Satellite-based estimation of evapotranspiration of an old-growth temperate mixed forest
Introduction
Evapotranspiration (ET) is the major link between the global energy budgets and hydrological cycles (Smith and Choudhury, 1990), and is a key component of the water budget at a wide range of scales. Accurate monitoring and estimation of ET continuously over large area is the basis for effective spatially resolved water related applications such as management of water resource projects, risk assessments for bushfires, and flooding (Hameed and Mario, 1993, Raupach, 2001). Satellite remote sensing provides routine observations, such as vegetation and land surface temperature, and offers possibilities for extending point measurements or empirical relationships to much larger areas, and many studies have incorporated remote sensing data into ET estimation (Seguin et al., 1994, Kustas and Norman, 1996, Carlson and Buffum, 1989, Allen et al., 2005, Garatuza-Payan and Watts, 2005).
The use of remote sensing models or models coupled with remote sensing is highly recommended to monitor ET over large areas continuously (Verstraeten et al., 2005). Two kinds of approaches have been taken to predict ET from remote sensing data: energy balance-based physical model and empirical model that relate ET to vegetation index (VI) measurements over growing season.
Physical models attempt to predict ET from the surface energy balance equation (Nagler et al., 2005a). Among these models, resistance–surface energy balance model has been widely reported (Cleugh and Dunin, 1995, Hall et al., 1992, Kalma and Jupp, 1990), while its performance has been shown to be unreliable (Cleugh et al., 2007). The Penman–Monteith model provides a more robust approach to estimating land surface evaporation, while its routine application is always hindered by requiring meteorological forcing data and the aerodynamic and surface resistances (Mauser and Schädlich, 1998, Moran et al., 1996, Cleugh et al., 2007).
Empirical methods utilize the scatter plot between VIs and Ts to derive surface resistance (Price, 1990, Yang et al., 1997, Jiang and Islam, 2001) following the idea of Nemani and Running (1989). However, this requires a continuum of soil moisture (from dry bare soil to saturated bare soil) and vegetation status (from water-stressed full-cover vegetation to well-watered full-cover vegetation) to provide a range of surface conditions. The effects of soil evaporation and vegetation stresses added scatter and uncertainty into the ET estimates (Nagler et al., 2005a). Several researchers tried to explore statistical models using remote sensing to extrapolate eddy covariance water and carbon flux data to regional scales. For example, Wylie et al. (2003) related normalized difference vegetation index (NDVI) to carbon fluxes in a sage-brush-steppe ecosystem, and Nagler et al., 2005a, Nagler et al., 2005b developed an a multivariate regression equation for ET prediction over large reaches of western U.S. rivers by combining remote sensing with flux site measurements with a relative root-mean-square error (RMSE) of 25%. These studies established the potential of using statistical techniques to extrapolate ET measured at eddy covariance flux towers to a regional scale.
Eddy covariance technique is a scale-appropriate way to scale ground measurements of ET and other biophysical processes over larger areas, and to project the results over periods of years, to be used for routine monitoring of regional ecosystems (Running et al., 1999) with relatively minimum uncertainties. It directly measures ET over area possessing longitudinal dimensions ranging between a 100 m and several kilometers (Schmid, 1994) across a spectrum of times scales, ranging from hours to years (Wofsy et al., 1993, Baldocchi et al., 2001). It can provide effective information of ET and biophysical processes which are critical for evaluating ET model and understanding the factors controlling seasonal dynamics of water fluxes (Law et al., 2002).
For vegetation covered area, the close coupling between transpiration and photosynthesis processes is observed (Stanhill, 1986, Steduto et al., 1997, Reichstein et al., 2002, Yu et al., 2008). There are a number of studies of photosynthesis, gross primary production (GPP), and net primary production (NPP) using several methods including measurements, remote sensing, modelling, etc. (Potter et al., 1993, Ruimy et al., 1994, Prince and Goward, 1995, Justice et al., 1998, Xiao et al., 2004a, Turner et al., 2006). GPP often increases with ET across vegetation types. The ratios of GPP to ET, which is defined as water use efficiency (WUE), are similar across vegetation types (Law et al., 2002) and strongly influenced by weather conditions (De Wit, 1958, Tanner and Sinclair, 1983, Law et al., 2002, Abbate et al., 2004, Yu et al., 2008). WUE has been estimated by complex physical model (Gutschick, 2007) or simple meteorology drived empirical models (Wang et al., 2007). There is little report on direct estimation of WUE from satellite images.
In this paper, our objectives are (1) to understand seasonal dynamics of WUE calculated from eddy flux tower data; (2) to develop the quantitative relationship between vegetation indices and WUE, here defined as the ratio of GPP to ET; (3) to predict ET with predicted WUE and GPP estimated from a satellite-based vegetation photosynthesis model (VPM) that uses satellite imagery, air temperature, and photosynthetically actively radiation (PAR; Xiao et al., 2004a, Xiao et al., 2004b, Xiao et al., 2005), we named as EvapoTranspiration on the Coupling between Photosynthesis and Transpiration (ET-CPT) model. The study site is mature mixed forest ecosystem in Changbai mountain, North-east of China. Flux tower has been established since 2002 in this forest ecosystem and many years of data has been reported (Zhang et al., 2006a, Zhang et al., 2006b).
Section snippets
Study sites and field data
The eddy flux tower site is located in No. 1 Plot at Forest Ecosystem Opened Research Station of Changbai Mountains (128°5′45.79E and 42°24′8.88″ N, Jilin Province, P.R. China), Chinese Academy of Sciences. The annual mean temperature is 0.9–4.0 °C, annual total precipitation is 600–810 mm year−1 (evaluated over a period of 20 years). The area is covered by on average 200-year-old, multi-storied, uneven-aged, multi-species mixed forest consisting of Korean pine (Pinus koraiensis), Tilia amurensis,
Quantitative relationships between WUE and VIs and meteorological variables
We conducted simple correlation analysis between WUE and vegetation indices, and between WUE and meteorological variables (Table 1). WUE was significantly correlated with both NDVI and EVI, but the correlation coefficients were generally higher for EVI than that for NDVI (Table 1 and Fig. 2). One possible reason for this may be that NDVI values were saturated in well-vegetated areas, as the LAI at the old-growth forest site is relatively high (up to 6.1 m2 m−2), while EVI does not appear to be as
Accuracy of eddy covariance data
Flux measurements require careful calibration to produce reliable data (Rana and Katerji, 2000, Scott et al., 2004). Eddy covariance results reported in this study have been validated in several studies. Wu et al. (2005) compared ET estimate by Bowen-ratio energy balance (BREB) method with corresponding ET measurements by eddy covariance technique. Over a 1 month measurement period, the BREB-based estimates for ET were within 6% of eddy covariance estimates. Diao et al. (2005) and Wang et al.
Acknowledgements
This paper is a contribution to the ChinaFLUX projects at Changbai Mountains Forest Ecosystem Opened Research Station. This work was supported by the National Natural Science Foundation of China (No. 30590381, 30500079, 90411020). Xiao was supported by grants from NASA Land Use and Land Cover Change program.
References (59)
- et al.
On estimating total evapotranspiration from remote surface temperature measurements
Remote Sensing of Environment
(1989) - et al.
Modelling sensible heat fluxes from a wheat canopy: an evaluation of the resistance energy balance method
Journal of Hydrology
(1995) - et al.
Regional evaporation estimates from flux tower and MODIS satellite data
Remote Sensing of Environment
(2007) Plant acclimation to elevated CO2—from simple regularities to biogeographic chaos
Ecological Modelling
(2007)- et al.
Estimating evaporation from pasture using infrared thermometry: evaluation of a one-layer resistance model
Agricultural and Forest Meteorology
(1990) - et al.
Environmental controls over carbon dioxide and water vapor exchange of terrestrial vegetation
Agricultural and Forest Meteorology
(2002) - et al.
Modelling the spatial distribution of evapotranspiration on different scales using remote sensing data
Journal of Hydrology
(1998) - et al.
Combining the Penman–Monteith equation with measurements of surface temperature and reflectance to estimate evaporation rates of semiarid grassland
Agricultural and Forest Meteorology
(1996) - et al.
Predicting riparian evapotranspiration from MODIS vegetation indices and meteorological data
Remote Sensing of Environment
(2005) - et al.
Evapotranspiration on western U.S. rivers estimated using the enhanced vegetation index from MODIS and data from eddy covariance and Bowen ratio flux towers
Remote Sensing of Environment
(2005)
Measurement and estimation of actual evapotranspiration in the field under Mediterranean climate: a review
European Journal of Agronomy
A global terrestrial monitoring network, scaling tower fluxes with ecosystem modeling and EOS satellite data
Remote Sensing of Environment
Interannual and seasonal variation in fluxes of water and carbon dioxide from a riparian woodland ecosystem
Agricultural and Forest Meteorology
Surface temperature and evapotranspiration: application of local scale methods to regional scales using satellite data
Remote Sensing of Environment
Water use efficiency
Advance in Agronomy
Water-use efficiency of sweet sorghum under water stress conditions gas-exchange investigations at leaf and canopy scales
Field Crop Research
Evaluation of MODIS NPP and GPP products across multiple biomes
Remote Sensing of Environment
Estimating evapotranspiration of European forests from NOAA-imagery at satellite overpass time: towards an operational processing chain for integrated optical and thermal sensor data products
Remote Sensing of Environment
Energy balance measurements and a simple model for estimating pecan water use efficiency
Agricultural Water Management
Energy balance closure at FLUXNET sites
Agricultural and Forest Meteorology
Calibration of remotely sensed, coarse resolution NDVI to CO2 fluxes in a sage-brush-steppe ecosystem
Remote Sensing of Environment
Sensitivity of vegetation indices to atmospheric aerosols: continental-scale observations in Northern Asia
Remote Sensing of Environment
Detecting leaf phenology of seasonally moist tropical forests in South America with multi-temporal MODIS images
Remote Sensing of Environment
Satellite-based modeling of gross primary production in an evergreen needleleaf forest
Remote Sensing of Environment
Modeling gross primary production of a deciduous broadleaf forest using satellite images and climate data
Remote Sensing of Environment
Seasonal variation in carbon dioxide exchange over a 200-year-old Chinese broad-leaved Korean pine mixed forest
Agricultural and Forest Meteorology
Climate and water availability effects on water-use efficiency in wheat
Crop Science
A landsat-based energy balance and evapotranspiration model in Western US water rights regulation and planning
Irrigation and Drainage Systems
FLUXNET: a new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor and energy flux densities
Bulletin of the American Meteorological Society
Cited by (34)
Insect Ecology: An Ecosystem Approach
2022, Insect Ecology: An Ecosystem ApproachWeather-sensitive height growth modelling of Norway spruce using repeated airborne laser scanning data
2021, Agricultural and Forest MeteorologyCitation Excerpt :Change detection by repeated ALS enables precise measurements of forest height growth (Socha et al., 2017; Tompalski et al., 2021; Yu et al., 2004). Repeated ALS allows for accurate measurement of the forest height increment (Yu et al., 2008, 2006; Zhang et al., 2009) and top height (TH) growth over time (Hopkinson et al., 2008; Næsset and Nelson, 2007; Socha et al., 2020a; Yu et al., 2006). Furthermore, the mean errors in ALS-based height increment measurements have been reported to decrease from 100% of the observed values for a 1-year period to 6% for a 5-year period (Hopkinson et al., 2008).
Potential of MODIS data to track the variability in ecosystem water-use efficiency of temperate deciduous forests
2016, Ecological EngineeringCitation Excerpt :Hence, how to extrapolate site-level WUE to large areas remains a challenge. Remote sensing techniques have been used to estimate GPP and ET with high spatial and temporal coverage (Sobrino et al., 2007; Wu et al., 2010; Donohue et al., 2014; Yang et al., 2013a, 2014), as both photosynthesis and evapotranspiration are closely related to biophysical properties and environmental factors (Zhang et al., 2009; Reichstein et al., 2014). By means of MODIS GPP and ET products, large-scale ecosystem WUE can be calculated by dividing GPP with ET as defined (Sur and Choi, 2013).
Parameterizing ecosystem light use efficiency and water use efficiency to estimate maize gross primary production and evapotranspiration using MODIS EVI
2016, Agricultural and Forest MeteorologyCitation Excerpt :Although previous studies have shown the potential of PRI in tracking LUE at the leaf level based on ground spectral measurements (Filella et al., 2009; Gamon et al., 1997) and satellite observations (Drolet et al., 2008; Hall et al., 2008; Hilker et al., 2009), high sensitivity of PRI to numerous extraneous effects such as canopy structure and the view observer geometry prevents its application at large spatial (landscape-to-global) scales, thereby requiring upscaling algorithms to account for structural differences in vegetation (Hilker et al., 2010). The assumption of a constant maximum value of ELUE or EWUE for the entire growing season, across years, and across sites for a biome type is very common in GPP and ET models (Alfieri et al., 2009; Running et al., 2004; Wagle et al., 2014; Zhang et al., 2009). Our results illustrated that there was a clear seasonality of ELUE and EWUE, and the maximum values of ELUE and EWUE varied year-to-year at the same site (Fig. 2), indicating that ELUE and EWUE are not constant properties.
A remote sensing model to estimate ecosystem respiration in Northern China and the Tibetan Plateau
2015, Ecological ModellingA MODIS-based Photosynthetic Capacity Model to estimate gross primary production in Northern China and the Tibetan Plateau
2014, Remote Sensing of EnvironmentCitation Excerpt :To further assess the performance of the PCM in GPP prediction, one-year data (the last year of the time series for each vegetation type) in Table 1 were chosen to compare the predictive accuracy of the PCM with that of the MODIS GPP products and the VPM. The MODIS GPP products are the principal source of GPP at regional and global scales (Running et al., 2004; http://daac.ornl.gov/MODIS/modis.html), and the VPM is a MODIS-based LUE model and has been validated in four of our studied vegetation types (Li et al., 2007; Wu et al., 2008, 2009; Zhang et al., 2009). Plots of the predicted GPP against the observed GPP for these three models are presented in Fig. 5.