium resolution instruments (Meroni et al 2009) The FLD method uses atmospheric 203

absorption bands characterized by lower incident light compared to wavelength regions 204

outside of these bands This configuration allows evaluating the in-filling of the absorption 205

bands by SIF (Damm et al 2014) In this study we used the broad O2-A oxygen absorption 206

band around 760 nm for the SIF760 retrieval For the ground data surface irradiance was 207

estimated from white reference measurements while perturbing atmospheric absorption and 208

scattering effects were assumed to be negligible on the 1 m pathway between canopy and 209

sensor (cf Damm et al (2014) for details on the set-up for SIF760 retrieval on the ground) 210

For the airborne measurements we compensated atmospheric absorption and scattering using 211

the atmospheric radiative transfer model MODTRAN5 (Berk et al 2005) Details of the 212

SIF760 retrievals using airborne data are provided in Appendix-B 213

214

25 Derivation of additional spectral indices 215

The relationship between two vegetation indices and GPP was investigated in parallel 216

to judge the performance of the new remote observation of SIF compared to commonly used 217

remote sensing approaches The photochemical reflectance index (PRI) (Gamon et al 1992) 218

was found to be sensitive to pigment changes related to the xanthophyll cycle and is 219

frequently used to quantify LUEp and GPP (Cheng et al 2013 Drolet et al 2008 Hilker et al 220

2009) The PRI was calculated as 221

(3) 222

while Rλ accounts for the surface reflectance at the wavelength λ 223

We further tested the soil adjusted Modified Chlorophyll Absorption Ratio Index 2 224

(MCARI2) (Haboudane et al 2004) that was found to be strongly related to green LAI thus 225

potential photosynthesis and is calculated as 226

10

2 $amp($)$+-amp0 (4) 227

228

26 Eddy-covariance measurements 229

Continuous turbulent fluxes of CO2 were measured at the agricultural and forest sites 230

between 2009 and 2013 using the EC technique (Baldocchi 2003) At each site the EC 231

instrumentation consisted of an open-path infrared gas analyzer (IRGA) (model LI-7500 LI-232

COR Inc Lincoln NE USA) and a three-dimensional sonic anemometer-thermometer 233

(models HS R3 and R2 Gill Instruments Lymington UK) EC measurements were made at a 234

frequency of 20 Hz and processed to half-hourly averages using the eth-flux EC software for 235

the forest and cropland (Mauder et al 2008) or a comparable custom-made EC software for 236

the grassland (Ammann et al 2007) Post-processing included corrections for damping losses 237

(Eugster and Senn 1995) air density fluctuations (Webb et al 1980) data screening for 238

optical sensor contamination stationary or low turbulence conditions (Foken and Wichura 239

1996) and statistical outliers Standardized gap filling and partitioning of CO2 fluxes was 240

performed using the methodology from Lasslop et al (2010) for the forest and cropland and 241

using the methodology from Falge et al (2001) for the grassland Measurements of 242

photosynthetic photon flux density (PPFD) quantified in micromol m-2 s-1 were converted to 243

radiance units (mW m-2 sr-1 nm-1) representing the photosynthetic active radiation (PAR) to 244

be consistent with Monteithrsquos LUE concept and the SIF radiance units 245

246

27 Leaf area index measurements 247

The Leaf Area Index (LAI) of the grassland and cropland was measured at the time of 248

APEX flights using an optical non-destructive approach based on LAI-2000 leaf area meter 249

observations (Li-Cor Lincoln USA) The LAI of the forest was estimated using digital 250

11

hemispherical photographs at two subplots following the VALERIE sampling scheme while 251

subsequent analysis were applied to obtain the LAI (Schneider et al 2014) 252

253

28 Statistical analysis of SIF760 - GPP relationship 254

SIF760 values were manually extracted from APEX-SIF760 maps for areas representing 255

the footprint of eddy-flux measurements (Figure 1) For the perennial grassland and cropland 256

we selected all pixels with SIF760 values ge 0 mW m-2 sr-1 nm-1 in respective fields although the 257

EC tower footprint can be slightly smaller than the entire field The increased number of 258

resulting pixels allows reducing data noise and covering small spatial variations within the 259

footprint Using the pixels of the entire field is justified by the large homogeneity of the 260

underlying fields (Figure 1A) For the forest site Laegeren all pixels representing an area that 261

covered 70 of all footprint occurrences were selected irrespective of spatial footprint 262

changes due to changing wind conditions Using the same area for all years reduces variations 263

in SIF due to changes in biomass and canopy structure but also reduces the representativeness 264

of selected pixels for actual spatial EC footprints 265

We used a hyperbolic model to relate GPP to SIF760 and to both vegetation indices 266

Although the hyperbolic model only empirically approximates the data behavior it is 267

supported by theoretical arguments for the SIF760-GPP relationship that are outlined hereafter 268

The photosynthetic efficiency LUEp is often described as a hyperbolic function of APAR 269

following the Michaelis-Menten theory (Michaelis and Menten 1913) 270

∙123467899 (5) 271

where c is a coefficient in radiance units and LUEpmax c represents GPPmax Following Eqs 272

(1) and (5) GPP rises asymptotically with APAR to GPPmax This equation is used in many 273

models for partitioning GPP eg Falge et al (2001) Substituting LUEp from Eq (5) and 274

APAR = SIF760 (LUEf fesc) from Eq (2) into Eq (1) results in 275

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

10

2 $amp($)$+-amp0 (4) 227

228

26 Eddy-covariance measurements 229

Continuous turbulent fluxes of CO2 were measured at the agricultural and forest sites 230

between 2009 and 2013 using the EC technique (Baldocchi 2003) At each site the EC 231

instrumentation consisted of an open-path infrared gas analyzer (IRGA) (model LI-7500 LI-232

COR Inc Lincoln NE USA) and a three-dimensional sonic anemometer-thermometer 233

(models HS R3 and R2 Gill Instruments Lymington UK) EC measurements were made at a 234

frequency of 20 Hz and processed to half-hourly averages using the eth-flux EC software for 235

the forest and cropland (Mauder et al 2008) or a comparable custom-made EC software for 236

the grassland (Ammann et al 2007) Post-processing included corrections for damping losses 237

(Eugster and Senn 1995) air density fluctuations (Webb et al 1980) data screening for 238

optical sensor contamination stationary or low turbulence conditions (Foken and Wichura 239

1996) and statistical outliers Standardized gap filling and partitioning of CO2 fluxes was 240

performed using the methodology from Lasslop et al (2010) for the forest and cropland and 241

using the methodology from Falge et al (2001) for the grassland Measurements of 242

photosynthetic photon flux density (PPFD) quantified in micromol m-2 s-1 were converted to 243

radiance units (mW m-2 sr-1 nm-1) representing the photosynthetic active radiation (PAR) to 244

be consistent with Monteithrsquos LUE concept and the SIF radiance units 245

246

27 Leaf area index measurements 247

The Leaf Area Index (LAI) of the grassland and cropland was measured at the time of 248

APEX flights using an optical non-destructive approach based on LAI-2000 leaf area meter 249

observations (Li-Cor Lincoln USA) The LAI of the forest was estimated using digital 250

11

hemispherical photographs at two subplots following the VALERIE sampling scheme while 251

subsequent analysis were applied to obtain the LAI (Schneider et al 2014) 252

253

28 Statistical analysis of SIF760 - GPP relationship 254

SIF760 values were manually extracted from APEX-SIF760 maps for areas representing 255

the footprint of eddy-flux measurements (Figure 1) For the perennial grassland and cropland 256

we selected all pixels with SIF760 values ge 0 mW m-2 sr-1 nm-1 in respective fields although the 257

EC tower footprint can be slightly smaller than the entire field The increased number of 258

resulting pixels allows reducing data noise and covering small spatial variations within the 259

footprint Using the pixels of the entire field is justified by the large homogeneity of the 260

underlying fields (Figure 1A) For the forest site Laegeren all pixels representing an area that 261

covered 70 of all footprint occurrences were selected irrespective of spatial footprint 262

changes due to changing wind conditions Using the same area for all years reduces variations 263

in SIF due to changes in biomass and canopy structure but also reduces the representativeness 264

of selected pixels for actual spatial EC footprints 265

We used a hyperbolic model to relate GPP to SIF760 and to both vegetation indices 266

Although the hyperbolic model only empirically approximates the data behavior it is 267

supported by theoretical arguments for the SIF760-GPP relationship that are outlined hereafter 268

The photosynthetic efficiency LUEp is often described as a hyperbolic function of APAR 269

following the Michaelis-Menten theory (Michaelis and Menten 1913) 270

∙123467899 (5) 271

where c is a coefficient in radiance units and LUEpmax c represents GPPmax Following Eqs 272

(1) and (5) GPP rises asymptotically with APAR to GPPmax This equation is used in many 273

models for partitioning GPP eg Falge et al (2001) Substituting LUEp from Eq (5) and 274

APAR = SIF760 (LUEf fesc) from Eq (2) into Eq (1) results in 275

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

11

hemispherical photographs at two subplots following the VALERIE sampling scheme while 251

subsequent analysis were applied to obtain the LAI (Schneider et al 2014) 252

253

28 Statistical analysis of SIF760 - GPP relationship 254

SIF760 values were manually extracted from APEX-SIF760 maps for areas representing 255

the footprint of eddy-flux measurements (Figure 1) For the perennial grassland and cropland 256

we selected all pixels with SIF760 values ge 0 mW m-2 sr-1 nm-1 in respective fields although the 257

EC tower footprint can be slightly smaller than the entire field The increased number of 258

resulting pixels allows reducing data noise and covering small spatial variations within the 259

footprint Using the pixels of the entire field is justified by the large homogeneity of the 260

underlying fields (Figure 1A) For the forest site Laegeren all pixels representing an area that 261

covered 70 of all footprint occurrences were selected irrespective of spatial footprint 262

changes due to changing wind conditions Using the same area for all years reduces variations 263

in SIF due to changes in biomass and canopy structure but also reduces the representativeness 264

of selected pixels for actual spatial EC footprints 265

We used a hyperbolic model to relate GPP to SIF760 and to both vegetation indices 266

Although the hyperbolic model only empirically approximates the data behavior it is 267

supported by theoretical arguments for the SIF760-GPP relationship that are outlined hereafter 268

The photosynthetic efficiency LUEp is often described as a hyperbolic function of APAR 269

following the Michaelis-Menten theory (Michaelis and Menten 1913) 270

∙123467899 (5) 271

where c is a coefficient in radiance units and LUEpmax c represents GPPmax Following Eqs 272

(1) and (5) GPP rises asymptotically with APAR to GPPmax This equation is used in many 273

models for partitioning GPP eg Falge et al (2001) Substituting LUEp from Eq (5) and 274

APAR = SIF760 (LUEf fesc) from Eq (2) into Eq (1) results in 275

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

12

lt= gtampABCampD (6) 276

where the coefficient b represents cLUEffesc accounting for the difference between LUEp 277

and LUEf GPPmax is a value that is asymptotically reached representing the photosynthetic 278

capacity or GPP of the canopy at light saturation It must be noted that Eq (6) is a 279

simplification of the complex relationships between GPP and SIF760 and thus performs best if 280

c LUEf fesc remains constant or SIF760 corresponds to APAR A constant LUEf is a 281

reasonable first approximation for vegetation that is only moderately stressed meaning that 282

variations in LUEf are substantially smaller than variations in LUEp (van der Tol et al 2014) 283

Moderate stress relates to any kind of environmental conditions (warm temperature strong 284

irradiance reduced water availability) that cause only short term and slight changes in 285

photochemical and non-photochemical protections However c and fesc are known to vary 286

across ecosystems and LUEf changes under environmental stress In order to evaluate their 287

impacts on the SIF-GPP relationships as expressed in Eq (6) we carried out simulations with 288

the Soil-Canopy Observations of Photosynthesis and Energy balance model SCOPE version 289

153 (van der Tol et al 2009b) (see Section 29) The goodness of fit for the hyperbolic model 290

was quantified using non-linear regression analysis and was quantitatively described using the 291

root mean square error normalized by the mean of measured data (rRMSE) and the coefficient 292

of determination R2 defined as the correlation between data and the best fit-curve determined 293

by non-linear regression 294

295

29 SCOPE simulations 296

SCOPE uses semi-empirical models for LUEp (Collatz et al 1991) and LUEf (van der 297

Tol et al 2014) embedded in a radiative transfer scheme for the canopy (van der Tol et al 298

2009b) The radiative transfer model at leaf level (Fluspect) is an extension of PROSPECT 299

(Jacquemoud and Baret 1990) and includes a fluorescence component The radiative transfer 300

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

13

within the canopy is calculated following Verhoef (1984) in the optical domain (SAIL) 301

following Verhoef et al (2007) in the thermal domain and Miller et al (2005) for the 302

fluorescence emissions (FluorSAIL) SCOPE requires a number of vegetation structural and 303

physiological parameters as well as weather information as input The model simulates GPP 304

and spectrally distributes SIF with a spectral resolution and a spectral sampling interval of 10 305

nm each Simulated PPFD values (micromol m-2 s-1) were also converted to radiance units (PAR) 306

We carried out three simulations using SCOPE In the first simulation we calculated 307

leaf level LUEf and LUEp of a typical C3 cropland as function of varying APAR For the 308

second simulation at canopy level we calculated LUEf and LUEp as function of varying 309

APAR for differently structured canopies We took the standard SCOPE parameters for a 310

typical C3 cropland and varied APAR and the structural parameters LAI leaf inclination 311

angle (LIDF) and the fraction of brown pigments (Cs) Both simulation experiments provide 312

theoretical evidence on the relationships between LUEf and LUEp between SIF760 and GPP 313

and their respective relationships to APAR For the third canopy level simulation SCOPE 314

was initialized using time series of mid-day (solar noon) meteorological data (PAR air 315

temperature) a seasonal cycle of LAI and the modelrsquos default soil and vegetation parameters 316

for a C3 cropland (cf the SCOPE user manual) The purpose of this simulation was to 317

evaluate seasonal SIF760-GPP relationships at various temporal aggregation levels For better 318

readability we use the following notation through the manuscript SIF760 O for APEX-SIF 319

SIF760 L and SIF760 C for SCOPE simulated leaf and canopy SIF GPPEC for flux tower GPP 320

and GPPL GPPC to relate to simulated GPP at leaf and canopy level 321

322

3 Results 323

31 SIF760 maps 324

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

14

Two examples of SIF760 O maps obtained from APEX are shown in Figures 2 and 3 325

The appearance of SIF760 O emissions in vegetated areas demonstrates that measured SIF760 O 326

signals originated solely from vegetation and were not contaminated by non-fluorescent non-327

vegetation surfaces Retrieved SIF760 O values range between 0 and 3 mW m-2 sr-1 nm-1 and 328

showed a high agreement (r2 = 07 rRMSE = 289) with in-situ measured reference data 329

Details of the SIF760 O validation are discussed in Appendix-C 330

331

lt Figure 2 gt 332

lt Figure 3gt 333

334

32 Relationships of GPPEC with SIF760 O PRI and MCARI2 335

Relating instantaneous measurements of SIF760 O and GPPEC from several years 336

revealed an asymptotic relationship with saturating GPPEC in presence of moderate to high 337

SIF760 O (Figure 4A Table 3) This general behavior was consistent for all investigated 338

vegetation types The fitted hyperbolic model shows a moderate to high goodness of fit for the 339

grassland (R2 = 059 rRMSE = 271) cropland (R2 = 088 rRMSE = 35) and forest (R2 = 340

048 rRMSE = 1588) The saturation of GPPEC appeared earlier for the cropland (SIF760 O 341

~05 mW m-2 sr-1 nm-1) than for the grassland and forest (SIF760 O gt 10 mW m-2 sr-1 nm-1) An 342

independent assessment considering a linear model to describe the relationship between 343

SIF760 O and GPPEC revealed a consistently lower goodness of fit justifying the use of a 344

hyperbolic model (data not shown) 345

Statistical analysis showed that a hyperbolic model represents well the relationship 346

between both vegetation indices (PRI and MCARI2) and GPPEC for the grassland with a 347

notable variation in LAI (range between 05 and 48 m2 m-2) However a hyperbolic model 348

does not allow well representing the relationship between PRI and MCARI2 with GPPEC for 349

the forest and cropland both showing only a small variation in LAI (less than plusmn03 m2 m-2) 350

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

15

(Figure 4B-C Table 3) The goodness of fit for the hyperbolic model describing the PRI and 351

GPPEC data for grassland (R2 = 095 rRMSE = 86) was even higher compared to SIF760 O 352

but lower for forest (R2 = 009 rRMSE = 2179) and cropland (R2 = 0008 rRMSE = 353

1176) With MCARI2 the goodness of fit for the hyperbolic model for grassland was high 354

(R2 = 098 rRMSE = 55) but low for cropland (R2 = 0005 rRMSE = 1029) and forest 355

(R2 = 034 rRMSE = 1798) Again considering a linear relationship between both 356

vegetation indices and GPPEC revealed no consistent improvement compared to the use of a 357

hyperbolic model (data not shown) 358

359

lt Figure 4 gt 360

lt Table 3 gt 361

362

33 Assessment of factors determining SIF760 GPPEC relationships 363

The asymptotic relationship between SIF760 O and GPPEC (Figure 4A) represented 364

instantaneous measurements of different canopies at different times and days over several 365

years covering a range of different light conditions phenological stages of the forest and 366

various amounts of green biomass for both investigated grasslands (crosses in Figure 5A-C 367

that indicate EC measurements during APEX overpasses) Relationships between SIF760 O and 368

GPPEC are consequently a complex function of light absorption in the canopy (APAR) of the 369

efficiency of light utilization (ie expressed as photosynthetic capacity and quantum 370

efficiency of photosynthesis) and of structural effects including fractions of non-371

photosynthetic vegetation components Both grass canopies were measured before and after 372

mowing resulting in a large range in LAI (05-48 m2 m-2) and GPPEC (4-27 micromol m-2 s-1) for 373

similar PAR values The forest LAI of 50 m2 m-2 was more constant over the years due to the 374

consistent timing of flights however senescence (eg lower canopy chlorophyll content and 375

light absorption) might explain the decrease in GPPEC of 15 micromol m-2 s-1 in September for 376

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

16

comparable irradiance conditions (Figure 5C) Investigated crops (winter wheat barley and 377

pea) were characterized by small variations in LAI (22-25 m2 m-2) and chlorophyll contents 378

corresponding to small variations in GPPEC (5 micromol m-2 s-1) while observed under comparable 379

irradiance conditions During the time of observation the barley and wheat canopies were in 380

the ripening phase with a pronounced ear layer 381

382

lt Figure 5 gt 383

384

The validity of the asymptotic relationship between SIF760 O and GPPEC was tested 385

using simulations made with the SCOPE model We emphasize that the regression coefficient 386

b in Eq (6) includes terms that vary in the model (b= cmiddotLUEfmiddotfesc) the fluorescence efficiency 387

LUEf varies with irradiance conditions (predominantly APAR) the fraction fesc depends on 388

vegetation structure while c=GPPmaxLUEpmax varies between sites and over time during the 389

season We first analysed how LUEf and LUEp and consequently SIF760 and GPP vary at 390

leaf (ie SIF760 L and GPPL) and canopy levels (ie SIF760 C and GPPC) using the model of 391

van der Tol et al (2014) that is embedded in SCOPE For the leaf level Figure 6A shows a 392

decrease of LUEp with increasing APAR consistent with Eq (5) The relation between APAR 393

and LUEf appears more complex LUEf first increases with increasing APAR under low light 394

conditions reaches a climax at intermediate light and then decreases with increasing APAR 395

under higher APAR values (gt 50 mW m-2 sr-1 nm-1) (Figure 6B) The initial increase in LUEf 396

is caused by the decline in photochemical quenching of the excitons whereas the decrease of 397

LUEf is caused by the growing number of non-photochemical quenching traps that dissipate 398

light energy before it can be emitted as fluorescence light (van der Tol et al 2009a) It is 399

important to note that the range of LUEp is nearly an order of magnitude greater compared to 400

the range of LUEf (Figure 6C) Although LUEf is not constant we can still conclude that the 401

difference in both LUE terms causes an earlier saturation of leaf-level GPPL (Figure 6D) 402

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

17

compared to leaf-level SIF760 L (Figure 6E) with APAR leading to an asymptotic relationship 403

between SIF760 L and GPPL (Figure 6F) Our canopy level simulations confirm these trends 404

(Figure 7A-D) The in-field measured asymptotic relationships between SIF760 O and GPPEC 405

are consequently caused by the covariance of GPPEC and SIF760 O with APAR the 406

photosynthetic capacity with APAR and the different value ranges of LUEf and LUEp 407

Additionally our canopy level simulations suggest that structure impacts the SIF760 O - GPPEC 408

relationship Figure 7D shows an example that changing LAI leaf inclination angles or the 409

fraction of senescent plant material determine the curvature of the hyperbolic SIF760 C - GPPC 410

relationships 411

412

lt Figure 6 gt 413

lt Figure 7 gt 414

415

The modelled SIF760 LC - GPPLC curves (Figures 6D and 7D) also indicate that under 416

reduced light intensity GPP and SIF760 are almost linearly related while GPP tends to saturate 417

under higher light intensities Changes in APAR at canopy level are caused by either sun 418

elevation or in the case of a constant solar angle by canopy structure (ie shading fractional 419

vegetation cover) In complex structured canopies with many foliage layers the saturation of 420

GPP with APAR may not be observable because variations of APAR are less related to PAR 421

falling on the leaves but rather to the variation of leaf area exposed to light Model 422

simulations suggest that in complex canopies SIF760 saturates with APAR before GPP due to 423

fesc being lower with high APAR (data not shown) This structural effect reduces the curvature 424

of the hyperbolic leaf level GPPL- SIF760 L relationship at canopy level (compare Figures 6F 425

and 7F) Furthermore the effect of structure and APAR covary in time since canopy structure 426

(ie LAI) varies in concert with APAR during a seasonal cycle especially for crops We 427

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

18

expect that this covariance makes the relationship between GPP and SIF760 more linear at the 428

canopy level especially if observations are aggregated over time 429

To evaluate the structure related alleviation of the hyperbolic relationship between 430

SIF760 and GPP at canopy level in response to temporal aggregation we modelled an annual 431

course of GPPC and SIF760 C considering varying environmental conditions (Figure 8A-B) 432

ie PAR ranging from 0-500 mW m-2 sr-1 nm-1 air temperature ranging from 0-32degC using 433

data collected during 2009 at a Nitro-Europe flux tower site (Speulderbos The Netherlands) 434

and canopy properties ie LAI ranging from 15-30 m2 m-2 (Figure 8C) The LAI data are 435

synthetic but have a realistic seasonal cycle For daily noon values of modelled SIF760 C and 436

GPPC an asymptotic behaviour can be observed while GPPC tends to saturate compared to 437

SIF760 C The SIF760C-GPPC relationship tends to become linear when monthly or quarterly 438

averages are considered (Figure 8D) 439

440

lt Figure 8 gt 441

442

This visual trend can be confirmed by statistical measures for daily aggregated values 443

the hyperbolic model (rRMSE = 174 R2 = 070) performed slightly better than the linear 444

model (rRMSE = 185 r2 = 068) In case of monthly aggregated data both the hyperbolic 445

and linear models performed similarly while the linear model gave better results (RMSE = 446

90 r2 = 090) than the hyperbolic model (rRMSE = 94 R2 = 088) for quarterly aggregated 447

data 448

449

4 Discussion 450

Significant relationships between remotely measured SIF and ecosystem GPP have 451

been reported in the past (Frankenberg et al 2011 Guanter et al 2014) however sufficient 452

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

19

process understanding for using SIF as proxy for GPP across ecosystems is still limited 453

(Porcar-Castell et al 2014) Below we discuss how our observational and modeling 454

approaches help bridging the gap between mechanistic understanding at the leaf level and 455

ecosystem-specific observations and thus provide insights on mechanisms determining the 456

complex relationship between SIF and GPP 457

458

41 Insights from the observational approach 459

Our results showed ecosystem-specific asymptotic relationships between SIF760 O and 460

GPPEC for the three investigated ecosystems mixed temperate forest cropland and perennial 461

grassland The varying SIF-GPP relationships across ecosystems were consistent with results 462

recently reported from combined satellite observations and model predictions (Guanter et al 463

2012) Our results showed that the relationships of both tested vegetation indices PRI and 464

MCARI2 to GPPEC experienced a higher variability across ecosystems compared to SIF760 O 465

SIF760 O showed an ecosystem-specific and overall moderate to good relationship (R2 between 466

048 and 088) to GPPEC for the three ecosystems Both vegetation indices showed only strong 467

relationships for the grassland (R2 = 095 and 098) that notably changed in LAI during the 468

various observations It is important to note that we do not further elaborate on the underlying 469

mechanisms relating vegetation indices and GPPEC and that it is unclear if hyperbolic models 470

best describe their relationships to GPPEC based on mechanistic principles The intention of 471

evaluating the relationships between GPPEC and both vegetation indices was to provide a base 472

to judge the performance of SIF760 O in relation to these commonly used remote observations 473

The diversity of sampled conditions and used data sources provide further evidence on 474

the robustness of observed SIF760 O-GPPEC relationships Our measurements represent various 475

canopies characterized by different biomasses sampled over several years Although most 476

observations were acquired under similar light condition and physiological states a few 477

measurements represented lower light conditions and other physiological states This suggests 478

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

20

that observed relationships are rather robust across different representations of APAR LUEp 479

and LUEf Furthermore we use data from different sources ie four EC flux towers the 480

APEX imaging spectrometer and modeling at leaf and canopy levels to reduce the 481

uncertainties associated with each method of measurements and to get a better understanding 482

of the SIF760 O-GPPEC relationship at ecosystem level 483

Flux tower measurements are representative of a certain source area (footprint) from 484

the ecosystem which changes constantly based on wind direction and turbulent state of the 485

atmosphere Depending on the heterogeneity of the ecosystem GPPEC signals can vary in 486

response to changing footprints (Barcza et al 2009 Kljun et al 2004) although this variation 487

should be rather small if the vegetation is homogeneous Remote measurements of SIF760 are 488

likely representative of the upper layer of the ecosystem thus vertical canopy heterogeneity 489

potentially leads to a mismatch when compared to GPPEC which integrates over the different 490

layers of the ecosystems This is expected to be especially the case in structurally complex 491

ecosystems such as forests Although the upper layer in forest ecosystems substantially 492

contributes to the total rate of GPPEC there are still many layers of leaves or needles within 493

the canopy photosynthesizing at different rates depending on their sunshade exposure In 494

addition understory vegetation in forest ecosystems can significantly contribute to GPPEC at 495

different rates depending on the ecosystem and time of year Even homogeneous vegetation 496

canopies like crops or grasslands can show vertical heterogeneity This is especially the case 497

at peak biomass during ripening or senescence During such phenological stages the top layer 498

of cereal crops consists of less fluorescing components (ie ears) and cereal crops or 499

grasslands can contain a higher fraction of brown pigments resulting in low SIF760 O values 500

while the entire canopy can still produce high photosynthetic rates and thus GPPEC (Damm et 501

al 2010) Indeed most of the crops investigated in our study were in the ripening phase with a 502

pronounced ear layer 503

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

21

This resulted in relatively low SIF760 O values and in high GPPEC and thus in a more 504

pronounced asymptotic relationship (cf Figure 4A Figure 7F) compared to results reported 505

for other crops (Guanter et al 2013 Guanter et al 2014) Further research is needed to better 506

understand the impact of vertical heterogeneity on GPPEC and SIF760 observations and their 507

respective relationships 508

Despite the technical aspects causing variations in retrieved SIF760 O (refer to 509

Appendix-D for a detailed discussion) canopy structure causes inherent variations in 510

measured SIF760 values and thus impacts SIF760-GPP relationships The combination of 511

complex canopy architecture and the distribution of absorbing and scattering elements 512

determine vegetation canopies to act as photon traps (Knyazikhin et al 2013 Lewis and 513

Disney 2007) Parts of emitted SIF760 photons from inner-canopy leaves are re-absorbed 514

while others escape the canopy in direction of the sensor The probability of photons escaping 515

the canopy is a function of canopy structure which implies sensitivity to measured SIF760 516

intensities for canopy structure In Eq (2) this sensitivity was approximated with the term 517

fesc In our analysis we minimized the impact of fesc by observing always the same forest area 518

representative for the EC tower footprint In less complex structured crop and grass canopies 519

we expect fesc to be less important However future research is required to quantify the impact 520

of fesc and to propose strategies to compensate its influence on SIF760 retrievals 521

522

42 Insights from the modeling approach 523

From a theoretical point of view radiative energy loss competes with photochemistry 524

and non-radiative energy dissipation for absorbed photons Considering that GPP and SIF760 525

can be conceptualized according to Eqs (1) and (2) SIF760 must show under certain 526

environmental conditions a sensitivity to photochemistry (Porcar-Castell et al 2014 van der 527

Tol et al 2009a van der Tol et al 2009b) This relationship is confirmed by our observational 528

data and other studies at leaf (Amoros-Lopez et al 2008) and canopy levels (Damm et al 529

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

22

2010) Our modeling results however indicated that the relationship between LUEf and LUEp 530

is subject to change hence impacting the sensitivity of SIF760 to track changes in GPP 531

(Figures 6C and 7C) 532

The asymptotic relationship between SIF760 and GPP theoretically predicted and 533

observed at leaf and canopy levels can be related in particular to the different value ranges of 534

both efficiency terms (LUEp and LUEf) (Figures 6C and 7C) while a complex set of 535

mechanisms causes an ecosystem dependency for SIF760-GPP relationships The 536

photosynthetic capacity of the entire ecosystem in combination with its vertical heterogeneity 537

determines differences in GPP and SIF760 and thus the slope and curvature of their respective 538

relationships Furthermore the initial slope of the asymptotic SIF760 - GPP relationship which 539

has a value of LUEpmax(LUEfmiddotfesc) is determined by a combination of the photosynthetic 540

efficiency and canopy structure Canopy structure and related shadowing effects alter the 541

average light interception and quality at leaf level (Nobel et al 1993 Stewart et al 2003) and 542

consequently the overall productivity of the canopy In the case of reduced APAR caused by 543

canopy shadowing the overall apparent stress level in the canopy and consequently the rate 544

of certain photoprotective mechanisms such as non-radiative energy dissipation is reduced 545

(Cheng et al 2015 Demmig-Adams 1998 Niinemets et al 2003) and observed GPP and 546

SIF760 fall in a range where they are more linearly related (cf Figure 6F and 7F) In addition 547

canopy level simulations show that the fraction of senescent plant material impacts the 548

curvature of the hyperbolic relationship 549

Recent literature reports linear relationships between satellite-based SIF and modeled 550

GPP (Frankenberg et al 2011 Guanter et al 2012 Guanter et al 2014) while we found 551

asymptotic relationships Both results are not contradicting if one considers the temporal 552

aggregation applied to the data Our results are based on instantaneous measurements while 553

monthly averages were calculated for the satellite data to minimize noise effects (Frankenberg 554

et al 2011 Guanter et al 2014) It is known from previous research that temporal averaging 555

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

23

tends to linearize the response between GPP and APAR (Ruimy et al 1995) as saturating 556

light conditions tend to be only a small fraction of averaged diurnalmonthly data 557

Aggregating GPPEC and APAR over different time scales also showed this effect (data not 558

shown) Because of the rather linear relationship between SIF760 and APAR we expect that 559

temporal aggregation tends to linearize the SIF760-GPP relationship as well A recent study 560

presenting continuous measurements of GPP and SIF760 over a temperate deciduous forest 561

support this hypothesis (Yang et al 2015) The increasing number of experiments using 562

autonomous spectrometer systems in combination with flux towers (Balzarolo et al 2011) 563

will likely provide more experimental results that supports this finding Our simulations 564

provide further evidence that the covariance of structure and APAR during a seasonal cycle 565

additionally tends to linearize the relationship between GPP and SIF760 if observations are 566

aggregated over time 567

The dependency of SIF760-GPP relationships on temporal aggregation requires special 568

attention since it has implications on the usage of SIF to constrain GPP In principle 569

conclusions and results obtained from temporally aggregated and instantaneous measurements 570

are valid if consistently related to the underlying temporal aggregation level Generalizing 571

conclusions obtained at various aggregation levels bears the risk for mis-interpretations eg 572

when vegetation models are refined or empirically re-calibrated using observations or for 573

projections of GPP using observational SIF data 574

575

43 Requirements for the use of SIF to assess GPP across ecosystems 576

The changing relationship between SIF760 and GPPEC with vegetation type challenges 577

the usage of SIF760 as a robust constraint to estimate GPP Specific strategies are required to 578

account for ecosystem (or species) specific sensitivities of SIF in general and SIF760 in 579

particular This includes i) integration platforms to combine observations and process 580

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

24

understanding and ii) accurate observations of additional vegetation and environmental 581

variables 582

Process-based models as data integration platform in combination with comprehensive 583

observations to sufficiently measure important environmental factors and vegetation 584

properties are considered as flexible and powerful framework Parazoo et al (2014) for 585

example proposed the combination of ensemble dynamic global vegetation models (DGVMs) 586

in combination with SIF as observational constraint Lee et al (2015) suggest incorporating 587

SIF in a climate model to utilize SIF as observational constraint of photosynthesis simulations 588

and subsequent calculations of carbon water and energy cycle information Yet other studies 589

propose using SIF to obtain physiological vegetation parameters ie Vcmax for improved 590

parameterizations of photosynthesis models to eventually model GPP (Zhang et al 2014) The 591

European space agencys (ESA) Earth Explorer 8 opportunity mission Fluorescence Explorer 592

(FLEX) suggests a combination of comprehensive observations (ie SIF over the full 593

emission spectrum vegetation ecosystems and actual environmental conditions) and process-594

based models (Guanter et al 2012) 595

Additional information is required to use SIF in process-based models as one has to 596

account for varying photosynthetic efficiencies of plants and prevailing stress responses 597

According to the conceptualization of GPP and SIF as given in Eqs (1) and (2) extracting 598

and evaluating the exact relationship between LUEp and LUEf is essential but requires 599

normalizing APAR and calculating both yields (Damm et al 2010 Govindjee 2004 Louis et 600

al 2005) Retrievals of APAR can be considered challenging (Garbulsky et al 2010 Gitelson 601

and Gamon 2015 Hanan et al 2002) as available approaches often utilize a-priori knowledge 602

of prevailing land cover or plant distribution functions and rely on the inversion of physical-603

based models or simple band ratios (Hilker et al 2008) Research is required to provide high 604

quality information on APAR not perturbed by light absorption of non-photosynthetic active 605

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

25

canopy elements (stems branches twigs) cf Zhang et al (2013 2012) that demonstrate 606

attempts to improve estimates of APAR 607

Observed SIF760 O - GPPEC relationships are additionally confounded by non-radiative 608

energy dissipation The PRI was designed to monitor pigment changes related to the 609

xanthophyll cycle (Gamon et al 1992) and provides access to this superimposing process 610

However reported sensitivities of PRI for other effects ie structure (Barton and North 611

2001) pigment pool sizes (Gamon and Berry 2012) illumination effects (Damm et al 2015) 612

and the results obtained in this study question the applicability of the PRI at canopy level 613

Alternative approaches such as the development of a canopy PRI as suggested by several 614

groups (Hernandez-Clemente et al 2011 Kovac et al 2013 Wu et al 2010) or the usage of 615

pigment compositions and other functional plant traits might represent interesting strategies 616

and should be elaborated in future research 617

Structural sensitivities of canopy leaving SIF signals are critical and require further 618

research The application of spectrally invariant correction factors eg the directional area 619

scattering factor (Knyazikhin et al 2013) or more advanced physically-based retrieval 620

schemes using combined atmosphere-canopy models (Damm et al 2015 Laurent et al 2013) 621

might provide potential strategies to compensate for canopy structural effects on retrieved 622

SIF760 623

Another important source of information that allows improving the usage of SIF as 624

constraint of GPP is to consider SIF emissions over the entire emission spectrum between 625

approximately 600 and 800 nm Recent research successfully demonstrated the 626

complementary information content of red-fluorescence retrieved in the O2-B band at 688 nm 627

(SIF688) compared to SIF760 because of the different contributions from photosystem I and II 628

(Rossini et al 2015) Yet other studies highlighted the potential to improve predictions of 629

LUEp and GPP using both SIF688 and SIF760 (Cheng et al 2013) 630

631

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

26

5 Conclusions 632

Remote sensing of SIF provides a new observational approach to assess terrestrial 633

GPP Compared to traditional greenness-based remote sensing parameters we found SIF760 to 634

be more consistently related to GPPEC These results in combination with the more direct 635

mechanistic link between SIF760 and photosynthesis suggest that using SIF760 provides a 636

strategy to decrease uncertainty in estimating GPP across ecosystems 637

Our observational and modeling approaches revealed asymptotic and ecosystem-638

specific relationships for SIF760 and GPPEC We identified a set of interlinked mechanisms 639

determining SIF760-GPPEC relationships to be ecosystem specific ie the photosynthetic 640

capacity and efficiency of ecosystems the confounding impact of non-radiative energy 641

dissipation (NPQ) as well as ecosystem structure and the fraction of non-photosynthetic 642

vegetation components We demonstrated that SIF760-GPPEC relationships scale with temporal 643

aggregation due to a covariance of canopy structure and APAR over the seasonal cycle 644

We suggest going beyond empirical and ecosystem-specific relationships instead 645

using instantaneous observations comprehensively providing information on the full SIF 646

emission vegetation status and functioning as well as environmental conditions in 647

combination with process-based models Such approaches eg suggested for ESArsquos potential 648

ldquoFluorescence Explorerrdquo (FLEX) Earth Explorer mission provide a flexible framework to 649

account for the various influencing factors identified and to successfully apply SIF as 650

constraint of unbiased global GPP estimates 651

Furthermore research is required to improve the comparison between observations of 652

GPPEC and SIF Both observations stem from conceptually contrasting technologies ie 653

measurements of gas exchange vs plant-light interactions and provide independent and 654

complementary access to study ecosystem photosynthesis Despite the need to harmonize 655

horizontally differing footprints the vertical heterogeneity of ecosystems including the 656

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

27

varying contributions from different ecosystem layers must be better understood to harmonize 657

and successfully integrate both measurements 658

659

Acknowledgements 660

This work was supported by a grant of the Swiss University Conference and ETH-661

Board in frame of the Swiss Earth Observatory Network (SEON) project and by a grant of the 662

European Space Agency (ESA) in frame of the FLEX-PARCS E Paul-Limoges was 663

supported in part by the Natural Sciences and Engineering Research Council of Canada 664

through a postgraduate scholarship We would like to thank the APEX team for acquiring the 665

APEX data Authors sequence is listed following the FLAE approach (doi 666

101371journalpbio0050018) 667

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

28

References 668

Ammann C Flechard CR Leifeld J Neftel A amp Fuhrer J (2007) The carbon budget of 669

newly established temperate grassland depends on management intensity Agriculture 670

Ecosystems amp Environment 121 5-20 671

Amoros-Lopez J Gomez-Chova L Vila-Frances J Alonso L Calpe J Moreno J amp 672

del Valle-Tascon S (2008) Evaluation of remote sensing of vegetation fluorescence by 673

the analysis of diurnal cycles International Journal of Remote Sensing 29 5423-5436 674

Baker NR (2008) Chlorophyll Fluorescence A probe of photosynthesis in vivo Annual 675

Review of Plant Physiology and Plant Molecular Biology 59 89-113 676

Baldocchi D Falge E Gu LH Olson R Hollinger D Running S Anthoni P 677

Bernhofer C Davis K Evans R Fuentes J Goldstein A Katul G Law B Lee 678

XH Malhi Y Meyers T Munger W Oechel W U KTP Pilegaard K Schmid 679

HP Valentini R Verma S Vesala T Wilson K amp Wofsy S (2001) FLUXNET A 680

new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide 681

water vapor and energy flux densities Bulletin of the American Meteorological Society 682

82 2415-2434 683

Baldocchi DD (2003) Assessing the eddy covariance technique for evaluating carbon 684

dioxide exchange rates of ecosystems past present and future Global Change Biology 9 685

479-492 686

Balzarolo M Anderson K Nichol C Rossini M Vescovo L Arriga N Wohlfahrt G 687

Calvet J-C Carrara A Cerasoli S Cogliati S Daumard F Eklundh L Elbers JA 688

Evrendilek F Handcock RN Kaduk J Klumpp K Longdoz B Matteucci G 689

Meroni M Montagnani L Ourcival J-M Sanchez-Canete EP Pontailler J-Y 690

Juszczak R Scholes B amp Pilar Martin M (2011) Ground-Based Optical Measurements 691

at European Flux Sites A Review of Methods Instruments and Current Controversies 692

Sensors 11 7954-7981 693

Barcza Z Kern A Haszpra L amp Kljun N (2009) Spatial representativeness of tall tower 694

eddy covariance measurements using remote sensing and footprint analysis Agricultural 695

and Forest Meteorology 149 795-807 696

Barton CVM amp North PRJ (2001) Remote sensing of canopy light use efficiency using 697

the photochemical reflectance index - Model and sensitivity analysis Remote Sensing of 698

Environment 78 264-273 699

Beer C Reichstein M Tomelleri E Ciais P Jung M Carvalhais N Roedenbeck C 700

Arain MA Baldocchi D Bonan GB Bondeau A Cescatti A Lasslop G Lindroth 701

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

29

A Lomas M Luyssaert S Margolis H Oleson KW Roupsard O Veenendaal E 702

Viovy N Williams C Woodward FI amp Papale D (2010) Terrestrial Gross Carbon 703

Dioxide Uptake Global Distribution and Covariation with Climate Science 329 834-838 704

Berk A Anderson GP Acharya PK Bernstein LS Muratov L Lee J Fox M 705

Adler-Golden SM Chetwynd JH Hoke ML Lockwood RB Gardner JA Cooley 706

TW Borel CC amp Lewis PE (2005) MODTRAN5 A reformulated atmospheric band 707

model with auxiliary species and practical multiple scattering options Proceedings of the 708

Society of Photo-Optical Instrumentation Engineer 5655 662-667 709

Cheng SJ Bohrer G Steiner AL Hollinger DY Suyker A Phillips RP amp 710

Nadelhoffer KJ (2015) Variations in the influence of diffuse light on gross primary 711

productivity in temperate ecosystems Agricultural and Forest Meteorology 201 98-110 712

Cheng Y-B Middleton EM Zhang Q Huemmrich KF Campbell PKE Corp LA 713

Cook BD Kustas WP amp Daughtry CS (2013) Integrating Solar Induced 714

Fluorescence and the Photochemical Reflectance Index for Estimating Gross Primary 715

Production in a Cornfield Remote Sensing 5 6857-6879 716

Collatz GJ Ball JT Grivet C amp Berry JA (1991) Physiological and environmental-717

regulation of stomatal conductance photosynthesis and transpiration - a model that 718

includes a laminar boundary-layer Agricultural and Forest Meteorology 54 107-136 719

DOdorico P Alberti E amp Schaepman ME (2010) In-flight spectral performance 720

monitoring of the Airborne Prism Experiment Applied Optics 49 3082-3091 721

DOdorico P Guanter L Schaepman ME amp Schlaumlpfer D (2011) Performance 722

assessment of onboard and scene-based methods for Airborne Prism Experiment spectral 723

characterization Applied Optics 50 4755-4764 724

Damm A Erler A Gioli B Hamdi K Hutjes R Kosvancova M Meroni M 725

Miglietta F Moersch A Moreno J Schickling A Sonnenschein R Udelhoven T 726

Van der Linden S Hostert P amp Rascher U (2010) Remote sensing of sun induced 727

fluorescence yield to improve modelling of diurnal courses of Gross Primary Production 728

(GPP) Global Change Biology 16 171-186 729

Damm A Erler A Hillen W Meroni M Schaepman ME Verhoef W amp Rascher U 730

(2011) Modeling the impact of spectral sensor configurations on the FLD retrieval 731

accuracy of sun-induced chlorophyll fluorescence Remote Sensing of Environment 115 732

1882-1892 733

Damm A Guanter L Laurent VCE Schaepman ME Schickling A amp Rascher U 734

(2014) FLD-based retrieval of sun-induced chlorophyll fluorescence from medium 735

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

30

spectral resolution airborne spectroscopy data Remote Sensing of Environment 147 256-736

266 737

Damm A Guanter L Verhoef W Schlaumlpfer D Garbari S amp Schaepman ME (2015) 738

Impact of varying irradiance on vegetation indices and chlorophyll fluorescence derived 739

from spectroscopy data Remote Sensing of Environment 156 202-215 740

Demmig-Adams B (1998) Survey of thermal energy dissipation and pigment composition in 741

sun and shade leaves Plant and Cell Physiology 39 474-482 742

Drolet GG Middleton EM Huemmrich KF Hall FG Amiro BD Barr AG Black 743

TA McCaughey JH amp Margolis HA (2008) Regional mapping of gross light-use 744

efficiency using MODIS spectral indices Remote Sensing of Environment 112 3064-3078 745

Eugster W amp Senn W (1995) A cospectral correction model for measurement of turbulent 746

NO2 fluxes Boundary-Layer Meteorology 74 321-340 747

Eugster W Zeyer K Zeeman M Michna P Zingg A Buchmann N amp Emmenegger 748

L (2007) Methodical study of nitrous oxide eddy covariance measurements using 749

quantum cascade laser spectrometery over a Swiss forest Biogeosciences 4 927-939 750

Falge E Baldocchi D Olson R Anthoni P Aubinet M Bernhofer C Burba G 751

Ceulemans R Clement R Dolman H Granier A Gross P Grunwald T Hollinger 752

D Jensen NO Katul G Keronen P Kowalski A Lai CT Law BE Meyers T 753

Moncrieff H Moors E Munger JW Pilegaard K Rannik U Rebmann C Suyker 754

A Tenhunen J Tu K Verma S Vesala T Wilson K amp Wofsy S (2001) Gap 755

filling strategies for defensible annual sums of net ecosystem exchange Agricultural and 756

Forest Meteorology 107 43-69 757

Falloon P amp Betts R (2010) Climate impacts on European agriculture and water 758

management in the context of adaptation and mitigation-The importance of an integrated 759

approach Science of the Total Environment 408 5667-5687 760

Foken T amp Wichura B (1996) Tools for quality assessment of surface-based flux 761

measurements Agricultural and Forest Meteorology 78 83-105 762

Fournier A Daumard F Champagne S Ounis A Goulas Y amp Moya I (2012) Effect 763

of canopy structure on sun-induced chlorophyll fluorescence Isprs Journal of 764

Photogrammetry and Remote Sensing 68 112-120 765

Frankenberg C Fisher JB Worden J Badgley G Saatchi SS Lee JE Toon GC 766

Butz A Jung M Kuze A amp Yokota T (2011) New global observations of the 767

terrestrial carbon cycle from GOSAT Patterns of plant fluorescence with gross primary 768

productivity Geophysical Research Letters 38 769

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

31

Gamon JA amp Berry JA (2012) Facultative and constitutive pigment effects on the 770

Photochemical Reflectance Index (PRI) in sun and shade conifer needles Israel Journal of 771

Plant Sciences 60 85-95 772

Gamon JA Coburn C Flanagan LB Huemmrich KF Kiddle C Sanchez-Azofeifa 773

GA Thayer DR Vescovo L Gianelle D Sims DA Rahman AF amp Pastorello 774

GZ (2010) SpecNet revisited bridging flux and remote sensing communities Canadian 775

Journal of Remote Sensing 36 S376-S390 776

Gamon JA Penuelas J amp Field CB (1992) A narrow-waveband spectral index that 777

tracks diurnal changes in photosynthetic efficiency Remote Sensing of Environment 41 778

35-44 779

Garbulsky MF Filella I Verger A amp Penuelas J (2014) Photosynthetic light use 780

efficiency from satellite sensors From global to Mediterranean vegetation Environmental 781

and Experimental Botany 103 3-11 782

Garbulsky MF Penuelas J Papale D Ardo J Goulden ML Kiely G Richardson 783

AD Rotenberg E Veenendaal EM amp Filella I (2010) Patterns and controls of the 784

variability of radiation use efficiency and primary productivity across terrestrial 785

ecosystems Global Ecology and Biogeography 19 253-267 786

Gege P Fries J Haschberger P Schoumltz P Schwarzer H Strobl P Suhr B Ulbrich 787

G amp Vreeling WJ (2009) Calibration facility for airborne imaging spectrometers 788

ISPRS Journal of Photogrammetry amp Remote Sensing 64 387ndash397 789

Gitelson AA amp Gamon JA (2015) The need for a common basis for defining light-use 790

efficiency Implications for productivity estimation Remote Sensing of Environment 156 791

196-201 792

Goetz SJ amp Prince SD (1999) Modelling terrestrial carbon exchange and storage 793

Evidence and implications of functional convergence in light-use efficiency Advances in 794

Ecological Research Vol 28 (pp 57-92) 795

Govindjee (2004) Chlorophyll a fluorescence A bit of basics and history In GC 796

Papageorgiou amp Govindjee (Eds) Chlorophyll a Fluorescence A Signature of 797

Photosynthesis Advances in Photosynthesis and Respiration (pp 1-42) Dordrecht 798

Springer 799

Guanter L Alonso L Gomez-Chova L Amoros-Lopez J Vila J amp Moreno J (2007) 800

Estimation of solar-induced vegetation fluorescence from space measurements 801

Geophysical Research Letters 34 doi 1010292007GL029289 802

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

32

Guanter L Alonso L Gomez-Chova L Meroni M Preusker R Fischer J amp Moreno 803

J (2010) Developments for vegetation fluorescence retrieval from spaceborne high-804

resolution spectrometry in the O2-A and O2-B absorption bands Journal of Geophysical 805

Research-Atmospheres 115 806

Guanter L Frankenberg C Dudhia A Lewis PE Goacutemez-Dans J Kuze A Suto H amp 807

Grainger R (2012) Retrieval and global assessment of terrestrial chlorophyll fluorescence 808

from GOSAT space measurements Remote Sensing of Environment 121 236-251 809

Guanter L Rossini M Colombo R Meroni M Frankenberg C Lee J-E amp Joiner J 810

(2013) Using field spectroscopy to assess the potential of statistical approaches for the 811

retrieval of sun-induced chlorophyll fluorescence from ground and space Remote Sensing 812

of Environment 133 52-61 813

Guanter L Zhang Y Jung M Joiner J Voigt M Berry JA Frankenberg C Huete 814

AR Zarco-Tejada P Lee J-E Moran MS Ponce-Campos G Beer C Camps-815

Valls G Buchmann N Gianelle D Klumpp K Cescatti A Baker JM amp Griffis 816

TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll 817

fluorescence Proceedings of the National Academy of Sciences of the United States of 818

America 111 E1327-E1333 819

Haboudane D Miller JR Pattey E Zarco-Tejada PJ amp Strachan IB (2004) 820

Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop 821

canopies Modeling and validation in the context of precision agriculture Remote Sensing 822

of Environment 90 337-352 823

Hanan NP Burba G Verma SB Berry JA Suyker A amp Walter-Shea EA (2002) 824

Inversion of net ecosystem CO2 flux measurements for estimation of canopy PAR 825

absorption Global Change Biology 8 563-574 826

Hernandez-Clemente R Navarro-Cerrillo RM Suarez L Morales F amp Zarco-Tejada 827

PJ (2011) Assessing structural effects on PRI for stress detection in conifer forests 828

Remote Sensing of Environment 115 2360-2375 829

Hilker T Coops NC Wulder MA Black TA amp Guy RD (2008) The use of remote 830

sensing in light use efficiency based models of gross primary production A review of 831

current status and future requirements Science of the Total Environment 404 411-423 832

Hilker T Gitelson A Coops NC Hall FG amp Black TA (2011) Tracking plant 833

physiological properties from multi-angular tower-based remote sensing Oecologia 165 834

865-876 835

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

33

Hilker T Lyapustin A Hall FG Wang Y Coops NC Drolet G amp Black TA 836

(2009) An assessment of photosynthetic light use efficiency from space Modeling the 837

atmospheric and directional impacts on PRI reflectance Remote Sensing of Environment 838

113 2463-2475 839

Hueni A Biesemans J Meuleman K DellEndice F Schlaumlpfer D Adriaensen S 840

Kempenaers S Odermatt D Kneubuehler M Nieke J amp Itten K (2009) Structure 841

Components and Interfaces of the Airborne Prism Experiment (APEX) Processing and 842

Archiving Facility IEEE Transactions on Geoscience and Remote Sensing 47 29-43 843

Hueni A Lenhard K Baumgartner A amp Schaepman M (2013) The APEX (Airborne 844

Prism Experiment - Imaging Spectrometer) Calibration Information System IEEE 845

Transactions on Geoscience and Remote Sensing 51 5169-5180 846

Hueni A Schlaepfer D Jehle M amp Schaepman M (2015) Impacts of Dichroic Prism 847

Coatings on Radiometry of the Airborne Imaging Spectrometer APEX Applied Optics 53 848

5344-5352 849

Imhoff ML Bounoua L Ricketts T Loucks C Harriss R amp Lawrence WT (2004) 850

Global patterns in human consumption of net primary production Nature 429 870-873 851

IPCC (2013) Summary for policymakers Climate change 2013 the physical science basis 852

Cambridge University Press Cambridge United Kingdom and New York NY USA 853

Jacquemoud S amp Baret F (1990) PROSPECT A model of leaf optical properties spectra 854

Remote Sensing of Environment 34 75-91 855

Jehle M Hueni A Damm A DOdorico P Kneubuumlhler M Meuleman K Schlapfer 856

D Schaepman ME amp Weyermann J (2010) APEX - current status performance and 857

validation concept IEEE Sensors 2010 Conference 533-537 858

Joiner J Guanter L Lindstrot R Voigt M Vasilkov AP Middleton EM Huemmrich 859

KF Yoshida Y amp Frankenberg C (2013) Global monitoring of terrestrial chlorophyll 860

fluorescence from moderate spectral resolution near-infrared satellite measurements 861

methodology simulations and application to GOME-2 Atmos Meas Tech Discuss 6 862

3883-3930 863

Jung M Reichstein M Margolis HA Cescatti A Richardson AD Arain MA 864

Arneth A Bernhofer C Bonal D Chen J Gianelle D Gobron N Kiely G Kutsch 865

W Lasslop G Law BE Lindroth A Merbold L Montagnani L Moors EJ 866

Papale D Sottocornola M Vaccari F amp Williams C (2011) Global patterns of land-867

atmosphere fluxes of carbon dioxide latent heat and sensible heat derived from eddy 868

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

34

covariance satellite and meteorological observations Journal of Geophysical Research-869

Biogeosciences 116 870

Kljun N Kastner-Klein P Fedorovich E amp Rotach MW (2004) Evaluation of 871

Lagrangian footprint model using data from wind tunnel convective boundary layer 872

Agricultural and Forest Meteorology 127 189-201 873

Kljun N Rotach MW amp Schmid HP (2002) A three-dimensional backward lagrangian 874

footprint model for a wide range of boundary-layer stratifications Boundary-Layer 875

Meteorology 103 205-226 876

Knyazikhin Y Schull MA Stenberg P Mottus M Rautiainen M Yang Y Marshak 877

A Latorre Carmona P Kaufmann RK Lewis P Disney MI Vanderbilt V Davis 878

AB Baret F Jacquemoud S Lyapustin A amp Myneni RB (2013) Hyperspectral 879

remote sensing of foliar nitrogen content Proceedings of the National Academy of 880

Sciences of the United States of America 110 E185-E192 881

Kovac D Malenovsky Z Urban O Spunda V Kalina J Ac A Kaplan V amp Hanus 882

J (2013) Response of green reflectance continuum removal index to the xanthophyll de-883

epoxidation cycle in Norway spruce needles Journal of Experimental Botany 64 1817-884

1827 885

Krausmann F Erb K-H Gingrich S Haberl H Bondeau A Gaube V Lauk C 886

Plutzar C amp Searchinger TD (2013) Global human appropriation of net primary 887

production doubled in the 20th century Proceedings of the National Academy of Sciences 888

of the United States of America 110 10324-10329 889

Lasslop G Reichstein M Papale D Richardson AD Arneth A Barr A Stoy P amp 890

Wohlfahrt G (2010) Separation of net ecosystem exchange into assimilation and 891

respiration using a light response curve approach critical issues and global evaluation 892

Global Change Biology 16 187-208 893

Laurent VCE Verhoef W Damm A Schaepman ME amp Clevers JGPW (2013) A 894

Bayesian object-based approach for estimating vegetation biophysical variables from at-895

sensor APEX radiance data Remote Sensing of Environment 139 6-17 896

Lee JE Berry JA Van der Tol C Yang X Guanter L Damm A Baker I amp 897

Frankenberg C (2015) Simulations of chlorophyll fluorescence in corporated into the 898

Community Land Model version 4 Global Change Biology in print 899

Lewis P amp Disney M (2007) Spectral invariants and scattering across multiple scales from 900

within-leaf to canopy Remote Sensing of Environment 109 196-206 901

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

35

Liu LY amp Cheng ZH (2010) Detection of Vegetation Light-Use Efficiency Based on 902

Solar-Induced Chlorophyll Fluorescence Separated From Canopy Radiance Spectrum Ieee 903

Journal of Selected Topics in Applied Earth Observations and Remote Sensing 3 306-312 904

Louis J Ounis A Ducruet JM Evain S Laurila T Thum T Aurela M Wingsle G 905

Alonso L Pedros R amp Moya I (2005) Remote sensing of sunlight-induced chlorophyll 906

fluorescence and reflectance of Scots pine in the boreal forest during spring recovery 907

Remote Sensing of Environment 96 37-48 908

Maier SW Guumlnther KP amp Stellmes M (2003) Sun-Induced Fluorescence A New Tool 909

for Precision Farming In M McDonald J Schepers L Tartly T van Toai amp D Major 910

(Eds) Digital Imaging and Spectral Techniques Applications to Precision Agriculture 911

and Crop Physiology (pp 209-222) ASA Special Publication 912

Mauder M Foken T Clement R Elbers JA Eugster W Gruenwald T Heusinkveld 913

B amp Kolle O (2008) Quality control of CarboEurope flux data - Part 2 Inter-914

comparison of eddy-covariance software Biogeosciences 5 451-462 915

Meroni M Busetto L Guanter L Cogliati S Crosta GF Migliavacca M Panigada 916

C Rossini M amp Colombo R (2010) Characterization of fine resolution field 917

spectrometers using solar Fraunhofer lines and atmospheric absorption features Applied 918

Optics 49 2858-2871 919

Meroni M Picchi V Rossini M Cogliati S Panigada C Nali C Lorenzini G amp 920

Colombo R (2008) Leaf level early assessment of ozone injuries by passive fluorescence 921

and photochemical reflectance index International Journal of Remote Sensing 29 5409-922

5422 923

Meroni M Rossini M Guanter L Alonso L Rascher U Colombo R amp Moreno J 924

(2009) Remote sensing of solar induced chlorophyll fluorescence review of methods and 925

applications Remote Sensing of Environment doi101016jrse200905003 926

Michaelis L amp Menten ML (1913) The kenetics of the inversion effect Biochemische 927

Zeitschrift 49 333-369 928

Middleton E McMurtney JE Campbell PK Corp LA Butcher LM amp Chappelle 929

EW (2002) Optical and fluorescence properties of corn leaves from different nitrogen 930

regimes In M Owe G Durso amp L Toulios (Eds) Conference on Remote Sensing for 931

Agriculture Ecosystems and Hydrology IV (pp 72-83) Agia Pelagia Greece 932

Miller JR Berger M Goulas Y Jacquemoud S Louis J Moise N Mohammed G 933

Moreno J Moya I Pedroacutes R Verhoef W amp Zarco-Tejada PJ (2005) Development 934

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

36

of a vegetation fluorescence canopy model Final Report May 2005 In European Space 935

Research and Technology Centre (ESTEC) 936

Monteith JL (1972) Solar-radiation and productivity in tropical ecosystems Journal of 937

Applied Ecology 9 747-766 938

Niinemets U Kollist H Garcia-Plazaola JI Hernandez A amp Becerril JM (2003) Do 939

the capacity and kinetics for modification of xanthophyll cycle pool size depend on growth 940

irradiance in temperate trees Plant Cell and Environment 26 1787-1801 941

Nobel PS Forseth IN amp Long SP (1993) Canopy structure and light interception 942

Ozanne CMP Anhuf D Boulter SL Keller M Kitching RL Korner C Meinzer 943

FC Mitchell AW Nakashizuka T Dias PLS Stork NE Wright SJ amp 944

Yoshimura M (2003) Biodiversity meets the atmosphere A global view of forest 945

canopies Science 301 183-186 946

Parazoo NC Bowman K Fisher JB Frankenberg C Jones DBA Cescatti A Peacuterez-947

Priego Oacute Wohlfahrt G amp Montagnani L (2014) Terrestrial gross primary production 948

inferred from satellite fluorescence and vegetation models Global Change Biology na-na 949

Plascyk JA (1975) MK II Fraunhofer Line Dicsriminator (FLD-II) for airborne and orbital 950

remote-sensing of solar-stimulated luminescence Optical Engineering 14 339-346 951

Porcar-Castell A Tyystjaumlrvi E Atherton J Van der Tol C Flexas J EE P Moreno 952

J Frankenberg C amp Berry JA (2014) Linking chlorophyll a fluorescence to 953

photosynthesis for remote sensing applications mechanisms and challenges Journal of 954

Experimental Botany 65 4065-4095 955

Rascher U Agati G Alonso L Cecci G Champagne S Colombo R Damm A 956

Daumard F De Miguel E Fernandez G Franch B Franke J Gerbig C Gioli B 957

Gomez JA Goulas Y Guanter L Gutierrez-De-La-Camara O Hamdi K Hostert 958

P Jimenez M Kosvancova M Lognoli D Meroni M Miglietta F Moersch A 959

Moreno J Moya I Neininger B Okujeni A Ounis A Palombi L Raimondi V 960

Schickling A Sobrino JA Stellmes M Toci G Toscano P Udelhoven T Van der 961

Linden S amp Zaldei A (2009) CEFLES2 The remote sensing component to quantify 962

photosynthetic efficiency from the leaf to the region by measuring sun-induced 963

fluorescence in the oxygen absorption bands Biogeosciences 6 1181-1198 964

Richter R amp Schlaumlpfer D (2002) Geo-atmospheric processing of airborne imaging 965

spectrometry data Part 2 Atmospherictopographic correction International Journal of 966

Remote Sensing 23 2631-2649 967

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

37

Rossini M Meroni M Migliavacca M Manca G Cogliati S Busetto L Picchi V 968

Cescatti A Seufert G amp Colombo R (2010) High resolution field spectroscopy 969

measurements for estimating gross ecosystem production in a rice field Agricultural and 970

Forest Meteorology 150 1283-1296 971

Rossini M Nedbal L Guanter L Ač A Alonso L Burkart A Cogliati S Colombo 972

R Damm A Drusch M Hanus J Julitta T Kokkalis P Moreno J Panigada C 973

Pinto F Schickling A Schuumlttemeyer D Zemek F amp Rascher U (2015) Red and far-974

red sun-induced chlorophyll fluorescence as a measure of plant photosynthesis 975

Geophysical Research Letters (accepted for publication) 976

Ruimy A Jarvis PG Baldocchi DD amp B S (1995) CO2 Fluxes over Plant Canopies 977

and Solar Radiation A Review Advances in Ecological Research 26 1-68 978

Running SW Nemani RR Heinsch FA Zhao MS Reeves M amp Hashimoto H 979

(2004) A continuous satellite-derived measure of global terrestrial primary production 980

Bioscience 54 547-560 981

Schaepman ME M J Hueni A DrsquoOdorico P Damm A Weyermann J Schneider F 982

Laurent VCE C P Seidel F Lenhard K Gege P Kuechler C Brazile J Kohler 983

P De Vos L Meuleman K Meynart R Schlaepfer D Kneubuumlhler M amp Itten K 984

(2015) Advanced radiometry measurements and Earth science applications with the 985

Airborne Prism Experiment (APEX) Remote Sensing of Environment 158 207-219 986

Schimel D Pavlick R Fisher JB Asner GP Saatchi S Townsend P Miller C 987

Frankenberg C Hibbard K amp Cox P (2014) Observing terrestrial ecosystems and the 988

carbon cycle from space Global Change Biology na-na 989

Schlaumlpfer D amp Richter R (2002) Geo-atmospheric processing of airborne imaging 990

spectrometry data Part 1 Parametric orthorectification International Journal of Remote 991

Sensing 23 2609-2630 992

Schneider FD Leiterer R Morsdorf F Gastellu-Etchegorry J-P Lauret N Pfeifer N 993

amp Schaepman ME (2014) Simulating imaging spectrometer data 3D forest modeling 994

based on LiDAR and in situ data Remote Sensing of Environment 152 235-250 995

Schroter D Cramer W Leemans R Prentice IC Araujo MB Arnell NW Bondeau 996

A Bugmann H Carter TR Gracia CA de la Vega-Leinert AC Erhard M Ewert 997

F Glendining M House JI Kankaanpaa S Klein RJT Lavorel S Lindner M 998

Metzger MJ Meyer J Mitchell TD Reginster I Rounsevell M Sabate S Sitch 999

S Smith B Smith J Smith P Sykes MT Thonicke K Thuiller W Tuck G 1000

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

38

Zaehle S amp Zierl B (2005) Ecosystem service supply and vulnerability to global change 1001

in Europe Science 310 1333-1337 1002

Sitch S Smith B Prentice IC Arneth A Bondeau A Cramer W Kaplan JO Levis 1003

S Lucht W Sykes MT Thonicke K amp Venevsky S (2003) Evaluation of ecosystem 1004

dynamics plant geography and terrestrial carbon cycling in the LPJ dynamic global 1005

vegetation model Global Change Biology 9 161-185 1006

Stewart DW Costa C Dwyer LM Smith DL Hamilton RI amp Ma BL (2003) 1007

Canopy structure light interception and photosynthesis in maize Agronomy Journal 95 1008

1465-1474 1009

Trnka M Dubrovsky M amp Zalud Z (2004) Climate change impacts and adaptation 1010

strategies in spring barley production in the Czech Republic Climatic Change 64 227-1011

255 1012

Turner DP Ritts WD Cohen WB Maeirsperger TK Gower ST Kirschbaum AA 1013

Running SW Zhao MS Wofsy SC Dunn AL Law BE Campbell JL Oechel 1014

WC Kwon HJ Meyers TP Small EE Kurc SA amp Gamon JA (2005) Site-level 1015

evaluation of satellite-based global terrestrial gross primary production and net primary 1016

production monitoring Global Change Biology 11 666-684 1017

van der Tol C Berry J Campbell PKE amp Rascher U (2014) Models of fluorescence 1018

and photosynthesis for interpreting measurements of solar-induced chlorophyll 1019

fluorescence Journal of Geophysical Research Biogeosciences IN PRESS 1020

van der Tol C Verhoef W amp Rosema A (2009a) A model for chlorophyll fluorescence 1021

and photosynthesis at leaf scale Agricultural and Forest Meteorology 149 96-105 1022

van der Tol C Verhoef W Timmermans A Verhoef A amp Su Z (2009b) An integrated 1023

model of soil-canopy spectral radiances photosynthesis fluorescence temperature and 1024

energy balance Biogeosciences 6 3109-3129 1025

Verhoef W (1984) Light scattering by leaf layers with application to canopy reflectance 1026

modeling The SAIL model Remote Sensing of Environment 16 125-141 1027

Verhoef W amp Bach H (2003a) Remote sensing data assimilation using coupled radiative 1028

transfer models Physics and Chemistry of the Earth 28 3-13 1029

Verhoef W amp Bach H (2003b) Simulation of hyperspectral and directional radiance 1030

images using coupled biophysical and atmospheric radiative transfer models Remote 1031

Sensing of Environment 87 23-41 1032

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

39

Verhoef W amp Bach H (2007) Coupled soil-leaf-canopy and atmosphere radiative transfer 1033

modeling to simulate hyperspectral multi-angular surface reflectance and TOA radiance 1034

data Remote Sensing of Environment 109 166-182 1035

Verhoef W Jia L Xiao Q amp Su Z (2007) Unified optical-thermal four-stream radiative 1036

transfer theory for homogeneous vegetation canopies Ieee Transactions on Geoscience 1037

and Remote Sensing 45 1808-1822 1038

Webb EK Pearman GI amp Leuning R (1980) Correction of flux measurements for 1039

density effects due to heat and water-vapour transfer Quarterly Journal of the Royal 1040

Meteorological Society 106 85-100 1041

Wu C Niu Z Tang Q amp Huang W (2010) Revised photochemical reflectance index 1042

(PRI) for predicting light use efficiency of wheat in a growth cycle validation and 1043

comparison International Journal of Remote Sensing 31 2911-2924 1044

Yang X Tang J Mustard JF Lee JE Rossini M Joiner J Munger JW Kornfeld 1045

A amp Richardson AD (2015) Solar-induced chlorophyll fluorescence that correlates with 1046

canopy photosynthesis on diurnal and seasonal scales in a temperate deciduous forest 1047

Geophysical Research Letters 1048

Zarco-Tejada PJ Berni JAJ Suarez L Sepulcre-Canto G Morales F amp Miller JR 1049

(2009) Imaging chlorophyll fluorescence with an airborne narrow-band multispectral 1050

camera for vegetation stress detection Remote Sensing of Environment 113 1262-1275 1051

Zarco-Tejada PJ Gonzalez-Dugo V amp Berni JAJ (2012) Fluorescence temperature and 1052

narrow-band indices acquired from a UAV platform for water stress detection using a 1053

micro-hyperspectral imager and a thermal camera Remote Sensing of Environment 117 1054

322-337 1055

Zarco-Tejada PJ Morales A Testi L amp Villalobos FJ (2013) Spatio-temporal patterns 1056

of chlorophyll fluorescence and physiological and structural indices acquired from 1057

hyperspectral imagery as compared with carbon fluxes measured with eddy covariance 1058

Remote Sensing of Environment 133 102-115 1059

Zhang Q Middleton EM Cheng YB amp Landis DR (2013) Variations of foliage 1060

chlorophyll fAPAR and foliage non-chlorophyll fAPAR (fAPARchl fAPARnonchl) at the 1061

Harvard Forest Ieee Journal of Selected Topics in Applied Earth Observations and Remote 1062

Sensing 6 2254-2264 1063

Zhang Q Middleton EM Gao BC amp Cheng YB (2012) Using EO-1 hyperion to 1064

simulate HyspIRI products for a coniferous forest The fraction of par absorbed by 1065

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

40

chlorophyll (fAPAR chl) and leaf water content (LWC) Ieee Transactions on Geoscience 1066

and Remote Sensing 50 1844-1852 1067

Zhang Y Guanter L Berry JA Joiner J van der Tol C Huete A Gitelson A Voigt 1068

M amp Koumlhler P (2014) Estimation of vegetation photosynthetic capacity from space-1069

based measurements of chlorophyll fluorescence for terrestrial biosphere models Global 1070

Change Biology 20 3727-3742 1071

1072

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

41

Appendix-A APEX data calibration 1073

The calibration of APEX images is based on the APEX Processing and Archiving Facility 1074

(PAF) which is the combined software and hardware designed to generate calibrated level 1 1075

at-sensor radiances from APEX raw image streams (Hueni et al 2009) Calibration 1076

coefficients for the radiometric geometric and spectral calibrations are generated from system 1077

calibrations at the APEX Calibration Home Base (CHB) at DLR Oberpfaffenhofen (Gege et 1078

al 2009) These CHB data are stored and processed within the APEX Calibration Information 1079

System (CAL IS) producing the so-called calibration cubes holding coefficients for each 1080

spatio-spectral pixel (Hueni et al 2013) 1081

For the multi-temporal dataset employed within this study all flight lines were 1082

reprocessed from the original raw image streams to ensure that the same version of the 1083

processor was applied The APEX PAF and CAL IS have significantly evolved over the 1084

timeframe of the dataset with calibrations getting more accurate over the years To ensure 1085

consistency all data were calibrated using coefficients obtained in May 2013 1086

The processor was specifically configured to retain the highest information content in 1087

the VNIR channel including i) no spectral resampling to compensate for spectral shifts and 1088

misregistrations (DOdorico et al 2010 DOdorico et al 2011) ii) no compensation of 1089

spectral shift related changes in radiometry (Hueni et al 2015) and iii) no spatial 1090

misregistration correction The calibration was thus effectively reduced to a linear dark 1091

current correction over time a de-smearing of the visible-NIR channel to compensate for 1092

sensor smear effects the application of radiometric gains and offsets followed by a destriping 1093

routine working in the spatial dimension only to remove pushbroom related along track 1094

striping Calibrated radiance data were directly used for the fluorescence retrieval (see Section 1095

24) while the feasibility to use APEX for SIF760 retrievals was theoretically demonstrated in 1096

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

42

a study of Damm et al (2011) Resulting SIF760 maps were afterwards geo-rectified using a 1097

parametric geocoding approach (Schlaumlpfer and Richter 2002) 1098

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

43

Appendix-B Retrieval of SIF760 from airborne data 1099

The quantification of SIF measured with ground or airborne-based spectrometers 1100

requires separating emitted SIF from the reflected radiance flux The Fraunhofer Line Depth 1101

(FLD) approach introduced by Plascyk (1975) serves as a de facto standard for medium 1102

resolution instruments such as APEX (Meroni et al 2009) and was applied in our study The 1103

FLD method uses atmospheric absorption bands characterized by lower incident light 1104

compared to wavelength regions outside of these bands This configuration increases the 1105

spectral information and allows the separation of SIF from reflected radiation In this study 1106

we used the broad O2-A oxygen absorption band around 760 nm and quantified the infilling of 1107

SIF760 using radiance measurements inside (subscript i) and outside (subscript o) of the O2-A 1108

band 1109

For the airborne case only atmospheric components can be simulated with 1110

MODTRAN-5 assuming perfect knowledge of the acquisition conditions The reflectance 1111

terms for diffuse and direct irradiance as well as the SIF760 contribution are usually unknown 1112

Assuming Lambertian surface reflectance behavior which introduces some inaccuracies but 1113

allows formulating the radiative transfer (RT) for radiance measurements at sensor level (LAtS) 1114

to obtain SIF760 both measurements inside and outside of the absorption band can be 1115

expressed as 1116

E9FA lang3H IJK LMNrangP QlangRSE rang T UlangVWWH VXWH rang∙langVYYH VYXH rangZH

Hlang[XXH rang T ABCHUlangVWWH ranglangVXWH rangZHlang[XXH rang ] ^ _ (B1) 1117

where E0 is the top-of-atmosphere irradiance including diffuse and direct irradiance 1118

components θil is the illumination zenith angle ρso is the path scattered radiance τss is the 1119

direct transmittance for sunlight τoo is the direct transmittance in view direction τsd is the 1120

diffuse transmittance of the atmosphere for sunlight τdo is the hemispherical-directional 1121

transmittance in view direction and ρdd is the spherical albedo If these atmospheric 1122

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

44

components are obtained from MODTRAN-5 simulations (Damm et al 2015 Verhoef and 1123

Bach 2003a b 2007) the radiative transfer equation (Eq B1) only contains four unknowns 1124

namely Ri Ro SIFi and SIFo The 3FLD method (Maier et al 2003) which is an adaptation of 1125

the original FLD method proposed by (Plascyk 1975) was used to linearly relate R and SIF 1126

inside and outside of the O2-A absorption band SIF can be finally retrieved with 1127

13 -0 abcMUdWecWlangRff_ rangZghcWUdMecMlangRff^ rangZiUdWecWlangRff_ rangZghUdMecMlangRff^ rangZ j

langVWWM ranglangVXWM rang (B2) 1128

kE lE9FA m lang3HW IJK LMNrangP langRSE rangn ] ^ _ (B3) 1129

E lang3HW IJK LMNrangP langoSSE TopSE rang ∙ langoE opE rang qEprs T Eprt] ^ _ (B4) 1130

r S13rlangoSSr rang T langopSr rang v13S$langoSSS rang T langopSS rangw (B5) 1131

where B is the factor relating SIFi and SIFo and was fixed to a value of 08 justified by 1132

simulations and experiments (Rascher et al 2009 Alonso et al 2008) Xj equals the at-1133

sensor radiance (reflected plus emitted radiation) without path radiance contribution and Ej 1134

expresses surface irradiance as measured at the sensor level A is the factor relating Ri and Ro 1135

and was derived from linear interpolation of R of the left (758 nm) and right (771 nm) O2-A 1136

band shoulder with 1137

xx+ (B6) 1138

y ```-0```z and y `-0`z

```z (B7) 1139

1140

Because APEX suffered from slight spectral smile effects (DOdorico et al 2010) 1141

updated center wavelength positions were involved in the convolution approach The spectral 1142

shift quantification for each pixel across track was based on the Cr method as described in 1143

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

45

Meroni et al (2010) The method assumes a high correlation between a measured and a 1144

convolved spectrum and iteratively minimizes a cost function which is 1 minus the Pearson 1145

correlation coefficient (r) calculated for both signals Remaining artifacts of the spectral shift 1146

correction and slightly inaccurate estimates of atmospheric properties cause SIF760 retrieval 1147

uncertainties We therefore applied a semi-empirical correction as described in Damm et al 1148

(2014) for each scan line across the track This technique employs reference surfaces which 1149

are free of any SIF760 emission (eg bare soil) to adjust the upward transmittance term inside 1150

the absorption feature oSSr The retrieval of SIF760 from top-of-canopy radiance data measured 1151

on ground follows the same approach as described in Eqs B2-B7 under the assumption that 1152

the surface irradiance can be approximated with the measurement of a reference panel and 1153

that ρso τdo and τdo can be set to 0 Furthermore the spectral resolution of APEX can be 1154

considered suboptimal for SIF760 retrievals although Damm et al (2011) demonstrate that 1155

APEX with its spectral resolution of 54 nm at 760 nm is sensitive for SIF760 However 1156

retrieved SIF760 values are too high and in an absolute sense incorrect We used reference 1157

SIF760 data obtained in parallel to the airborne acquisitions using the ASD field spectrometers 1158

(Section 23) to estimate the actual offset The observed scaling factor of 193 was afterwards 1159

applied to the retrieved SIF760 values to transfer them to a realistic value range 1160

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

46

Appendix-C SIF retrieval accuracy 1161

We applied a quantitative accuracy evaluation of resulting SIF760 maps using ground 1162

measured SIF760 values (Figure C1) Considering all reference data acquired during five 1163

campaigns a good linear relationship was achieved between ground and airborne-based 1164

SIF760 values (r2 = 071) but with a proportionality factor much higher than 1 (Figure C1-A) 1165

This experimentally obtained systematic overestimation of APEX SIF760 is comparable to 1166

simulation results documented in Damm et al (2011) Further we found slightly changing 1167

correlations between the different years (several linear models in Figure C1-A) which are 1168

likely caused by canopy structural effects spectral and radiometric performances of used 1169

ground instruments or slight deviations of the performance of APEX over time 1170

To compensate for the overestimation of APEX SIF760 we applied an empirical correction 1171

factor of 193 derived from the comparison of reference data and APEX which resulted in a 1172

rRMSE of 289 (Figure C1-B) 1173

1174

lt Figure C1 gt 1175

1176

The Laegeren site was covered with three adjacent flight lines having an overlap of 1177

approximately 10 (cf Figure 3) This configuration provides the opportunity to 1178

quantitatively evaluate the consistency of SIF retrievals obtained from different measurements 1179

and view zenith angles (ie plusmn 135deg) The comparison of SIF760 retrievals obtained from flight 1180

lines 1 and 2 reveals a moderate linear relationship for homogeneous agricultural surfaces (r2 1181

= 081 rRMSE = 57) We observed a systematic underestimation of flight line 1 compared 1182

to flight line 2 (factor 06) for a view zenith angle difference of ~26deg for SIF760 indicating 1183

either a directionality of SIF760 emissions or an insufficient decoupling of surface HCRF 1184

affected by reflectance anisotropy With increasing surface heterogeneity scattering increases 1185

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

47

and the agreement of retrieved SIF from both flight lines slightly decreases (ie for forested 1186

areas r2 of 05 and rRMSE of 88) This lower correlation can be mainly associated to slight 1187

geometric mis-registrations (ie max 2 pixel ~ 40 m) in combination with surface 1188

heterogeneity (eg shaded and sunlit tree crowns) and emission anisotropy of SIF760 Also 1189

negative SIF760 values are present in the SIF760 maps and caused by sensor noise and 1190

especially for forests due to shadowing effects (Damm et al 2015) 1191

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

48

Appendix-D Technical aspects determining the SIF760 retrieval accuracy 1192

A rigorous evaluation of the SIF760 retrieval accuracy is crucial to judge the reliability 1193

of our assessment on the relationship between SIF760 and GPPEC across ecosystems 1194

Furthermore additional factors imposing variations on SIF760 retrieved from airborne 1195

spectroscopy data that cannot be related to the physiological origin of SIF760 needs to be 1196

critically discussed to justify the proposed approach 1197

Ground-based measurements of SIF760 were made using field spectrometers to validate 1198

the airborne retrievals Field spectroscopy provides a good reference for validating airborne 1199

retrievals as it is likely not affected by biases resulting from atmospheric composition 1200

estimates (eg aerosol load) due to the simultaneous measurement of irradiance fluxes at 1201

canopy level using a reference panel The reliability of the applied field spectroscopic 1202

approach to measure SIF760 was previously demonstrated using field studies (Damm et al 1203

2010 Liu and Cheng 2010 Meroni et al 2009 Rascher et al 2009) and theoretical 1204

assessments (Damm et al 2011) It must be noted that field spectroscopic approaches are 1205

subject to uncertainties particularly for heterogeneous vegetation canopies Damm et al 1206

(2015) discussed retrieval errors caused by illumination effects likely introducing slight 1207

errors in reference SIF760 This source of uncertainty must be considered when judging the 1208

accuracy of APEX SIF760 The nevertheless good agreement between ground and airborne 1209

observations indicated that atmospheric absorption and scattering effects could be sufficiently 1210

compensated and thus demonstrated the reliability of retrieved SIF760 measurements In 1211

particular the empirical correction applied making use of high spatial resolution airborne IS 1212

data in combination with reference surfaces not emitting SIF760 allowed to bypass retrieval 1213

uncertainties caused by atmospheric effects (Damm et al 2014) 1214

Instrumental effects in combination with retrieval methods were found to cause 1215

systematic differences in SIF760 retrievals (Damm et al 2011) the comparatively low 1216

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

49

resolution of APEX around the O2-A band used to retrieve SIF760 (ie ~54 nm) resulted in an 1217

overestimation of APEX SIF760 compared to retrievals from high resolution instruments 1218

Indeed maximum APEX SIF760 values are 6 mW m-2 sr-1 nm-1 while typical SIF760 values 1219

obtained from high resolution instruments range between 0 and 3 mW m-2 sr-1 nm-1 (Guanter 1220

et al 2013) Considering SIF760 values retrieved from ASD data in original resolution (3 nm) 1221

APEX showed an overestimation of 1102 which confirmed that low spectral resolution 1222

resulted in an overestimation of SIF760 The reasoning for this absolute error is given by the 1223

sub-optimal spectral resolution of APEX with 54 nm at 760 nm causing a reduced sensitivity 1224

for SIF760 on the one hand and a wider distance of the O2-A shoulder values used to 1225

parameterize the 3FLD approach on the other hand The bias was corrected using the ASD 1226

reference data and does not impact the results of our analysis Further the relative variation of 1227

SIF760 is unaffected by the bias 1228

Surface heterogeneity complicates the retrieval of SIF760 According to Damm et al 1229

(2015) SIF760 retrievals for forest sites are especially subject to uncertainties due to 1230

assumptions on surface irradiance Crown shapes and associated surface orientations 1231

determine the intensity and quality of irradiance including fractional amounts of diffuse and 1232

direct irradiance To account for pixel wise changes in surface irradiance high resolution 1233

digital object models (DOM) (eg derived from LiDAR observations) are required In 1234

absence of such a high resolution DOMrsquos crowns are assumed to be flat which results in 1235

uncertainties of up to 58 in case of cast shadows (Damm et al 2015) Since we used 1236

averaged observations over larger areas representing flux tower footprints and most of our 1237

data were flown under comparable illumination conditions (ie around solar noon) we 1238

expect these potential biases to be less influential in this analysis 1239

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

1

Tables 1

Table 1 Date and time of APEX flights and corresponding eddy covariance flux tower 2

measurements for the three investigated ecosystems (ie grassland cropland mixed 3

temperate forest) 4

Date Time [UTC] Location Ecosystem Flux tower

17062009 950 Laegeren forest 1 tower

17062009 1030 Oensingen cropland 1 tower

17062009 1030 Oensingen grassland 2 tower

25062010 1030 Oensingen cropland 1 tower

25062010 1030 Oensingen grassland 2 tower

26062010 1500 Laegeren forest 1 tower

26062010 1030 Oensingen cropland 1 tower

26062010 1030 Oensingen grassland 2 tower

29062010 1030 Laegeren forest 1 tower

26062011 1030 Laegeren forest 1 tower

28062011 1030 Oensingen cropland 1 tower

28062011 1030 Oensingen grassland 1 tower

16062012 1025 Laegeren forest 1 tower

16062012 900 Oensingen cropland 1 tower

16062012 1100 Oensingen cropland 1 tower

12072013 730 Laegeren forest 1 tower

03092013 1050 Laegeren forest 1 tower

5

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

2

Table 2 Availability of reference SIF760 data obtained from field spectroscopy measurements 6

to validate APEX based SIF760 retrievals 7

Date time [UTC] Location Measured crops Nr points

17062009 1030 Oensingen maize rye pea sugar beet wheat white clover 29 points

25062010 1030 Oensingen maize pea sugar beet wheat white clover 15 points

16062012 900 Oensingen carrot maize winter wheat 11 points

16062012 1100 Oensingen carrot maize winter wheat 10 points

8

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

3

Table 3 Goodness of fit for the hyperbolic model describing the relationship between GPPEC 9

and SIF760 O PRI and MCARI2 for the three ecosystem types perennial grassland cropland 10

and mixed temperate forest Calculated statistics are based on all data acquired from 2009-11

2013 as described in Table 1 12

Optical variable Ecosystem Regression coefficient Relative RMSE []

SIF crops 088 355

grass 059 2708

forest 048 1588

PRI crops 0008 1176

grass 095 860

forest 009 2179

MCARI2 crops 0005 1029

grass 098 550

forest 034 1798

13

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

1

Figures 1

2

Figure 1 Study sites A Oensingen agricultural site B Laegeren forest site The white 3

rectangles represent in A the boundary of investigated fields that comprise most of the flux-4

tower footprints and in B a coarse outline of the footprint area of the flux-tower The white 5

crosses show the locations of the flux-towers Background RGB images from Google-Earth 6

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

2

7

Figure 2 APEX reflectance and SIF760 image for the Oensingen site acquired on the 2606 8

2010 1030 UTC A top-of-canopy hemispherical-conical reflectance data as true color 9

composite B SIF760 map The white boxes indicate the approximate position of the flux 10

tower 11

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

3

12

Figure 3 APEX reflectance and SIF760 image for the Laegeren site acquired on the 2606 13

2010 1510 UTC A top-of-canopy hemispherical-conical reflectance data as true color 14

composite B SIF760 map The black arrows indicate the flight direction of flight line 1 to 3 15

(FL1-FL3) The white line marks a transect along the overlapping region of FL1 and FL2 The 16

white boxes indicate the approximate position of the flux tower 17

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

4

18

Figure 4 Relationships between GPPEC from the eddy-covariance measurements and SIF760 O 19

instantaneously measured between 2009 and 2013 (see Table 1) with the airborne 20

spectrometer APEX (A) PRI index (B) and MCARI2 index (C) for three vegetation types 21

(black dots perennial grassland dark grey rectangles cropland including winter wheat pea 22

and barley grey triangles mixed temperate forest) The solid curves correspond to a 23

hyperbolic model fitted to the measured data The dashed curves show the extrapolated 24

behavior 25

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

5

26

Figure 5 Daily light response curves from eddy flux measurements for A cropland B 27

perennial grassland and C mixed temperate forest Data were acquired between 2009 and 28

2013 (see Table 1) Grey curves represent a fitted hyperbolic model to approximate the 29

behavior of measured PAR-GPP values Crosses show GPPEC measurements that were used 30

for the SIF760 O - GPPEC comparison based on the time of APEX flights PAR values are 31

binned in 20 mW m-2

sr-1

nm-1

increments 32

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

6

33

Figure 6 Leaf level relationships between photosynthetic (LUEp) and fluorescence yield 34

(LUEf) for a typical C3 crop Dependency of LUEp (A) and LUEf (B) on APAR C 35

Relationship between LUEp and LUEf D Relationship between APAR and GPPL GPPL was 36

calculated as APARLUEp E Relationship between APAR and SIF760 SIF760 was calculated 37

as APARLUEf assuming a constant relationship between the SCOPE output SIFtotalL 38

(integral between 640-850nm) and SIF760L F Relationship between SIF760 L and GPPL All 39

data were simulated using SCOPE The grey areas indicate illumination conditions 40

comparable to APEX flights 41

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

7

42

Figure 7 Canopy level relationships between photosynthetic (LUEp) and fluorescence yield 43

(LUEf) for different canopy representations of a C3 cropland in SCOPE Dependency of 44

LUEp (A) and LUEf (B) on APAR LUEp was calculate as GPPCAPAR and LUEf was 45

directly obtained from SCOPE C Relationship between LUEp and LUEf D Relationship 46

between APAR and GPPC GPPC was directly obtained from SCOPE E Relationship 47

between APAR and SIF760C SIF760C was directly obtained from SCOPE F Relationship 48

between SIF760C and GPPC The black dashed line corresponds to a C3 cropland with LAI = 49

30 fraction of brown pigments (Cs) = 00 and a spherical (sp) leaf inclination distribution 50

function (LIDF) For the grey line Cs was changed to 20 for the black dashed line the LAI 51

was changed to 10 and for the grey dasheddotted line the LIDF was changed to plagiophile 52

(pg) The grey areas indicate illumination conditions comparable to APEX flights 53

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

8

54

Figure 8 Impact of temporal scaling on the SIF760C-GPPC relationship SIF760C and GPPC 55

data were simulated using SCOPE Model input parameters air temperature (A) PAR (B) and 56

LAI (C) were measured within the NitroEurope project (httpwwwnitroeuropeeu) at the 57

Speulderbos forest site Netherlands D Relationship between SIF760C and GPPC including a 58

fitted hyperbolic model for daily averages (grey dots and grey line) monthly averages (dark 59

grey triangles and line) and quarterly data (black dots and black dashed line) 60

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67

9

61

Figure C1 Comparison of ground and airborne-based SIF760 retrievals from various crops 62

Data were acquired during four campaigns between 2009 and 2012 in the agricultural site 63

Oensingen (see Table 2) Left Fitted linear regression lines for the four considered campaigns 64

are overlaid in black The dashed line corresponds to the 11 line Right SIF retrievals from 65

APEX after applying an empirical correction factor (193) The black line represents the fitted 66

linear model the dashed line corresponds to the 11 line 67