Elsevier

Remote Sensing of Environment

Volume 124, September 2012, Pages 581-595
Remote Sensing of Environment

Global estimation of evapotranspiration using a leaf area index-based surface energy and water balance model

https://doi.org/10.1016/j.rse.2012.06.004Get rights and content

Abstract

Studies of global hydrologic cycles, carbon cycles and climate change are greatly facilitated when global estimates of evapotranspiration (E) are available. We have developed an air-relative-humidity-based two-source (ARTS) E model that simulates the surface energy balance, soil water balance, and environmental constraints on E. It uses remotely sensed leaf area index (Lai) and surface meteorological data to estimate E by: 1) introducing a simple biophysical model for canopy conductance (Gc), defined as a constant maximum stomatal conductance gsmax of 12.2 mm s 1 multiplied by air relative humidity (Rh) and Lai (Gc = gs max × Rh × Lai); 2) calculating canopy transpiration with the Gc-based Penman–Monteith (PM) E model; 3) calculating soil evaporation from an air-relative-humidity-based model of evapotranspiration (Yan & Shugart, 2010); 4) calculating total E (E0) as the sum of the canopy transpiration and soil evaporation, assuming the absence of soil water stress; and 5) correcting E0 for soil water stress using a soil water balance model.

This physiological ARTS E model requires no calibration. Evaluation against eddy covariance measurements at 19 flux sites, representing a wide variety of climate and vegetation types, indicates that daily estimated E had a root mean square error = 0.77 mm d 1, bias =  0.14 mm d 1, and coefficient of determination, R2 = 0.69. Global, monthly, 0.5°-gridded ARTS E simulations from 1984 to 1998, which were forced using Advanced Very High Resolution Radiometer Lai data, Climate Research Unit climate data, and surface radiation budget data, predicted a mean annual land E of 58.4 × 103 km3. This falls within the range (58 × 103–85 × 103 km3) estimated by the Second Global Soil Wetness Project (GSWP-2; Dirmeyer et al., 2006). The ARTS E spatial pattern agrees well with that of the global E estimated by GSWP-2. The global annual ARTS E increased by 15.5 mm per decade from 1984 to 1998, comparable to an increase of 9.9 mm per decade from the model tree ensemble approach (Jung et al., 2010). These comparisons confirm the effectivity of the ARTS E model to simulate the spatial pattern and climate response of global E. This model is the first of its kind among remote-sensing-based PM E models to provide global land E estimation with consideration of the soil water balance.

Highlights

► The first of its kind to provide evapotranspiration (E) with consideration of the soil water balance. ► Introducing a new leaf area index (Lai) based canopy conductance (Gc) model (Gc = gsmax × Rh × Lai). ► No calibration and validation at 19 flux sites indicates nice daily and monthly error statistics. ► Global gridded ARTS E agrees well with spatial pattern of the global E estimated by GSWP-2. ► The global annual ARTS E increased by 15.5 mm per decade from 1984 to 1998.

Introduction

As a crucial process in the terrestrial ecosystem connecting atmosphere, vegetation, and soil spheres, land evapotranspiration (E) is an important component of the water and energy cycles, and plant transpiration is driven by the same stomatal conductance term that governs carbon cycle. Global E consumes more than 50% of absorbed solar energy (Trenberth et al., 2009), and returns about 60% of annual land precipitation to the atmosphere (Oki & Kanae, 2006). Much evidence, mainly drawn from precipitation and runoff datasets, has confirmed the modification of the hydrologic cycle (Alkama et al., 2011, Huntington, 2006, Labat et al., 2004).

Direct observational evidence of this intensification of global land E is, unfortunately, lacking because there are only about 400 flux stations worldwide and their temporal records are very short (Huntington, 2006, Jung et al., 2010). However, large-scale E estimation is required for answering questions related to climate change. Climate change is expected to increase the global available renewable freshwater resources, but the increasing probability of drought and changes to regional precipitation patterns may offset this effect and lead to water stresses in many regions (Oki & Kanae, 2006). Since leaf stomata control the exchange of water and carbon between vegetation and atmosphere, and high stomatal conductance leads to higher transpiration and photosynthesis, an understanding of global E variation will help to elucidate the effects of climate change on biogeochemical cycling (Dang et al., 1997, Huntington, 2006, Jarvis, 1976, Kelliher et al., 1995, Nemani and Running, 1989, Shugart, 1998).

The surface energy balance partitions the available energy (Rn  G) between turbulent heat fluxes (λE and H):λE=RnGH,where λE is latent heat flux (λ is the latent heat of vaporization, and E is evapotranspiration), Rn is net radiation, G is ground heat flux, and H is sensible heat flux. E is mainly controlled by three factors: available water, available energy, and conductivity of the ecosystem to water vapor (Batra et al., 2006).

Satellite remote sensing can supply temporally and spatially continuous observations of key biophysical variables of the land surface, such as Lai, vegetation index (VI), albedo, land surface temperature, and emissivity. It has ushered in a new era for the development of land E models (Cleugh et al., 2007, Fisher et al., 2008, Leuning et al., 2008, Mu et al., 2007, Mu et al., 2011, Nagler et al., 2005, Su, 2002, Wang and Liang, 2008). There are two principal types of remote sensing E models: empirical and physical.

Section snippets

Empirical E models

These models often apply statistical regression to estimate E, using satellite VI and other meteorological data, such as air temperature and surface net radiation (Nagler et al., 2005, Wang and Liang, 2008). More recently, Jung et al. (2010) developed a model tree ensemble (MTE) approach that predicts global land E based on a set of explanatory variables (remote sensing-based fraction of absorbed photosynthetically active radiation data, and surface meteorological data), according to model

Physical E models

Physical E models use different biophysical metrics, derived from remote sensing. They can be further classified into two types:

  • (1)

    Energy balance E models. They estimate instantaneous E rates as a residual of the land surface energy balance using thermal infrared temperature as the most important input, combined with other data. Examples of this type include the Surface Energy Balance Algorithm for Land (SEBAL; Bastiaanssen et al., 1998), the Surface Energy Balance System (SEBS; Su, 2002), and the

Evapotranspiration algorithm

We propose a two-source E model to calculate actual E, in two steps. The first is to estimate plant transpiration and soil evaporation using respective equations, under the assumption of plentiful soil water. The second is to account for the effects of soil water stress, using a SWB model. The main improvements to the PM model in this study are explicit consideration of soil water stress impact on E.

Naturally, the available energy A is partitioned to two parts: the soil part (As) and canopy

Observation data for model evaluation

ET and meteorological data, measured at 19 AmeriFlux flux-tower sites (Table 2) by the eddy-covariance (EC) method, were used in model evaluation. The EC method is widely accepted for directly measuring heat fluxes (Paw et al., 2000) and is widely applied to global E measurements at flux tower sites in FLUXNET (Baldocchi et al., 2001). The AmeriFlux network is a core part of the global FLUXNET network. It includes sites from North, Central, and South America and continuously observes

Model evaluation at 19 flux sites

Statistics of model performance at the daily scale for all 19 sites (Table 4) show that the ARTS E goodness of fit and error varied from site to site. The slopes of the linear regression of estimated E vs. observed E ranged from 0.58 at Vaira to 1.32 at Bartlett, and the intercepts varied from − 0.08 mm d 1 at Tonzi to 0.48 mm d 1 at Donaldson. The E model had an average RMSE of 0.75 mm d 1 for all sites, ranging from 0.45 mm d 1 at Santa to 1.09 mm d 1 at Donaldson. An average bias of − 0.11 mm d 1 was

Discussion

The evaluation of ARTS E at flux sites was affected by measurement error of flux E data. The EC method has an energy imbalance problem, i.e., net radiation Rn minus ground heat flux G is greater than the sum of latent heat flux λE and sensible heat flux H at many eddy flux sites (Leuning et al., 2008, Wilson et al., 2002, Yan and Shugart, 2010). The ratio of λE + H to Rn  G is about 0.8 for global FLUXNET measurements (Wilson et al., 2002). Thus, a correction method, i.e., energy closure ratios

Conclusions

The ARTS E model uses remote sensing observations to predict the rates of E globally. It accounts for the impact of net radiation, air temperature, air moisture deficit, soil water deficit, and vegetation LAI, thereby adequately representing the principles of surface energy balance and water balance. The ARTS E model shows good agreement with observed E at 19 flux sites, at daily and monthly scales. The evaluation also indicates that the PM equation provides a practical framework with which to

Acknowledgments

The authors would like to thank the flux site investigators for allowing us to use their flux data through AmeriFlux program for the development of ARTS E model. This work was supported by National Natural Science Foundation of China (41171284, 40801129), Special Fund for Meteorological Research in the Public Interest (GYHY201106027, 200906022) and Chinese Academy of Sciences (XDA05050602-1). Flux observations at UMBS were supported by US DoE grant # DE-SC0006708. Finally the reviewers and Dr.

References (105)

  • W.P. Kustas et al.

    Evaluation of soil and vegetation heat flux predictions using a simple two-source model with radiometric temperatures for partial canopy cover

    Agricultural and Forest Meteorology

    (1999)
  • D. Labat et al.

    Evidence for global runoff increase related to climate warming

    Advances in Water Resources

    (2004)
  • J.J. Landsberg et al.

    A generalised model of forest productivity using simplified concepts of radiation-use efficiency, carbon balance and partitioning

    Forest Ecology and Management

    (1997)
  • S.Y. Ma et al.

    Inter-annual variability in carbon dioxide exchange of an oak/grass savanna and open grassland in California

    Agricultural and Forest Meteorology

    (2007)
  • A.M. Moffat et al.

    Comprehensive comparison of gap-filling techniques for eddy covariance net carbon fluxes

    Agricultural and Forest Meteorology

    (2007)
  • Q. Mu et al.

    Development of a global evapotranspiration algorithm based on MODIS and global meteorology data

    Remote Sensing of Environment

    (2007)
  • Q. Mu et al.

    Improvements to a MODIS global terrestrial evapotranspiration algorithm

    Remote Sensing of Environment

    (2011)
  • R.B. Myneni et al.

    Global products of vegetation leaf area and fraction absorbed PAR from year one of MODIS data

    Remote Sensing of Environment

    (2002)
  • P.L. Nagler et al.

    Predicting riparian evapotranspiration from MODIS vegetation indices and meteorological data

    Remote Sensing of Environment

    (2005)
  • J.M. Norman et al.

    Source approach for estimating soil and vegetation energy fluxes in observations of directional radiometric surface temperature

    Agricultural and Forest Meteorology

    (1995)
  • Y. Ryu et al.

    On the temporal upscaling of evapotranspiration from instantaneous remote sensing measurements to 8-day mean daily-sums

    Agricultural and Forest Meteorology

    (2012)
  • B. Saugier et al.

    Some plant factors controlling evapotranspiration

    Agricultural and Forest Meteorology

    (1991)
  • J.B. Stewart

    Modelling surface conductance of pine forest

    Agricultural and Forest Meteorology

    (1988)
  • J.B. Stewart et al.

    Preliminary modelling of transpiration from the FIFE site in Kansas

    Agricultural and Forest Meteorology

    (1989)
  • A.E. Suyker et al.

    Interannual water vapor and energy exchange in an irrigated maize-based agroecosystem

    Agricultural and Forest Meteorology

    (2008)
  • R. Suzuki et al.

    Interannual covariability between actual evapotranspiration and PAL and GIMMS NDVIs of northern Asia

    Remote Sensing of Environment

    (2007)
  • T.E. Twine et al.

    Correcting eddy-covariance flux underestimates over a grassland

    Agricultural and Forest Meteorology

    (2000)
  • R.K. Vinukollu et al.

    Global estimates of evapotranspiration for climate studies using multi-sensor remote sensing data: Evaluation of three process-based approaches

    Remote Sensing of Environment

    (2011)
  • K. Wilson et al.

    Energy balance closure at FLUXNET sites

    Agricultural and Forest Meteorology

    (2002)
  • J. Xiao et al.

    Estimation of net ecosystem carbon exchange for the conterminous United States by combining MODIS and AmeriFlux data

    Agricultural and Forest Meteorology

    (2008)
  • W. Yuan et al.

    Global estimates of evapotranspiration and gross primary production based on MODIS and global meteorology data

    Remote Sensing of Environment

    (2010)
  • R. Alkama et al.

    Trends in global and basin-scale runoff over the late twentieth century: Methodological issues and sources of uncertainty

    Journal of Climate

    (2011)
  • R.G. Allen

    Crop evapotranspiration: Guidelines for computing crop water requirements

    (1998)
  • M.C. Anderson et al.

    A climatological study of evapotranspiration and moisture stress across the continental United States based on thermal remote sensing: 1. Model formulation

    Journal of Geophysical Research-Atmospheres

    (2007)
  • D. Baldocchi et al.

    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

    (2001)
  • J. Ball et al.

    A model predicting stomatal conductance and its contribution to the control of photosynthesis under different environmental conditions

  • W.G.M. Bastiaanssen et al.

    A remote sensing surface energy balance algorithm for land (SEBAL)

    Journal of Hydrology

    (1998)
  • G.B. Bonan

    A land surface model (LSM version 1.0) for ecological, hydrological, and atmospheric studies: technical description and user's guide

  • R.J. Bouchet

    Evapotranspiration reelle et potentielle, signification climatique

    International Association of Science and Hydrology

    (1963)
  • W. Buermann et al.

    Evaluation of the utility of satellite-based vegetation leaf area index data for climate simulations

    Journal of Climate

    (2001)
  • D. Chen et al.

    Comparison of the Thornthwaite method and pan data with the standard Penman–Monteith estimates of reference evapotranspiration in China

    Climate Research

    (2005)
  • K.L. Clark et al.

    Carbon dynamics along a chronosequence of slash pine plantations in north Florida

    Ecological Applications

    (2004)
  • Y. Dai et al.

    A two-big-leaf model for canopy temperature, photosynthesis, and stomatal conductance

    Journal of Climate

    (2004)
  • Q.L. Dang et al.

    Regulation of branch-level gas exchange of boreal trees: Roles of shoot water potential and vapor pressure difference

    Tree Physiology

    (1997)
  • P.A. Dirmeyer et al.

    GSWP-2: Multimodel analysis and implications for our perception of the land surface

    Bulletin of the American Meteorological Society

    (2006)
  • D. Dragoni et al.

    Uncertainty of annual net ecosystem productivity estimated using eddy covariance flux measurements

    Journal of Geophysical Research-Atmospheres

    (2007)
  • M. Falk et al.

    Flux partitioning in an old-growth forest: Seasonal and interannual dynamics

    Tree Physiology

    (2008)
  • R.F. Grant et al.

    Net biome productivity of irrigated and rainfed maize–soybean rotations: Modeling vs. measurements

    Agronomy Journal

    (2007)
  • S.K. Gupta et al.

    A parameterization for longwave surface radiation from satellite data: Recent improvements

    Journal of Applied Meteorology

    (1992)
  • S.K. Gupta et al.

    The Langley Parameterized Shortwave Algorithm (LPSA) for Surface Radiation Budget Studies. NASA/TP-2001-211272

    (2001)
  • Cited by (134)

    View all citing articles on Scopus
    View full text