Upload
independent
View
0
Download
0
Embed Size (px)
Citation preview
Agricultural and Forest Meteorology 149 (2009) 1129–1139
Evapotranspiration from understory vegetation in an eastern Siberian boreallarch forest
Shin’ichi Iida a,*, Takeshi Ohta b, Kazuho Matsumoto c, Taro Nakai d, Takashi Kuwada b,Alexander V. Kononov e, Trofim C. Maximov e, Michiel K. van der Molen f, Han Dolman f,Hiroki Tanaka g, Hironori Yabuki h
a Forestry and Forest Products Research Institute, Matsunosato, Tsukuba, Ibaraki 305-8687, Japanb Graduate School of Bioagricultural Sciences, Nagoya University, Nagoya 464-8601, Japanc Graduate School of Agriculture, Kyoto University, Kyoto 606-8502, Japand Institute of Low Temperature Science, Hokkaido University, Sapporo, Hokkaido 060-0819, Japane Institute for Biological Problems of the Cryolithozone, Siberia Division, RAS, 41 Lenin Ave., Yakutsk 678891, Russiaf Department of Geo-Environmental Sciences, Free University, de Boeleaan 1085, 1081 HV Amsterdam, The Netherlandsg Graduate School of Environmental Studies, Nagoya University, Nagoya 464-8601, Japanh Institute of Observational Research for Global Change, Natsushima-cho, Yokosuka, Kanagawa 237-0061, Japan
A R T I C L E I N F O
Article history:
Received 16 July 2008
Received in revised form 20 January 2009
Accepted 3 February 2009
Keywords:
Aerodynamic conductance
Deciduous larch
Decoupling coefficient
Evergreen understory vegetation
Surface conductance
A B S T R A C T
We measured evapotranspiration in an eastern Siberian boreal forest, in which the understory was
cowberry and the overstory was larch, during the entire growing seasons of 2005 and 2006. We
compared evapotranspiration from the understory vegetation above the forest floor EU with
evapotranspiration from the whole ecosystem above the overstory canopy EO. The EU/EO ratio had a
seasonal trend with a flat-bottomed U-shape during the growing season (4 May–30 September). High-
EU/EO ratios at the beginning and end of the growing season were observed because larch, one of the two
sources of EO, was a deciduous tree, while the understory was the evergreen cowberry. The mean daily EU
values during the foliated period of larch (1 June–31 August) were 0.8 and 0.9 mm day�1, or 51.4 and
51.8% of EO in 2005 and 2006, respectively. The understory vegetation was one of the most important
components of the hydrologic cycle in this forest. A significant amount of EU was caused by plant
physiological control, due to the aerodynamic conductance, which was much larger than the surface
conductance, leading to a smaller decoupling coefficient. We found that 71% of EU was caused by the
vapour pressure deficit above the forest floor.
Crown Copyright � 2009 Published by Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Agricultural and Forest Meteorology
journal homepage: www.e lsev ier .com/ locate /agr formet
1. Introduction
Forests cover approximately 30% of the total land surface of theEarth (FAO, 2005). They make effective use of radiation input, asthe albedo is lower and net radiation is higher above forestcanopies than above other land cover types (e.g., McNaughton andJarvis, 1983). Thus forests play an important role in energy andwater partitioning processes. Approximately 25% of the globalforest area is boreal forest, of which around 76% is in Russia (FAO,2005). Investigation of water and energy exchanges between theforest and the atmosphere in boreal forests, especially those inRussia, is therefore essential to our understanding of the globalhydrological cycle and energy balance.
Boreal forests have a number of features that make themdifferent from forests in other climate regions: monotonous
* Corresponding author. Tel.: +81 29 829 8234; fax: +81 29 874 3720.
E-mail address: [email protected] (S. Iida).
0168-1923/$ – see front matter . Crown Copyright � 2009 Published by Elsevier B.V. A
doi:10.1016/j.agrformet.2009.02.003
vegetation, relatively sparse main canopy trees, and denseunderstory vegetation. Black and Kelliher (1989) reviewed earlierpublications that calculated the evapotranspiration from unders-tory vegetation (EU) using lysimeter or chamber methods, orthrough stomatal conductance measurements. They showed thatthe contribution of EU to evapotranspiration by the wholeecosystem (EO) ranged from 8 to 65%. Thus the boreal forestecosystem has two main sources of evapotranspiration: overstorytrees and understory vegetation on the forest floor. To understandthe water and energy exchange processes between the borealforest ecosystem and the atmosphere, we must observe EU andevaluate its contribution to EO.
Since the 1990s, many researchers have attempted to observethe latent and sensible heat fluxes above understory vegetationusing the eddy-covariance (EC) method (Baldocchi and Meyers,1991; Baldocchi and Vogel, 1996; Baldocchi et al., 1997, 2000a;Blanken et al., 1997). The EC has many advantages for EU
measurements; namely, it allows direct evaluation, high time-resolution, spatially averaged observation. The EC method permits
ll rights reserved.
S. Iida et al. / Agricultural and Forest Meteorology 149 (2009) 1129–11391130
measurement of most of the turbulent eddies above the understoryvegetation if the shapes of the power spectra and co-spectra aresimilar to those of the surface boundary layer (Baldocchi et al.,2000a), and if the energy closure ratio is similar to general values(i.e., those reported by Wilson et al., 2002). Constantin et al. (1999)observed EU using the EC method in a boreal Swedish forest for twoshort periods: from 27 May to 23 June 1994, and from 27 June to 7July 1995. They reported that EU contributed typically 10–15% ofEO. The Boreal Ecosystem–Atmosphere Study (BOREAS) project(Sellers et al., 1995) observed EU using the EC method over anentire growing season in a Canadian boreal aspen forest (Blacket al., 1996; Blanken et al., 2001) and in a jack pine forest (Baldocchiet al., 1997). They reported EU/EO ratios of 24% (Black et al., 1996),27% (Blanken et al., 2001), and 20–40% (Baldocchi et al., 1997).Baldocchi et al. (2000b) noted that observations of EO for entiregrowing seasons are very rare in boreal forests. Indeed, for both EO
and EU, there have been very few observations performed overentire growing seasons, except for those of the BOREAS project.
A number of researchers have evaluated EU/EO ratios in Siberianboreal forests. In a pioneering study, Kelliher et al. (1997) observedEO above a Siberian larch canopy using the EC method, and EU usinglysimeters, during 9 days in summer. They reported that 50% of thetotal evapotranspiration emanated from the understory vegeta-tion. A similarly high-EU/EO ratio (54%) was observed in a Siberianpine forest (Kelliher et al., 1998). However, Kelliher and colleaguesobserved EU and EO by the EC method for only 18 days in mid-summer. Ohta et al. (2001) used the EC method to measure EO andsap-flow observations to measure the transpiration of overstorylarch. They estimated EU as the difference between those valuesand reported an EU/EO ratio of 35%. Hamada et al. (2004), whoestimated EO by the EC method and EU by lysimeter measurements,reported that EU/EO ranged from 25 to 50% in a Siberian pine forest.In all these studies, the results were based on short- or long-termEU data obtained by a low-time-resolution method (Kelliher et al.,1997; Hamada et al., 2004), on long-term but indirect EU
estimations (Ohta et al., 2001), or on short-term EU measurementsby the EC method (Kelliher et al., 1998). However, no studies can befound that have used the EC method to observe both EO and EU inSiberian boreal forest over an entire growing season. Thus, our aimwas to observe EU directly for the entire growing season and toevaluate its quantitative contribution to the hydrologic cycle inSiberian boreal forests.
Because the contribution of EU to EO was likely to be large, itwas important to clarify the factor controlling EU in boreal forests.Evapotranspiration is affected by the equilibrium evaporation Eeq,the ‘‘imposed’’ evaporation Eimp, and the decoupling factor V,which indicates the relative importance of Eeq (McNaughton andJarvis, 1983). Baldocchi and Meyers (1991) and Baldocchi et al.(2000a) reported less coupling between EU and the availableenergy at the forest floor AEU, which is the driving energy of Eeq atthe forest floor (i.e., Eeq-U). They argued that large-scale eddiessweep moisture out of the canopy air space before EU can reachEeq-U. As a result, EU is mainly determined by Eimp at the forest floor(i.e., Eimp-U) and is driven by the vapour pressure deficit. ForSiberian boreal forests, Kelliher et al. (1997) obtained similarresults as Baldocchi and Meyers (1991) and Baldocchi et al.(2000a), but their dataset spanned only 9 days. Long-termobservations of EU and the micrometeorology above the unders-tory vegetation are necessary to understand the factors thatquantitatively control EU in Siberian boreal forest. As mentionedabove, most boreal forest is located in Siberia; hence, thisinvestigation may be essential for understanding the hydrologiccycle in boreal forests globally.
We measured EU by the EC method for two growing seasons(2005–2006) in a Siberian boreal forest. Our objectives were to (i)clarify the seasonal variation in the daily contribution of EU to EO,
(ii) evaluate the contribution of EU to EO quantitatively for entiregrowing seasons, and (iii) clarify the energy sources for EU and EO.
2. Materials and methods
2.1. Site description
The observations were made at a site on the west bank of themiddle reaches of the Lena River in eastern Siberia (6281501800N,12981402900E; 220 m a.s.l.), approximately 20 km north of the cityof Yakutsk. The site was situated in a continuous permafrost regionreferred to as the ‘‘Spasskaya Pad’’ experimental forest of theInstitute for Biological Problems of the Cryolithozone, RussianAcademy of Sciences (RAS). The annual mean air temperature was�10.4 8C (Dolman et al., 2004) and annual mean precipitation forthe period from 1998 to 2006 was 259.2 mm (Ohta et al., 2008).During summer, the permafrost near the ground surface thawed,and its front retreated to >1 m in depth (Ohta et al., 2008). Ameteorological observation tower 32 m high was installed at thesite; the topography around the tower was mostly flat and inclinedslightly northwards.
The experimental forest was typical light taiga. The dominantoverstory species was deciduous coniferous larch (Larix cajanderi),which had a mean tree height of 18.0 m, basal area of 27.6 m2 ha�1,and stand density of 840 trees ha�1. The leaf area index (LAI) of thelarch was determined to be 1.6 by a plant canopy analyser (LAI2000;LI-COR Inc., USA) and 2.0 by the litter trap method. The low-standdensity allowed light to easily reach the forest floor, where verydense understory vegetation (evergreen cowberry, Vaccinium vitis-
idaea) covered the ground completely. Although the cowberryunderstory was low (<0.1 m), it had an LAI of 2.1, as measured by thedestructive method (Miyahara, 2005). Therefore, the LAI of theunderstory cowberry was almost equal to that of the overstory larch.
2.2. Observations
2.2.1. Definition of EO and EU
Micrometeorological observations were conducted on thetower during two growing seasons in 2005 and 2006. To evaluateevapotranspiration from the whole ecosystem (EO) and from theunderstory (EU) separately, we installed EC systems at heights of32.0 and 3.3 m (Table 1). We used the EC method to measure EU,and thus EU includes both the transpiration of the understoryvegetation and the evaporation from soil. EO was composed of thetranspiration of the larch overstory Elarch, EU, and the change inlatent heat storage below the height of 32.0 m Jl. Hence,EO = Elarch + EU + Jl/l, where l is the latent heat of vapourisationof water.
2.2.2. Environmental variables
Table 1 lists the sensors used in this study. We measuredupward and downward shortwave radiation Su and Sd, upward anddownward longwave radiation Lu and Ld, downward photosyn-thetic photon flux density PPFD, air temperature Ta, and relativehumidity RH on the observation tower. These variables weredistinguished using subscripts of ‘O’ and ‘U’ for above the overstoryand above the understory, respectively. For example, the upwardshortwave radiation observed above the overstory was designatedas Su-O. Net radiation above the overstory (Rn-O) was calculated asthe balance of four radiation components Rn-O = (Sd-O � Su-O) + (Ld-
O � Lu-O), and net radiation above the understory vegetation Rn-U
was observed with a radiometer. We measured the volumetric soilwater content VSWC at five depths. Because the root system rarelyextended below 0.5 m in depth (Ohta et al., 2008), the weightedmean of the VSWC was calculated for depths of 0–0.5 m (VSWC0–
50). Soil temperature Ts was measured at seven depths. The soil
Table 1List of sensors used in this study.
Observed items Height (m) Sensor type
Downward shortwave radiation above the canopy (Sd-O) 32.0 MS-402F; Eiko, Japan
Downward longwave radiation above the canopy (Ld-O) 32.0 MS-202F; Eiko, Japan
Upward shortwave radiation above the canopy (Su-O) 31.4 CM-6F; Kipp & Zonen, The Netherlands
Uownward longwave radiation above the canopy (Lu-O) 31.4 MS-201F; Eiko, Japan
Downward photosynthetic photon flux density above the canopy (PPFDO) 27.9 LI-190; LI-COR, USA
Downward photosynthetic photon flux density above
the understory vegetation (PPFDU)
1.8 LI-190; LI-COR, USA
Net all-wave radiation above the understory vegetation (Rn-U) 1.2 Q-7; REBS, USA
Air temperature (Ta) and relative humidity (RH) 31.4, 24.6, 5.7, 1.8 HMP-45D; Vaisala, Finland
Wind speed 32.0, 27.0, 24.8, 14.9, 5.7, 1.9 AC-750; Makino, Japan
Sensible heat flux above the canopy (HO) 32.0 R3-50; Gill, UK
Sensible heat flux above the understory vegetation (HU) 3.3 R3-50; Gill, UK
Latent heat flux above the canopy (EO) 32.0 LI-7500; LI-COR, USA
Latent heat flux above the understory vegetation (EU) 3.3 LI-7500; LI-COR, USA
Soil heat flux (G) �0.01 PHF-01; Hukseflux, The Netherlands
Soil temperature 0, �0.1, �0.2, �0.4, �0.6, �0.8, �1.2 TS101; Hakusan Kougyo, Japan
Volumetoric soil water content (VSWC) �0.1, �0.2, �0.4, �0.6, �0.8 TRIME-FM2/P2; IMKO, Germany
Fig. 1. (A) Power spectra of vertical wind velocity w, temperature T, and humidity q
above the understory vegetation. The spectral density on the y-axis was multiplied
by the natural frequency n and normalised by the variance. The natural frequency
on the x-axis was divided by the wind speed. (B) Co-spectra for the covariances
between w and T, and w and q, above the understory vegetation. The spectral
density on the y-axis was multiplied by n and normalised by the covariance. The
data were obtained from 11:30 to 12:00 on 23 July 2006.
S. Iida et al. / Agricultural and Forest Meteorology 149 (2009) 1129–1139 1131
heat flux G was observed at a depth of 0.01 m. These variables wererecorded by data loggers every 5 min. Gross rainfall P wasmeasured at an open site located approximately 1 km south ofthe tower using a tipping-bucket rain gauge with 0.13-mmaccuracy.
We observed Rn-U and PPFDU at one station under the overstorycanopy. Because Sd-U has high-spatial variability, we measured Sd-U
at 42 points within a circle of 50-m diameter from the tower usinga calibrated silicon photodiode (Hamamatsu Photonics K.K., Japan,type S2387-66R). In total, we conducted observations of the spatialvariability nine times during the 2005 growing season. Onephotodiode was fixed at the top of tower and measured shortwaveradiation above the overstory (Sd-O-Sfix), and another photodiodewas fixed above the forest floor very near the station for Rn-U andPPFDU and measured shortwave radiation (Sd-U-Sfix). The finalphotodiode was moved above the forest floor to measureshortwave radiation at the 42 points (Sd-U-Smove). The fixedphotodiode measured a mean value of relative shortwave radiation(Sd-U-Sfix/Sd-O-Sfix) of 0.426, whereas the moving photodiodemeasured a mean value (Sd-U-Smove/Sd-O-Sfix) of 0.418. To evaluatethe estimation error of the area mean Sd-U derived from its spatialvariation, we used a statistical equation that has been widelyapplied to estimate the observation error for throughfall (e.g., Iidaet al., 2005). The maximum observation error in the ninemeasurement times for Sd-U-Smove/Sd-O-Sfix was 0.086, and theminimum value was 0.042. Thus, there was no significantdifference between the mean values of Sd-U-Sfix/Sd-O-Sfix and Sd-U-
Smove/Sd-O-Sfix, and the station could be considered reasonablyrepresentative for Sd-U. On the other hand, the net radiationincludes longwave radiation. In general, Ld-U is large under denseoverstory canopy, whereas Sd-U is small, producing a counter-balancing relationship between Ld-U and Sd-U (e.g., Hashimoto et al.,1997). Moreover, radiation that is time averaged over longerperiods is less effected by spatial variation (Hashimoto et al., 1997).In this study we used a mid-daytime (10:00 to 14:00 h) or dailytime scale for the analysis; therefore spatial variations of radiationwould not be significant for our analysis.
2.2.3. Turbulent fluxes
We measured sensible H and latent heat fluxes lE above boththe overstory and understory vegetation using the EC method witha three-dimensional ultrasonic anemometer and open-pathinfrared gas analyser (Table 1). To avoid damage to theinstruments, the EC system ran only during the warm periodfrom mid-April to the end of October. The three components ofwind speed, air temperature, and water vapour concentration were
sampled at a rate of 10 Hz using a data logger. Because the ECmethod was originally developed to measure mass, energy, andmomentum exchanges in the atmospheric surface layer, there wasa possibility that this method could fail to detect true fluxes withinforest canopies (Baldocchi et al., 2000a). To confirm the accuracy ofthe EC method above understory vegetation, we calculatedturbulence spectra for a few typical 30-min datasets. The spectraof vertical wind speed w, temperature T, and water vapour densityq were free of noise, the inertial subrange slopes were nearly equal
Fig. 2. Relationships between the observed and estimated latent heat fluxes lE of
(A) the whole ecosystem and (B) the understory vegetation in 2006. Solid symbols
show daytime (8:00 to 18:00 h) data, and open symbols show nighttime (18:00 to
8:00 h) data.
S. Iida et al. / Agricultural and Forest Meteorology 149 (2009) 1129–11391132
to the classical value of �2/3, and the inertial subrange slopes ofthe w–T and w–q co-spectra were close to �4/3 (Fig. 1). Thus weconcluded that the EC method was able to detect most of theturbulent eddies above the understory vegetation. To obtainreliable turbulent fluxes at 30-min intervals, we used a three-dimensional coordinate rotation to make the lateral and verticalwind speeds zero (McMillen, 1988; Kaimal and Finnigan, 1994).We also corrected for the effects of density fluctuation (Webb et al.,1980) and for the error caused by the ultrasonic anemometer angleof attack (Nakai et al., 2006).
2.3. Data processing
2.3.1. Estimation of heat storage in the air mass, soil layer, and
overstory trees
We estimated the changes in heat storage in the air mass belowthe EC system but above the understory Ja, in a soil layer betweenthe surface and a buried heat flux plate Js, and in overstory trees Jwd.All of the heat storage changes were calculated by the samemethod as used by Ohta et al. (2008). Ja was the sum of the sensibleheat storage Jh and the latent heat storage Jl. The values of Jh and Jl
were calculated from the unit-time change in the vertical profilesof air temperature and mixing ratio of water vapour in dry air,respectively. Js was estimated using the heat storage capacities ofsoil and water and the representative temperature of the soil layerbetween the surface and the heat flux plate. Jwd was estimatedwithout wood temperature data using a simple model thatincluded the biomass volume and air temperature just abovethe canopy (Tanaka et al., 2008). The equations and parametersused for calculation of all of these quantities have already beendescribed in detail by Ohta et al. (2008).
2.3.2. Quality control, energy closure, and gap-filling for
turbulent fluxes
The turbulent fluxes calculated during and just after rainfallwere not used because the open-path infrared gas analyser couldhave been affected by wet conditions. Toba and Ohta (2002)observed the rainfall interception processes at this site andreported that forest canopies dried within 10 h after rainfallceased. Therefore, we removed the fluxes during rain events andfor 10 h after rainfall stopped. To delete other errors, we usedquadratic regression equations between net radiation, and sensibleand latent heat fluxes, which were determined from half-hourlydata. The regression equations for the above-overstory data werecalculated for each month from May to September. However, therewere large differences in the radiation that reached the understoryvegetation during the leafless and foliated periods of the overstorydeciduous larch. Therefore, we performed regression analyses forthe understory dataset over three periods: May, June–August, andSeptember. We obtained the differences between the observed andpredicted fluxes using the regression equations. Outliers, whichwere defined as values greater than twice the standard deviation,were excluded from the dataset.
After all these quality-control measures, the remaining fluxdata were used to calculate the energy balances above theoverstory and understory using the following equations:
HO þ lEO þ Ja þ Jwd þ Js ¼ aO � ðRn-O � GÞ þ bO; (1)
HU þ lEU þ Js ¼ aU � ðRn-U � GÞ þ bU: (2)
The slope aO was 0.94 (2005) and 0.96 (2006), and aU was 0.80(2005) and 0.83 (2006). The y-intercept bO was�4.3 W m�2 (2005)and �1.6 W m�2 (2006), and bU was 3.8 W m�2 (2005) and6.0 W m�2 (2006). The coefficients of determination (R2) ofEq. (1) were 0.94 (2005) and 0.93 (2006), and those of Eq. (2)were 0.78 (2005) and 0.73 (2006). These values of a, b, and R2 were
similar to the results of Wilson et al. (2002), implying that thedataset we used was reasonable.
To estimate the accumulated evapotranspiration over certainperiods (e.g., daily, monthly) we had to fill gaps in the dataset. Weapplied the multiple-imputation (MI) method to fill the gaps (Huiet al., 2004), using the ‘‘NORM’’ software (http://www.stat.p-su.edu/�jls/misoftwa.html). We used net radiation, downwardshortwave radiation, wind speed, air temperature, and vapourpressure deficit to impute the latent and sensible heat fluxes. Wefilled the gaps separately for each month of the overstory andunderstory datasets. For each month, 2000 iterations of imputationwere performed, and 10 imputed datasets were picked up at every100-iteration interval from the last 1000 imputations (i.e., iterationnumbers 1100, 1200, . . ., 2000). We calculated the ensemble meanof the 10 imputed datasets and replaced the gap with the mean. Toevaluate the accuracy of gap-filling using this method, additionalrandom gaps of 10% were artificially generated from the originaldata and filled using the MI method. The observed lE and thatestimated using the MI method showed a close linear relationship
S. Iida et al. / Agricultural and Forest Meteorology 149 (2009) 1129–1139 1133
(Fig. 2), with daytime (from 8:00 to 18:00) root mean square errors(RMSEs) between observed and imputed lEs of 37.4 W m�2 (2005)and 33.2 W m�2 (2006) above the overstory canopy and20.0 W m�2 (2005) and 22.0 W m�2 (2006) above the understoryvegetation. The daytime relative RMSE (R_RMSE), calculated as theRMSE divided by the average value of observed lE, was 0.45 (2005)and 0.40 (2006) above the overstory canopy, and 0.41 (2005) and0.43 (2006) above the understory vegetation. Because the obtainedR_RMSEs were similar to the values reported by Alavi et al. (2006),we concluded that the MI method filled the gaps in the fluxes withreasonable accuracy.
2.3.3. Calculation of surface conductance, aerodynamic conductance,
and decoupling factor
We calculated surface conductance gs using the invertedPenman–Monteith equation (Dolman et al., 1991):
g�1s ¼ D
gb� 1
� �g�1
a þrC p
glEVPD; (3)
where D is the rate of change in saturation water vapour pressurewith temperature, g is the psychrometric constant, b is the Bowenratio, ga is the aerodynamic conductance, r is the density of moistair, Cp is the specific heat of air at constant pressure, and VPD is thevapour pressure deficit. The ga was estimated from the measuredfriction velocity u* as
ga ¼u2�
u; (4)
where u is the wind speed.McNaughton and Jarvis (1983) reformulated the Penman–
Monteith equation as
lE ¼V � lEeq þ ð1�VÞ � lEimp; (5)
where Eeq is the equilibrium evaporation, Eimp is the imposedevapotranspiration, and V is the decoupling factor, given as
Eimp ¼rC p
lg� gs � VPD; (6)
Eeq ¼D � AE
lðDþ gÞ ; (7)
V ¼ D=g þ 1
D=g þ 1þ ga=gs
: (8)
In the above, AE is the available energy (above overstory,AEO = Rn-O� G � Ja � Jwd� Js, and above understory, AEU = Rn-
U � G� Js). TheV term indicates the relative importance of radiative
Fig. 3. Seasonal changes in (A) mid-daytime (10:00 to 14:00 h) mean albedo observed ab
photosynthetic photon flux density observed above the overstory canopy, PPFDO, to tha
effect of the direct component of PPFDO, only cases of PPFDO less than 1000 mmol m�2
energy, lEeq, and advective energy, lEimp, for lE. In completelycoupled conditions (V = 0), exchange between the surface and theair occurs smoothly. Then, lE is affected only by lEimp and hence iscontrolled by the surface conductance. In contrast, the plantphysiological control becomes weaker as V approaches 1. Wecalculated these parameters for both the whole ecosystem and theunderstory vegetation using the overstory and forest floor datasets,respectively. Note that all of the parameters described in thissubsection were calculated from the quality-controlled datasets, inwhich the gaps had not been filled using the MI method.
3. Results and discussion
3.1. Seasonal changes in the phenology of the larch overstory and
environmental variables
The snow-cover condition was estimated using the mid-daytime (10:00 to 14:00 h) mean observed albedo of the overstorycanopy (Fig. 3(A)). The albedo decreased during April, indicatingthe melting of snow cover. After April, albedo was relativelyconstant until the end of September. In both 2005 and 2006, thesnow cover melted completely on 4 May. This date was validatedby daytime mean soil temperature measurements at 0.1 m depth:the soil temperature rose above 0 8C after the snow cover hadmelted. Albedo greatly increased in the middle of October, whensnow again began to cover the forest. The leaf phenology of thelarch overstory was evaluated using the ratio of the mid-daytimemean of the downward photosynthetic photon flux density abovethe overstory, PPFDO, to that above the understory, PPFDU. ThePPFDU/PPFDO ratio decreased from May to June (Fig. 3(B)),indicating the larch leaf flush. The completion dates of the leafflush were different in 2005 and 2006: in 2006, the completiondate was near the end of June, approximately 20 days later thanthat in 2005. After early September, the PPFDU/PPFDO ratio began toincrease, indicating the larch leaf fall. The phenology was alsoconfirmed by a digital camera fixed on the tower, and was verysimilar to our previous result (Ohta et al., 2001). Based on thesnow-cover conditions and the overstory larch phenology, wedetermined the growing season to be from 4 May to 31 Septemberin both 2005 and 2006, almost corresponding with the snow-freeperiod. The snow-free period was suitable for our primaryobjective of measuring EU by the EC method. After the snowfallperiod begins, the low-lying understory vegetation can easilybecome covered by snow, and both its transpiration and theevaporation from soil decreases significantly. The leaf-foliatedperiods of the larch were approximately from the beginning of Juneto the end of August in both years.
ove the overstory canopy and (B) the ratio of the mid-daytime mean of downward
t observed above the understory vegetation, PPFDU, in 2005 and 2006. To avoid the
s�1 are plotted in (B).
Fig. 4. Seasonal variations in the daily mean value of (A) net radiation Rn, (B) air temperature Ta, (C) vapour pressure deficit VPD, (D) daily rainfall amount P, and (E) the daily
mean volumetric soil water content VSWC0–50 in 2005 and 2006. For Rn, Ta, and VPD, the overstory canopy data are shown by the open circles, and the understory vegetation
data are shown by the solid lines.
Table 2Evapotranspiration from the whole ecosystem EO, the understory vegetation EU, and
the larch, Elarch = EO � EU. Also, the contribution of EU to EO for each month, for the
foliated period from June to August (JJA), and for the whole growing seasons in 2005
and 2006.
Month EO (mm) EU (mm) Elarch (mm)a EU/EO (%)
2005
May 20.6 15.4 5.2 74.9
June 57.6 29.3 28.2 50.9
July 52.4 27.6 24.7 52.8
August 35.4 17.8 17.7 50.2
September 15.9 9.2 6.7 58.0
JJA 145.4 74.7 70.6 51.4
Total 181.9 99.3 82.5 54.6
2006
May 15.3 16.2 – 105.6
June 54.9 31.8 23.1 57.9
July 63.3 32.0 31.2 50.6
August 37.3 16.8 20.5 45.0
September 9.4 5.7 3.6 61.1
JJA 155.5 80.6 74.8 51.8
Total 180.2 102.5 78.4 56.9
a Elarch = EO � EU.
S. Iida et al. / Agricultural and Forest Meteorology 149 (2009) 1129–11391134
The daily mean net radiation above the understory vegetation,Rn-U, was significantly smaller than that above the overstorycanopy, Rn-O, because the overstory shaded the understory(Fig. 4(A)). However, there were relatively small differencesbetween the daily mean vapour pressure deficit above theunderstory, VPDU, and that above the canopy, VPDO (Fig. 4(C)).This phenomenon has also been observed in a temperate forest(Wilson et al., 2000) and a boreal forest (Blanken et al., 1997). Thedaily mean air temperature above the overstory Ta-O exceeded15 8C at the beginning of May 2005, more than 5 8C higher than atthe same date in 2006. The warmer Ta-O in the spring of 2005caused the leaf flush to finish earlier (Fig. 3(B)). We observedrelatively large amounts of rainfall P during the growing season in2005. In contrast, the values of P were highest in August andSeptember in 2006 (Fig. 4(D)). Ohta et al. (2008) argued that thevolumetric soil water content VSWC in this forest significantlydepended on the amount of rainfall in the previous summer. As P
was much smaller in the summer of 2004 (Ohta et al., 2008), theVSWC0–50 during June and July in 2005 was smaller than that in2006 (Fig. 4(E)).
3.2. Seasonal variation in daily values of EO, EU, and EU/EO
The daily values of EU were almost equal to those of EO at thebeginning of May in 2005 and 2006. Towards the end of May, wefrequently observed daily values of EO that were larger than thoseof EU (Fig. 5(A)). These trends indicated that the transpiration of thelarch overstory, Elarch, gradually increased, and thus the daily EU/EO
ratio decreased during the leaf flush period (Fig. 5(B)). Because the
monthly mean Jl value was less than 7% of EO, we could evaluateElarch as the difference between EO and EU on a monthly basis:Elarch = EO � EU (Table 2). In May, monthly values of Elarch weremuch smaller than those of EU because larch leaf flush had justbegun. Especially in May 2006, when leaf flush was completed
Fig. 5. Seasonal changes in (A) daily amounts of evapotranspiration from the whole ecosystem EO and from the understory vegetation EU, and (B) the contribution of EU to EO in
2005 and 2006.
S. Iida et al. / Agricultural and Forest Meteorology 149 (2009) 1129–1139 1135
much later than in 2005, Elarch was too small to measure (Table 2).EO and EU increased at the beginning of June, and EU showed aparallel seasonal trend to EO throughout the foliated period (Figs. 5and 6). The reasons for this trend are discussed in Section 3.4.Because of this, the EU/EO ratio was relatively constant during thefoliated period (Fig. 5(B)). However, the EU/EO ratio increasedduring September because the larch overstory leaves fell, causingElarch to decrease faster than EU (Fig. 5(B)). The difference in theseasonal trend of EO and EU, especially the difference in thetranspiration activity between deciduous overstory trees andevergreen understory vegetation, resulted in a seasonal variationof the EU/EO ratio that had a flat-bottomed U-shape.
During the foliated period (1 June–31 August), the daily meanEO was 1.6 and 1.7 mm day�1 in 2005 and 2006, respectively(Table 3). Previous studies in eastern or central Siberia havereported daily EO values of 2.0 mm day�1 in larch forest (Kelliheret al., 1997), 1.5 mm day�1 (Ohta et al., 2001; Dolman et al., 2004)in the same forest as that examined here, and 1.6 and2.2 mm day�1 in pine forests (Tchebakova et al., 2002; Hamadaet al., 2004). Our results agree closely with the relatively long-termmeasurements of Ohta et al. (2001) and Dolman et al. (2004). The
Fig. 6. Relationship between the daily evapotranspiration amounts from the whole
ecosystem EO and from the understory vegetation EU. Data collected during June,
July, and August are indicated by solid symbols, and data collected during May and
September are indicated by open symbols. The regression lines were calculated for
the data during the foliation period (June, July and August).
daily mean EO values in July, calculated from Table 2, were 1.7 and2.0 mm day�1 in 2005 and 2006, respectively. These values aresimilar to the value obtained from the measurements in larchforest during July 1993 by Kelliher et al. (1997).
The daily mean EU values during the foliated period were 0.8and 0.9 mm day�1 in 2005 and 2006, respectively (Table 3). Thesevalues are comparable with previous relatively long-term mea-surements in larch forest (0.5 mm day�1; Ohta et al., 2001) and inpine forest (0.8 mm day�1; Hamada et al., 2004). The daily mean EU
value in July was 0.9 and 1.0 mm day�1 in 2005 and 2006,respectively, similar to the 0.9 mm day�1 reported by Kelliher et al.(1997). Lopez et al. (2007) estimated Elarch based on sap fluxobservations in the same forest examined here and reported thatthe daily mean Elarch values from 7 to 31 July were 1.3 and0.9 mm day�1 in 2003 and 2004, respectively. We calculated themonthly mean daily amount of Elarch in July as 0.8 and1.0 mm day�1 in 2005 and 2006, respectively, from Table 2. Thisclose agreement with the results of Lopez et al. (2007) suggeststhat our observations of EO and EU were reasonable.
3.3. Contribution of EU to EO during foliated periods and
entire growing seasons
During the foliated period, the EU/EO ratio was 51.4% in 2005and 51.8% in 2006 (Table 3). Over the entire growing season, thecontribution was 54.6% in 2005 and 56.9% in 2006 (Table 2). Thiswas slightly larger than the foliated period contribution because itincluded the leaf flushing and falling periods of larch. In this forest,EU was slightly larger than Elarch throughout the growing season,even during the foliated period. Our measurements of EU/EO werethe first to use the EC method over entire growing seasons in aSiberian boreal forest. Although the long-term values of EU/EO thatwe calculated could not be compared with other previous studiesconducted in Siberia, they are probably reasonable because eachabsolute value of EO and EU was validated against values reportedin previous papers as discussed above.
The contributions of understory vegetation mentioned abovewere taken from dry canopy datasets and thus did not include theamount of rainfall interception loss, I, because we used the ECmethod to measure EO and EU. Ohta et al. (2008) estimated I usingan interception model developed in the same forest as thatexamined here (Toba and Ohta, 2002, 2005). They found I values of48.9 and 50.5 mm over the entire growing season in 2005 and2006, respectively. The contributions of understory vegetation tothe whole-ecosystem evapotranspiration from both dry and wetcanopies (EU/[EO + I]) were 43.0 and 43.2% in 2005 and 2006,respectively. Therefore, at this site, the contribution of the
Table 3Comparison of evapotranspiration from the whole ecosystem EO, with that from the understory vegetation, EU, and the contribution of EU to EO observed in boreal forests.
Cited paper Overstory Understory Period EO
(mm day�1)
EU
(mm day�1)
EU/EO
(%)
LAI of
overstory
Observation method
Siberian forests
This study Larix cajanderi Vaccinium vitis-idaea 1 June–31
August 2005
1.6 0.8 51.4 2.0 EO: eddy-covariance
method (EC), EU: EC
This study 1 June–31
August 2006
1.7 0.9 51.8
Kelliher et al. (1997) L. cajanderi Vaccinium and
Arctostaphylos spp.
15, 21, 22 and
23 July 1993
2.0a 0.9a 45.0a 1.5 EO: EC, EU: lysimeters
Ohta et al. (2001) Same forest as this study 1 June–31
August 1998
1.5 0.5b 35.0 EO: EC, EU: difference
between EO and
transpiration of larch
estimated by sap flux
measurements
Dolman et al. (2004) Same forest as this study Growing seasons
in 2000 and 2001c
1.5 – – EO: EC, EU: not
measured
Kelliher et al. (1998) Pinus sylvestris Cladonia and Cladina spp. 8–10, 12–18,
20–25 July 1996
1.3d 0.8d 61.5d 1.5 EO: EC, EU: EC and
lysimeter
Tchebakova et al. (2002) P. sylvestris Cladonia and Cladina spp. 1 June–31
August 1999
and 2000
1.6e – – 1.5 EO: EC, EU: not
measured
Hamada et al. (2004) P. sylvestris Arctostaphylos uva-ursi 1 May–31
August 2000
2.2 0.8b 25–50 2.5 EO: EC, EU: lysimeters
Canadian and Swedish forests
Baldocchi et al. (1997) Pinus banksiana Ahus crispa, A. uva-ursi,
V. vitis-idaea, Cladina spp.
1 June–31
August 1994
1.5f 0.3f 20.0f 1.9–2.2 EO: EC, EU: EC
Black et al. (1996) Populus tremuloides Corylus cornuta Marsh.
and Alnus crispa (Ait.)
Pursch
1 June–31
August 1994
2.7g 0.7g 25.9g 3.3 EO: EC, EU: EC
Constantin et al. (1999) P. sylvestris and
Picea abies
Vaccinium myrtillus 4–23 June
1994 (missing
data on 5 and
18 June)
2.6h 0.3h 11.5h 3.0–4.0 EO: EC, EU: EC
a Calculated from Table 3 in Kelliher et al. (1997).b These values were not described in original articles, and calculated as EO�EU/EO. For Hamada et al. (2004), the mean value of EU/EO was used for the calculation.c Detailed information was not provided in the original paper, but the growing season should be mainly from May to September.d Calculated from Tables 1 and 2 in Kelliher et al. (1998).e Calculated from Table 3 in Tchebakova et al. (2002).f Digitized and calculated from Figs. 8 and 9 in Baldocchi et al. (1997).g Digitized and calculated from Fig. 6 in Black et al. (1996).h Digitized and calculated from Fig. 6 in Constantin et al. (1999).
Table 4Mean values of aerodynamic conductance ga, surface conductance gs, and
decoupling factor V for the whole growing seasons in 2005 and 2006.
Parameters Unit Year
2005 2006
Whole ecosystem
ga-O m s�1 0.129 (�0.066) 0.130 (�0.066)
gs-O m s�1 0.0043 (�0.0026) 0.0046 (�0.0027)
VO – 0.093 (�0.068) 0.102 (�0.075)
Understory vegetation
ga-U m s�1 0.038 (�0.033) 0.043 (�0.028)
gs-U m s�1 0.0031 (�0.0020) 0.0030 (�0.0016)
VU – 0.212 (�0.138) 0.194 (�0.110)
Mean (�standard deviation) values for the daytime (8:00 to 18:00 h) for the whole
growing season from 4 May to 30 September.
S. Iida et al. / Agricultural and Forest Meteorology 149 (2009) 1129–11391136
understory vegetation was very significant. It is likely thatunderstory vegetation cannot be neglected if the hydrologic cyclein Siberian boreal forests is to be modelled accurately.
3.4. The factors controlling EU and EO
In this section, we discuss the energy sources for EU and EO.Evapotranspiration can be expressed as the sum of a radiation termV�lEeq, and aerodynamic term (1 �V)�lEimp, and therefore can bedetermined by V, lEeq, and lEimp (Eq. (5)). The mean values of lEeq
and lEimp above the overstory (lEeq-O and lEimp-O) shown inFig. 7(A) were 152 and 80 W m�2, respectively, and those above theunderstory (lEeq-U and lEimp-U) were 51 and 50 W m�2, respec-tively. There was a large difference (72 W m�2) between lEeq-O andlEimp-O, but lEeq-U and lEimp-U were almost equal. This differencebetween the overstory and the understory was observed becauseof a much smaller Rn-U than Rn-O (Fig. 4(A)) and a smaller differencebetween VPDU and VPDO (Fig. 4(C)). lEU had a linear relationshipwith both lEeq-U and lEimp-U, but we found a higher determinationcoefficient between lEU and lEimp-U (�1) than between lEU andlEeq-U (Fig. 7(B)). A similar relationship existed between lEO andlEimp-O (Fig. 7(A)). The high-determination coefficients betweenlEU and lEimp-U and between lEO and lEimp-O imply that lEU andlEO were strongly affected by lEimp-U and lEimp-O, respectively.The decoupling factors at both heights may have been small. Infact, the decoupling factor calculated above the understory, VU,was small: 0.212 in 2005 and 0.194 in 2006 (Table 4). The wholeecosystem value, VO, was also small: 0.093 in 2005 and 0.102 in
2006. The coupling was so high because gs was significantlysmaller than ga in both cases (Table 4).
To evaluate how lEU was controlled by the radiation and thevapour pressure deficit, we calculated the contributions of theradiation term VU�lEeq-U and the aerodynamic term(1 �VU)�lEimp-U to lEU. The slope and the coefficient ofdetermination of the linear regression between lEU and(1 �VU)�lEimp-U were clearly larger than those between lEU
and VU�lEeq-U (Fig. 8(B)). Because the y-intercept value was almostzero, the contribution of (1 �VU)�lEimp-U to lEU was almostequivalent to the slope of the regression equation, i.e., 71%. In otherwords, lEU was mainly determined by (1 �VU)�lEimp-U. We
Fig. 7. Comparison between the latent heat flux lE and the equilibrium evaporation
lEeq, and between lE and the imposed evaporation lEimp above (A) the overstory
and (B) the understory. We plotted all data obtained in 2005 and 2006. Fig. 8. Relationships between the latent heat flux lE and the radiation term of the
Penman–Monteith equation V�lEeq, and between lE and the aerodynamic term
(1 �V)�lEimp above (A) the overstory and (B) the understory vegetation. We
plotted all data obtained in 2005 and 2006.
S. Iida et al. / Agricultural and Forest Meteorology 149 (2009) 1129–1139 1137
obtained similar results for the overstory dataset, including acontribution of (1 �VO)�lEimp-O to lEO of 81% (Fig. 8(A)). Thusboth lEU and lEO were driven mainly by VPD rather than AE, andwere dominantly controlled by the plant physiological factors.Linear relationships were found between the daily mean VU�lEeq-U
and VO�lEeq-O, and between (1 �VU)�lEimp-U and (1 �VO)�lEimp-
O (Fig. 9). These relationships resulted from the linear relationshipsbetween the daily mean values of Rn-U and Rn-O, with R2 = 0.91, andbetween VPDU and VPDO, with R2 of 0.96, which were also thereason for the linear relationship between EU and EO during thefoliated period (Fig. 6). As mentioned above, we measured Rn-U atone station above the forest floor. To validate the accuracy ofVU�lEeq-U calculated by the measured Rn-U, we obtained VU�lEeq-U
as the difference between lEU and (1 �VU)�lEimp-U (Eq. (5)).Calculations of VU and lEimp-U do not include Rn-U (Eqs. (3), (4), (6)and (8)), and thus we can obtain VU�lEeq-U as the differenceindependent of Rn-U. The obtained values corresponded very wellwith the original values on a daily scale, (lEU � [1 �VU]�lEimp-
U) = 1.0VU�lEeq-U + 0.6, with R2 = 0.93. Therefore, the measured Rn-
U was considered reasonable for our analysis.
The absolute value of EU and the EU/EO ratio are likely to belarger in Siberian forest than in Swedish and Canadian forests(Table 3). In all the three forest types, VPDU is the main factorcontrolling EU (Baldocchi et al., 1997; Black et al., 1996; Constantinet al., 1999). Thus the larger EU and EU/EO ratio in Siberian forestmay be caused by a larger gs-U and/or VPDU. Unfortunately,previous reports of gs-U and VPDU have been too limited to comparewith our results. However, Blanken et al. (1997) reported that inthe same Canadian aspen forest as that investigated by Black et al.(1996), the mean canopy conductance value of the understoryvegetation during the foliated period was 0.0029 m s�1, and themaximum value of half-hourly VPDU was as high as 25 hPa. Thesevalues are similar to our gs-U (Table 4) and half-hourly VPDU results.Hence, it is reasonable that the mean daily EU value for the foliatedperiod in Canadian aspen forest is nearly the same as our result(Table 3). We could not compare our gs-U and VPDU results withthose for Canadian jack pine forest (Baldocchi et al., 1997) andSwedish spruce/pine forest (Constantin et al., 1999) because these
Fig. 9. Comparison between (A) the radiation term V�lEeq above the overstory and
the understory and (B) the aerodynamic term (1 �V)�lEimp above the overstory
and the understory. The closed circles show the data during the foliation period
(June, July, and August), and the open circles show the data during the leaf flushing
and falling periods (May and September) in 2005 and 2006. The regression lines
were calculated from the dataset collected during the foliation period.
S. Iida et al. / Agricultural and Forest Meteorology 149 (2009) 1129–11391138
values were not recorded there. However, the daily mean VPDO
observed at our site was larger than that reported by Baldocchiet al. (1997). This implies that the VPDU at our site was larger thanthat in Canadian jack pine forest. It is possible that larger VPDU wasone of reasons why EU was larger at our site than in Canadian jackpine forest. However, more observations of EU and micrometeor-ology above the forest floor are needed to validate the difference inEU and the EU/EO ratio between Siberian and other boreal forests,and to clarify the reason for the difference.
Overstory trees in boreal forests generally have small LAI,constricting the maximum value of gs (Kelliher et al., 1995), andsparse density, which enhances ga. These forest structures,therefore, likely have small V, indicating the importance of plantphysiological control on the evapotranspiration process. However,tropical forests have high VO, and therefore high-lEO values thatnearly correspond to lEeq-O (e.g., Kumagai et al., 2004, 2005). In the
forest we studied, lEeq-O was much larger than lEimp-O (Fig. 7(A)).If the LAI increased and VO tended towards 1.0, lEO wouldincrease. Therefore, the significantly low-lEO value compared withvalues for warmer climate regions (e.g., Hamada et al., 2004) iscaused by low-VO values in boreal forests. On the other hand, alarger overstory LAI decreases ga significantly, thereby increasingnot only VO but also VU. It also decreases lEeq-U, due to largerabsorption of radiation by leaves, and finally results in smaller EU.Therefore, overstory LAI is a key factor affecting the EU/EO ratio,with larger LAI resulting in a smaller ratio (Table 3). Our suggestioncorresponds well with the relationship demonstrated by Daikokuet al. (2008), whereby the EU/EO ratio decreased exponentially withincrease in the LAI.
4. Conclusions
We measured the evapotranspiration in a Siberian deciduouslarch forest from the entire ecosystem above the overstory canopyEO and from the evergreen understory vegetation above the forestfloor EU using the eddy-covariance method. We conductedobservations for two complete seasons (2005 and 2006) andobtained the following three conclusions for the objectives given inthe introduction:
(i) T
he mean daily values of EO during the foliated period (1 June–31 August) were 1.6 (2005) and 1.7 mm day�1 (2006), whilethose of EU were 0.8 (2005) and 0.9 mm day�1 (2006). EU wasnearly equal to EO at the beginning of the larch leaf flush periodand at the end of the larch leaf falling period. During thefoliated period, daily EU values varied linearly with EO. Thusthe daily contribution of EU to EO (EU/EO) had a flat-bottomed,U-shaped seasonal trend.(ii) T
he contributions of EU to EO were 51.4% (2005) and 51.8%(2006) during the foliated period. During the growing season(4 May–30 September), the contributions were 54.6% (2005)and 56.9% (2006). Including the rainfall interception loss Icalculated by the model, the contributions of EU to EO + I were43.0% (2005) and 43.2% (2006).
(iii) B
ecause of the small LAI and sparse density of the overstorytrees, the aerodynamic conductance was much larger than thesurface conductance for both the whole ecosystem and theunderstory vegetation. As a result, their decoupling coeffi-cients V were small. The contribution of the aerodynamicterm to EU at the understory (1 �VU)�lEimp-U/lEU was 71%,and that to EO at the overstory (1 �VO)�lEimp-O/lEO was 81%.Therefore, both EU and EO were controlled by plant physio-logical factors and were determined by the energy resultingfrom the vapour pressure deficit, not by the available energy.The small EO values in boreal ecosystems, compared to those inforests in warmer climate regions, are due to the small VO values.The structure of boreal forests, especially small LAIs and sparse treedensities, enhances plant physiological control of evapotranspira-tion and the contribution of EU to EO.
Acknowledgements
This study was part of the Water and Energy Cycles in NorthernForests (WECNoF) framework (PI: Prof. Takeshi Ohta, NagoyaUniversity, Japan), which is supported by the Core Research forEvolutional Science and Technology (CREST) program of the JapanScience and Technology Agency (JST). This study was alsosupported by the Programme International Nature ManagementCentral and Eastern Europe (PIN-MATRA) project (PIs: Prof. HanDolman, Amsterdam Free University, and Prof. Eddy J. Moors,Alterra, The Netherlands) and the Institute of Observational
S. Iida et al. / Agricultural and Forest Meteorology 149 (2009) 1129–1139 1139
Research for Global Change (IORGC) program (PI: Dr. Tetsuo Ohata,IORGC, Japan). We thank all of the members of these projects fortheir financial and operational support, as well as their friendlycooperation. We especially thank the members of the Institute forBiological Problems in the Cryolithozone, Siberia Division, RussianAcademy of Sciences (RAS), for their help in obtaining themeasurements and conducting the analyses. We thank Dr.Motomu Toda from the Institute of Low Temperature Science,Hokkaido University, Japan, for his helpful comments on thismanuscript. Finally, we would like to thank the editor and thereviewers for their many useful and constructive comments.
References
Alavi, N., Warland, J.S., Berg, A.A., 2006. Filling gaps in evapotranspiration measure-ments for water budget studies: evaluation of a Kalman filtering approach.Agric. For. Meteorol. 141, 57–66.
Baldocchi, D.D., Meyers, T.P., 1991. Trace gas exchange at the floor of a deciduousforest. I. Evaporation and CO2 efflux. J. Geophys. Res. Atmos. 96, 7271–7285.
Baldocchi, D.D., Vogel, C., 1996. A comparative study of water vapor, energy and CO2
flux densities above and below a temperate broadleaf and a boreal pine forest.Tree Physiol. 16, 5–16.
Baldocchi, D.D., Vogel, C., Hall, B., 1997. Seasonal variation of energy and watervapor exchanges rates above and below a boreal jack pine forest canopy. J.Geophys. Res. 102, 28939–28951.
Baldocchi, D.D., Law, B.E., Anthoni, P.M., 2000a. On measuring and modeling energyfluxes above the floor of a homogeneous and heterogeneous conifer forest.Agric. For. Meteorol. 102, 187–206.
Baldocchi, D.D., Kelliher, F.M., Black, T.A., Jarvis, P., 2000b. Climate and vegetationcontrols on boreal zone energy exchange. Global Change Biol. 6 (Suppl. 1), 69–83.
Black, T.A., Kelliher, F.M., 1989. Processes controlling understory evapotranspira-tion. Philos. Trans. R. Soc. Lond., Ser. B 324, 207–231.
Black, T.A., den Hartog, G., Neumann, H.H., Blanken, P.D., Yang, P.C., Russell, C., Nesic,Z., Lee, X., Chen, S.G., Staebler, R., Novak, M.D., 1996. Annual cycles of watervapour and carbon dioxide fluxes in and above a boreal aspen forest. GlobalChange Biol. 2, 219–229.
Blanken, P.D., Black, T.A., Yang, P.C., Neumann, H.H., Nesic, Z., Staebler, R., denHartog, G., Novak, M.D., Lee, X., 1997. Energy balance and canopy conductanceof a boreal aspen forest: partitioning overstory and understory components. J.Geophys. Res. 102, 28915–28927.
Blanken, P.D., Black, T.A., Neumann, H.H., den Hartog, G., Yang, P.C., Nesic, Z., Lee, X.,2001. The seasonal water and energy exchange above and within a boreal aspenforest. J. Hydrol. 245, 118–136.
Constantin, J., Grelle, A., Ibrom, A., Morgenstern, K., 1999. Flux partitioning betweenunderstorey and overstorey in a boreal spruce/pine forest determined by theeddy covariance method. Agric. For. Meteorol. 98–99, 629–643.
Daikoku, K., Hattori, S., Deguchi, A., Aoki, Y., Miyashita, M., Matsumoto, K., Akiyama,J., Iida, S., Toba, T., Fujita, Y., Ohta, T., 2008. Influence of evaporation from theforest floor on evapotranspiration from the dry canopy. Hydrol. Process 22,4083–4096.
Dolman, A.J., Gash, J.H.C., Roberts, J., Shuttleworth, W.J., 1991. Stomatal and surfaceconductance of tropical rainforest. Agric. For. Meteorol. 54, 303–318.
Dolman, A.J., Maximov, T.C., Moors, E.J., Maximov, A.P., Elbers, J.A., Kononov, A.V.,Waterloo, M.J., van der Molen, M.K., 2004. Net ecosystem exchange of carbondioxide and water of far-eastern Siberian larch (Larix cajanderii) on permafrost.Biogeosciences 1, 133–146.
FAO, 2005. State of the World’s Forests 2005. Food and Agriculture Organization ofthe United Nations, Rome, 166 p.
Hamada, S., Ohta, T., Hiyama, T., Kuwada, T., Takahashi, A., Maximov, T.C., 2004.Hydrometeorological behaviour of pine and larch forests in eastern Siberia.Hydrol. Process 18, 23–39.
Hashimoto, T., Ohta, T., Nakamura, T., 1997. Spatial representativeness of thedownward radiation and snowmelt rates in a forest based on the three dimen-sional forest structure (1) estimation method. J. Jpn. Soc. Hydrol. Water Resour.10, 524–536 (in Japanese with English summary).
Hui, D., Wan, S., Su, B., Katul, G., Monson, R., Luo, Y., 2004. Gap-filling missing data ineddy covariance measurements using multiple imputation (MI) for annualestimation. Agric. For. Meteorol. 121, 93–111.
Iida, S., Tanaka, T., Sugita, M., 2005. Change of interception process due to thesuccession from Japanese red pine to evergreen oak. J. Hydrol. 315, 154–166.
Kaimal, J.C., Finnigan, J.J., 1994. Boundary Layer Flows. Oxford University Press,Oxford, 289 pp.
Kelliher, F.M., Leuning, R., Raupach, M.R., Schulze, E.-D., 1995. Maximum con-ductances for evaporation from global vegetation types. Agric. For. Meteorol.73, 1–16.
Kelliher, F.M., Hollinger, D.Y., Schulze, E.-D., Vygodskaya, N.N., Byers, J.N., Hunt, J.E.,McSeveny, T.M., Milukova, I., Sogatchev, A., Varlagin, A., Ziegler, W., Arneth, A.,Bauer, B., 1997. Evaporation from an eastern Siberian larch forest. Agric. For.Meteorol. 85, 135–147.
Kelliher, F.M., Lloyd, J., Arneth, A., Byers, J.N., McSeveny, T.M., Milukova, I.,Grigoriev, S., Panfyorov, M., Sogatchev, A., Varlargin, A., Ziegler, W., Bauer,G., Schulze, E.-D., 1998. Evaporation from a central Siberian pine forest. J.Hydrol. 205, 279–296.
Kumagai, T., Saitoh, T.M., Sato, Y., Morooka, T., Manfroi, O.J., Kuraji, K., Suzuki, M.,2004. Transpiration, canopy conductance and the decoupling coefficient of alowland mixed dipterocarp forest in Sarawak, Borneo: dry spell effects. J.Hydrol. 287, 237–251.
Kumagai, T., Saitoh, T.M., Sato, Y., Takahashi, H., Manfroi, O.J., Morooka, T., Kuraji, K.,Suzuki, M., Yasunari, T., Komatsu, H., 2005. Annual water balance and season-ality of evapotranspiration in a Bornean tropical rainforest. Agric. For. Meteorol.128, 81–92.
Lopez, M.L., Saito, H., Kobayashi, Y., Shirota, T., Iwahaba, G., Maximov, T.C., Fukuda,M., 2007. Interannual environmental-soil thawing rate variation and its controlon transpiration from Larix cajanderi, Central Yakutia, Eastern Siberia. J. Hydrol.338, 251–260.
McMillen, R.T., 1988. An eddy correlation technique with extended applicability tonon-simple terrain. Boundary Layer Meteorol. 43, 231–245.
McNaughton, K.G., Jarvis, P.G., 1983. Predicting effects of vegetation changes ontranspiration and evaporation. In: Kozlowski, T.T. (Ed.), Water Deficits and PlantGrowth, vol. V. Academic Press, New York, pp. 1–48.
Miyahara, M., 2005. Responses of Boreal Vegetation to Soil Moisture. MastersThesis. Nagoya University, Japan (in Japanese).
Nakai, T., van der Molen, M.K., Gash, J.H.C., Kodama, Y., 2006. Correction of sonicanemometer angle of attack errors. Agric. For. Meteorol. 136, 19–30.
Ohta, T., Hiyama, T., Tanaka, H., Kuwada, T., Maximov, T.C., Ohata, T., Fukushima, Y.,2001. Seasonal variation in the energy and water exchanges above and below alarch forest in eastern Siberia. Hydrol. Process 15, 1459–1476.
Ohta, T., Maximov, T.C., Dolman, H., Nakai, T., van der Molen, M., Kononov, A.,Maximov, A., Hiyama, T., Iijima, Y., Moors, E.J., Tanaka, H., Toba, T., Yabuki, H.,2008. Interannual variation in water balance and summer evapotranspiration inan eastern Siberian larch forest over a 7-year period (1998–2006). Agric. For.Meteorol. 148, 1941–1953.
Sellers, P., Hall, F., Margolis, H., Baldocchi, D., den Hartog, G., Cihlar, J., Ryan,M.G., Goodison, B., Crill, P., Ranson, K.J., Lettenmaier, D., Wickland, D.E.,1995. The Boreal Ecosystem–Atmosphere Study (BOREAS): an overview andearly results from the 1994 field year. Bull. Am. Meteorol. Soc. 76, 1549–1577.
Tanaka, H., Hiyama, T., Kobayashi, N., Yabuki, H., Ishii, Y., Desyatkin, R.V., Maximov,T.C., Ohta, T., 2008. Energy balance and its closure over a young larch forest ineastern Siberia. Agric. For. Meteorol. 148, 1954–1967.
Tchebakova, N.M., Kolle, O., Zolotoukhine, D., Arneth, A., Styles, J.M., Vygodskaya,N.N., Schulze, E.-D., Shibistova, O., Lloyd, J., 2002. Inter-annual and seasonalvariations of energy and water fluxes above a Pinus sylvestris forest in theSiberian middle taiga. Tellus 54B, 537–551.
Toba, T., Ohta, T., 2002. Modeling the characteristics of interception loss in forests. J.Jpn. Soc. Hydrol. Water Resour. 15, 345–362 (in Japanese with English sum-mary).
Toba, T., Ohta, T., 2005. An observational study of the factors that influenceinterception loss in boreal and temperate forests. J. Hydol. 313, 208–220.
Webb, E.K., Pearman, G.I., Leuning, R., 1980. Correction of flux measurements fordensity effects due to heat and water vapour transfer. Q. J. R. Meteorol. Soc. 106,85–100.
Wilson, K.B., Hanson, P.J., Baldocchi, D.D., 2000. Factors controlling evaporation andenergy partitioning beneath a deciduous forest over an annual cycle. Agric. For.Meteorol. 102, 83–103.
Wilson, K., Goldstein, A., Falge, E., Aubinet, M., Baldocchi, D., Berbigier, P., Bernhofer,C., Ceulemans, R., Dolman, H., Field, C., Grelle, A., Ibrom, A., Law, B.E., Kowalski,A., Meyers, T., Moncrieff, J., Monson, R., Oechel, W., Tenhunen, J., Valentini, R.,Verma, S., 2002. Energy balance closure at FLUXNET sites. Agric. For. Meteorol.113, 223–243.