ECOSTRESS L1 and L2 Calibration and Validation

ECOSTRESS L1 and L2 Calibration and Validation

Simon J. Hook and the ECOSTRESS Team

Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA

(c) 2014 California Institute of Technology. Government sponsorship acknowledged.

National Aeronautics and Space Administration

Outline

2

•  Introduction – Science Use

•  Theoretical Basis –  Methods

•  L1 and L2 Product Flow •  Validation

–  T-val –  R-val

•  Uncertainties •  Summary and Conclusions

Evapotranspiration (drought monitoring)

Surface Energy Balance Models Urban Heat Island Studies

Earth Science Use of LST&E Understanding Climate Change

Theoretical Basis: Planck Formula

⎥⎥⎦

⎤

⎢⎢⎣

⎡−⎟⎟⎠

⎞⎜⎜⎝

⎛=

1exp 25

1

TCCBλλ

λ

constant.radiation second

constant.radiation first re. temperatuabsolute

wavelength=

exitance. spectralblackbody :where

2

1

=

=

=

=

CC

B

Tλ

λ

0

20

40

60

80

100

Ra

dia

nc

e (

W/m

*m*m

)/1

.0e

64 6 8 10 12 14 16 18 20

Wavelength (micrometers)

450K

350K

273.15K

As the temperature increases the peak in the Planck function shifts to shorter and shorter wavelengths

5

Theoretical Basis: Temperature and Spectral Emissivity

Materials are not perfect blackbodies, but instead emit radiation in accordance with their own characteristics. The ability of a material to emit radiation can be expressed as the ratio of the spectral radiance of a material to that of a blackbody at the same temperature. This ratio is termed the spectral emissivity:

Blackbody)( / Material)( LL λλλε =

(T) ΒL ε=

or for a material at a given wavelength

∫+⋅−+⋅⋅=↓

⋅ iiiiiSiiii dPTBLeTBeL τθτθτθ ))(()1()()()()(

Surface Emission

Surface Reflection

Skin Temperature & Surface Emissivity

Thermal Infrared Radiative Transfer

Surface Radiance

Atmospheric Emission

TOA Radiance

↓iL

)(θ↑iL

!!"#,! = !! !! !!! !"# + ! 1− !!! !!!"# ,!↓ !!! + !!!"# ,!↑ !

Thermal Infrared Radiative Transfer

HyspIRI (nominal)

x xxxx ECOSTRESS

x x x x x x x xMASTER x x x x x x x x x

Thermal Infrared Physics

ECOSTRESS L1/L2 Product Flow

Calibration

Atmospheric Correction

T/E Separation

Atmospheric Profiles,

Scene graybodies, Cloud Masks

System response functions

Blackbody temperatures and

radiances

L2 - Temperature

Product

Process

L0 – Raw Data

L1 – Radiance at Sensor or Brightness Temperature

Calibration Curve

L2 – Radiance at Surface

L2 - Emissivity

Legend

9

ECOSTRESS (ECOsystem

Spaceborne Thermal Radiometer

Experiment on Space Station)

ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer)

MODIS* (Moderate Resolution

Imaging Spectroradiometer) AIRS

(Atmospheric Infrared Sounder)

Satellite ISS (2017)

Terra (2000)

Terra/Aqua (2000/2002)

Aqua (2002)

Algorithm TES Calibration Curve with

WVS TES Calibration

Curve with WVS

TES Calibration Curve with

WVS Regression plus

retrieval Temporal sampling

5 day repeat (varying times)

16 day repeat (1030 AM/PM)

Twice-daily (10:30/1:30 AM/

PM) Twice-daily (10:30 AM/

PM) Nadir Spatial

resolution 40 x 60 m 90 m 1 km 13 km

Spectral resolution

5 TIR bands (8-12 µm)

5 TIR bands (8-12 µm)

7 MIR/TIR bands (3.7-14 µm)

39 ‘hinge-points’

(3.7-15.4 µm) Swath Width 384 km 60 km 2330 km 1650 km

NASA L2 LST&E Product Characteristics

* VIIRS also available at 1 km

L1 Calibration and Validation •  Calibration

– Two blackbodies, one temperature controlled

•  Validation – Water targets

Lake Tahoe CA/NV Salton Sea CA

PHyTIR Test Configuration

Room-temperature reference blackbody. Flat plate with corrugated, painted surface. Emissivity <1 acceptable.

• Instrument is in air. Vacuum enclosure around focal-plane is evacuated (to be described in detail in mechanical presentation). Scan mirror rotating.

Variable-temperature blackbody: ≥ room temperature. Flat plate with corrugated, painted surface. Emissivity <1 acceptable.

MCS Target Projector with slit source. Will underfill PHyTIR aperture.

Small hot source. 295K to > 500 K.

Test Sources Placed Within PhyTIR Scan Range

03/29/2012 11

Radiometric Spatial

Saturation

Blackbodies Added for ISS

2012-05-20 12

Primary CoolerElectronics A (Thales COTS)Redundant CoolerElectronics A (Thales COTS)

16W

Filters andDetector

M3

M2

M1

300K Blackbody

Nadir Baffle

60 K Focal Plane

200 K Cold Shield

Vacuum EnclosureFor Ground Testing

190K

55KScanMirror

Motor andEncoder

Scan MotorControl (SIMNew Layout)

Card Cage(SMAP build to print)

64 GB Memory(SMAP Heritage)

Power Supply(SMAP

Build to Print)

FPAVirtex-5Digital Board(OCO-3

Build to Print)

Window

Focal-PlaneInterface

32 Channels10 Mpix/sec(M 3 Heritage)

ExistingRadiometer

Build to printaccumulator assembly

Build to printelectronics

COTSelectronics

New design or layoutbased on heritage

0.5W4.4W

45W

45W

CryocoolerCompressor A80W80W

28V PowerConditioningElectronics(OCO-3 buildto Print)C&DH Virtex-5

Digital Board(OCO-3

Build to Print)

10W

10W

4W

CryocoolerCompressor B

1553B120Vdc

Fluid CoolingLoop Return

Fluid CoolingLoop Supply

Cooling LoopAccumulator

Assembly (OCO-3Build to Print)

Ethernet12W

15.5W20W

13.1W

15W

ISS JEM-EF

HeaterControl

Housekeeping

RadiationFrom Scene

15W

1.3W

10WFoldMirror

SurvivalHeaters

120Vdc

JEM-EF Payload Interface Unit

10W

340KBlackbody

(45W)

48V PowerConditioningElectronics #1(RapidScatHeritage)

PowerConditioningElectronics #2

15.5W Primary CoolerElectronics B (Thales COTS)Redundant CoolerElectronics B (Thales COTS)

9W

ECOSTRESS Design Layout

2000-09-20-D

•  Large 35 km x 16 km •  High 2 km •  Available year round (does not

freeze in winter). •  Homogenous compared with land. •  Large annual temperature range

5-25 C. •  Freshwater (kind to instruments!) •  Good infrastructure and easy access.

14

TB3 Installed 11-04-2002

Skin temperature

Air temperature & Rel. Humidity Wind Speed

& Direction

Logging System

Batteries Bulk Water Temperature

3m 2012-05-20

Radiometer Calibration

Residual Error after Calibration

-100.00

-80.00

-60.00

-40.00

-20.00

0.00

20.00

40.00

60.00

80.00

100.00

0.00 500.00 1000.00 1500.00 2000.00 2500.00 3000.00 3500.00 4000.00 4500.00

Rad

Tem

pera

ture

Err

or (m

K)

Chubb Temperature (C*100)

Rad 409 Validation at 4C, 4%

Calibration runValidation run

Photo courtesy Brant Allen UCD, 2012-03-06 17 2012-05-20

Data clearly fall on 1x1 line. High radiance values from Salton Sea

5.25

5.75

6.25

6.75

7.25

7.75

8.25

8.75

9.25

9.75

10.25

5.25 5.75 6.25 6.75 7.25 7.75 8.25 8.75 9.25 9.75 10.25

OB

C R

adia

nce

(W/m

2.µm

.sr)

Vicarious Radiance (W/m2.µm.sr)

ASTER Std Filter Vicarious and OBC Thermal Infrared Derived Radiances at L. Tahoe and Salton Sea CY2000-2012, Std Filter, v3.0x

Band 10 (8.29 µm)

Band 11 (8.63 µm)

Band 12 (9.08 µm)

Band 13 (10.66 µm)

Band 14 (11.29 µm)

1x1 line

Std FilterBuoy Range < 0.1km Std Dev > 0.2N Good = 9Skin Effect < 1.0

Excellent best fit lines obtained in all bands. Rsquared is typically 0.98. Data follow 1x1 line.

y = 1.0034x - 0.0388R² = 0.9874

y = 1.0068x - 0.0635R² = 0.99

y = 1.0184x - 0.1215R² = 0.9906

y = 1.0155x - 0.0994R² = 0.9896

y = 1.0172x - 0.1219R² = 0.9875

5.25

5.75

6.25

6.75

7.25

7.75

8.25

8.75

9.25

9.75

10.25

5.25 5.75 6.25 6.75 7.25 7.75 8.25 8.75 9.25 9.75 10.25

OB

C R

adia

nce

(W/m

2.µm

.sr)

Vicarious Radiance (W/m2.µm.sr)

ASTER Std Filter Vicarious and OBC Thermal Infrared Derived Radiances at L. Tahoe and Salton Sea, CY2000-2012, Std Filter, v3.0x

Band 10 (8.29 µm) Band 11 (8.63 µm)

Band 12 (9.08 µm) Band 13 (10.66 µm)

Band 14 (11.29 µm) 1x1 line

Linear (Band 10 (8.29 µm)) Linear (Band 11 (8.63 µm))

Linear (Band 12 (9.08 µm)) Linear (Band 13 (10.66 µm))

Linear (Band 14 (11.29 µm))

Band 13 Band 14

Band 10 Band 11 Band 12


If you look at the individual points they typically scatter between +/- 1K with a few outliers. ASTER specification for 270-340K is 1K. Need to check outliers more!

-4

-3

-2

-1

0

1

2

3

4

03/01/00 03/01/01 03/01/02 03/01/03 02/29/04 02/28/05 02/28/06 02/28/07 02/28/08 02/27/09 02/27/10 02/27/11 02/27/12

Vica

rious

-O

BC

BT

Time

Delta Vicarious and OBC Brightness Temp. for ASTER TIR Channels at L. Tahoe and Salton Sea CY2000-2012, Std Filter v3.x

b10

b11

b12

b13

b14


If look at mean radiance differences for each year they are typically less than 0.5% or 0.3K.

-3

-2

-1

0

1

2

3

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 All years

Avg

% R

adia

nce

Diff

(v-o

)/v *

100

Year

Percent Radiance Change in TIR Channels for ASTER at Lake Tahoe and Salton Sea CY2000-2012, Std Filter, v3.x

b10 b11 b12 b13 b14

MODIS other TIR bands

Band 31: 11.01 µm 1% radiance change ≈ 0.65K

ASTER specification is 1K for at sensor temps between 270-340K


If look at individual bands see that night data problem in 2005

-3

-2

-1

0

1

2

3

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Avg

% R

adia

nce

Diff

(v-o

)/v *

100

Year

Percent Radiance Change in TIR Channels for ASTER at Lake Tahoe 2000-2012, Std Filter, Day/Night sep. v3.x

b10-night b10-day b11-night b11-day b12-night

b12-day b13-night b13-day b14-night b14-day




If look at different between day and night for two clear TIR channels (minimum atmospheric effect) see that daytime values tend to be lower than nighttime. Cause for this is unknown but is observed with other sensors, e.g. Landsat. Likely explanation is RT model or bulk-skin effect.

-3

-2

-1

0

1

2

3

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 All years

Avg

% R

adia

nce

Diff

(v-o

)/v *

100

Year

Percent Radiance Change in TIR Channels for ASTER at Lake Tahoe and Salton Sea CY2000-2012, Std Filter v3.x

b13-night b14-night b13-day b14-day




Channels 13 and 14 used to check bias since least affected by the atmosphere. Band 10 is most affect by the atmosphere.

-5

-4

-3

-2

-1

0

1

2

3

4

5

8 8.5 9 9.5 10 10.5 11 11.5

Del

ta V

icar

ious

-O

BC

Brig

htne

ss T

empe

ratu

re

Band

Delta Vicarious and OBC Brightness Temp. as a function of Wavelength at L. Tahoe and Salton Sea CY2000-2012, Std Filter v3.x

Band 1 Band 2 Band 3 Band 4

Band 5 Night mean Day mean Day+Night mean


25

Level 2 Validation – Temperature and Emissivity

•  Good correlation at regional scale, but differences when look in detail. Differences due to different spatial, spectral, and temporal characteristics of the sensors, including algorithmic differences.

•  Currently validate data LST&E data from ASTER, MODIS, AIRS, Landsat, VIIRS, will use same techniques with ECOSTRESS

NAALSED Summertime Emissivity (Jul-Sep 2000-2010), Band 12 (9.1 µm), 5km

Number of pixels averaged for a given location ->

Lowest emissivity over southwest

ASTER-GEM Band 12 emissivity (9.1 µm) 5km resolution: 2000-2010 (~112,000 scenes)

Certain regions have far more coverage than others!

ASTER-GEM Band 12 emissivity (9.1 µm) 5km resolution: 2000-2010 (~57,446 scenes)

L2 Validation •  Two methods:

– Temperature validation (Tval) – measure the temperature at the same time as the overpass and compare with the temperature retrieved by the satellite

– Radiance validation (Rval) – measure the emissivity and the atmospheric profile independently. Forward calculate the ground temperature needed to match the at-sensor radiance and compare to retrieved temperature.

Method Requirements Advantages Disadvantages

T-val Accurate radiometer measurement(s) at the same time of overpass

Direct comparison Can also be used to validate calibration of sensor

Requires in situ measurement at time of overpass Difficult to perform over targets where temperatures vary rapidly over short distances

R-val Surface emissivity measurement (not coincident with overpass) Atmospheric profile at the time over overpass

Does not require in situ emissivity measurement at time of overpass

Requires atmospheric profile at time of overpass Requires surface emissivity measurement Indirect measurement cannot be used to validate calibration of sensor

Both approaches are typically used over homogenous targets (either in temperature or emissivity)

-1.50

-1.00

-0.50

0.00

0.50

1.00

1.50

2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 AllYears

Year

Vica

rious

min

us P

rodu

ct D

eriv

ed (K

)

AST08MOD11-L2ASTER and MODIS LST Spec. ±1K

Tahoe LST validation

The MODIS product is accurate to (± 0.2K), while the ASTER product has a bias of 1-2 K due to residual atmospheric correction effects

Great Sands, CO

Killpecker, WY Algodones, CA

White Sands, NM Kelso, CA

Little Sahara, UT

Moses Lake, WA

Stovepipe Wells, CA

Coral Pink, UT

Pseudo-invariant sand dune validation sites

Pseudo-Invariant Sand Dune Validation Sites

Hulley, G. C., Hook, S. J., and A.M. Baldridge, Validation of the North American ASTER Land Surface Emissivity Database (NAALSED) Version 2.0, Remote Sensing of Environment (2009), accepted

ASTER MINUS LAB EMISSIVITY (%)

Dune site Band 10

Band 11

Band 12

Band 13

Band 14

Mean

Algodones 0.68 0.60 0.13 0.02 1.40 0.57 Stovepipe Wells

0.17 0.77 1.02 0.34 0.37 0.53

White Sands 0.34 2.76 0.16 0.92 1.08 1.05 Kelso Dunes 1.57 1.04 1.33 1.91 0.81 1.33 Great Sands 1.44 0.97 1.42 1.64 0.69 1.23 Moses Lake 0.69 0.52 0.42 0.61 1.01 0.65 Sand Mountain 7.74 6.47 9.01 1.82 1.10 5.23

Coral Pink 7.48 6.44 7.32 2.50 1.70 4.90 Little Sahara 3.55 2.39 2.60 0.96 0.19 1.94 Killpecker 2.34 1.99 2.26 1.33 0.81 1.75

< 1.6% (~1 K)

Algodones, CA

Radiance-based LST Validation at Algodones Dunes

Summary and Conclusions

35

•  Land surface temperature and emissivity (LST&E) are important measurements for understanding the earth system. They are used in a wide variety of studies from measuring evapotranspiration to predicting volcanic eruptions.

•  LST&E measurements are available at a variety of spatial, spectral and temporal resolutions and generated using multiple algorithms. Different algorithms result in one product performing better in one region and a different product performing better in a different region.

•  Well understood procedure to calibrate to L1 data using 2 blackbodies but must also validate in-flight to obtain independent validation. Will use large lakes to validate L1 radiance at sensor

•  L2 LST&E data can be validated using two approaches T-val and R-val. Approaches complement each other and allow validation over a broader range of cover types.

National Aeronautics and Space Administration Jet Propulsion Laboratory California Institute of Technology Pasadena, California

Documents

ECOSTRESS L1 and L2 Calibration and Validation