Actual evapotranspiration in drylands derived from in-situ and satellite data: Assessing biophysical constraints
Highlights
► A remote sensing evapotranspiration model was evaluated in two dryland sites. ► Soil moisture was estimated with a Thermal Inertia index. ► Results in the African savanna were similar to those with measured soil moisture. ► High potential for regionalization as most of inputs are global satellite data.
Introduction
Evapotranspiration (or latent heat flux expressed in energy terms, λE) represents 90% of the annual precipitation in water-limited regions which cover 40% of the Earth's surface (Glenn et al., 2007). In these regions there is a close link between carbon and water cycles (Baldocchi, 2008) where water availability is the main control for biological activity (Brogaard et al., 2005). λE rates also determine groundwater recharge (Huxman et al., 2005) and feedbacks to continental precipitation patterns (Huntington, 2006). The Sahel and the Mediterranean basin are both located in transitional climate regions and are thus expected to be extremely sensitive to climate change (Giorgi & Lionello, 2008). The land surface is a strong amplifier on the inter-annual variability of the West African Monsoon leading to the observed persistency patterns (Nicholson, 2000, Taylor et al., 2011, Timouk et al., 2009). Therefore, improving estimates of temporal and spatial variations of λE is crucial for understanding land surface–atmosphere interactions and to improve hydrological and agricultural management (Yuan et al., 2010).
λE can be estimated at regional scales using remote sensing data. One way is to use models based on the bulk resistance equation for heat transfer (Brutsaert, 1982), relying on the difference between surface temperature (Ts) and air temperature (Ta) and the aerodynamic resistance to turbulent heat transport. In this case, λE is estimated indirectly as a residual of the surface energy balance equation (Anderson et al., 2007, Chehbouni et al., 1997). This approach circumvents the problem of estimating soil and canopy surface resistances to water vapor, needed to compute λE, that tend to be more critical in λE modeling than aerodynamic resistances in dryland regions (Verhoef, 1998, Were et al., 2007). In those regions, two-source models treating the land surface as a composite of soil and vegetation elements with different temperatures, fluxes, and atmospheric coupling provide better results than single-source models (Anderson et al., 2007). However, despite the strong physical basis of two-source models (Kustas and Norman, 1999, Norman et al., 1995) their spatialization is difficult because the task of estimating aerodynamic resistances at instantaneous time scales is not trivial, requiring knowledge about atmospheric stability, several vegetation and soil parameters as well as meteorological data (Fisher et al., 2008). Further complications arise from the partition of Ts between soil and vegetation (Kustas & Norman, 1999) because the radiative surface temperature differs from the aerodynamic surface temperature especially over sparsely vegetated surfaces (Chehbouni et al., 1997).
A second group of models using remote sensing data directly solves the λE term using the Penman–Monteith (PM) combination equation. In this case, λE can be partitioned into soil and vegetation components (Leuning et al., 2008). With this approach, the challenge is to characterize the spatial and temporal variation in surface conductances to water vapor without using field calibration (Zhang et al., 2010). A simple way to estimate surface conductances is to use prescribed sets of parameters based on biome-type maps (Zhang et al., 2010). Other approaches perform optimization with field data but can lead to a lack of estimates over vast regions of the globe, such as the Sahel, due to the scarcity of field measurements (Yuan et al., 2010). One of the first attempts to characterize surface conductance without optimization proposed an empirical relationship with LAI derived from MODIS (Moderate Resolution Imaging Spectroradiometer) (Cleugh et al., 2007). Mu et al., 2007, Mu et al., 2011 refined this approach using the empirical multiplicative model proposed by (Jarvis, 1976) estimating moisture and temperature constraints on stomatal conductance and upscaling leaf stomatal conductance to canopy. Alternatively, Leuning et al. (2008) used a biophysical model for surface conductance based on Kelliher et al. (1995) method. However, this method required optimization with field data for gsx, the maximum stomatal conductance of leaves, and for the soil water content. As both parameters were held constant along the year λE was overestimated at drier sites. To address this shortcoming, Zhang et al. (2008) introduced a variable-soil moisture fraction dependent on rainfall, and optimized gsx using outputs from an annual water balance model or a Budyko-type model (Zhang et al., 2008, Zhang et al., 2010). Although this represented a step-forward for operational applications, results at dry sites were still poorer than at more humid sites (Zhang et al., 2008, Zhang et al., 2010).
A solution to overcome those parameterization problems using the Penman–Monteith equation, was the simplification proposed by Priestley and Taylor (1972) (PT) for equilibrium evapotranspiration over large regions by replacing the surface and aerodynamic resistance terms with an empirical multiplier αPT (Zhang et al., 2009). The PT equation is theoretically less accurate than PM although uncertainties in parameter estimation using PM can result in higher errors (Fisher et al., 2008). Fisher et al. (2008) proposed a model based on PT to estimate monthly actual λE. The authors used biophysical constraints to reduce λE from a maximum potential value, λEp, in response to multiple stresses. One advantage of this approach is that it does not require information regarding biome-type or calibration with field data. The modeling framework can be seen as conceptually similar to the so-called Production Efficiency Models (PEM) for estimating GPP (Gross Primary Productivity) (Houborg et al., 2009, Monteith, 1972, Potter et al., 1993, Verstraeten et al., 2006a) where maximum light use efficiency (ε) of conversion of absorbed energy fAPAR into carbon is reduced below its maximum potential due to environmental stresses. In fact, part of the formulation from the PT-JPL model has been introduced into some PEM models (Yuan et al., 2010). The main model assumption is that plants optimize their capacity for energy acquisition in a way that changes in parallel with the physiological capacity for transpiration (Fisher et al., 2008, Nemani and Running, 1989). This idea is to some extent related to the hydrological equilibrium hypothesis stating that in water-limited natural systems, plants adjust canopy development to minimize water losses and maximize carbon gains (Eagleson, 1986) but applied over shorter time-scales. The modeling approach described above neglects the behavior of individual leaves and considers the canopy response to its environment in bulk for which it can be referred to as a top–down approach (Houborg et al., 2009). Top–down approaches use simpler scaling rules compared to bottom–up models that require detailed mechanistic descriptions of leaf-level processes up-scaled to the canopy (Schymanski et al., 2009). Although top–down approaches require less parameters than bottom–up approaches, they are subjected to a higher degree of empiricism with high uncertainty on the functional responses of ecosystem processes to environmental stresses (Yuan et al., 2010).
The use of global satellite vegetation products and meteorological gridded databases as input to top–down approaches based on the PM or the PT equations has made possible to obtain regional estimates of evapotranspiration (Mu et al., 2007). However, there are still limitations regarding the use of such databases. One hand, existing global climatic data sets interpolated from observations such as the Climatic Research Unit data set (CRU, University of East Anglia) are available on a monthly but not a daily basis (New et al., 2000). Moreover, data from reanalyses such as ECMWF (European Centre for Medium-Range Weather forecasts) or NCEP/NCAR present coarse spatial resolutions (≈ 1.25°) (Mu et al., 2007) being desirable to minimize the use of climatic data when possible.
On the other hand, PM and PT satellite-based approaches have taken advantage of optical remote sensing data to estimate vegetation properties but thermal remotely sensed data has been used only marginally and with coarse spatial resolution data such as the microwave AMSR-E at 0.25° (Miralles et al., 2011). Incorporation of longwave infrared thermal data at spatial resolutions of 1–3 km available from the MODIS (Moderate Resolution Imaging Spectroradiometer) or the SEVIRI (Spinning Enhanced Visible and Infrared Imager) sensors could help to track changes in surface conductance (Berni et al., 2009, Boegh et al., 2002), soil evaporation (Qiu et al., 2006), surface water deficit (Boulet et al., 2007, Moran et al., 1994) or soil water content (Gillies and Carlson, 1995, Nishida, 2000, Sandholt et al., 2002). In relation to soil moisture a promising approach is the mapping of soil moisture based on soil thermal inertia (Cai et al., 2007, Sobrino et al., 1998, Verstraeten et al., 2006b), following the early work of Price (1977) and Cracknell and Xue (1996).
The objective of this work was to adapt and evaluate a daily version of the PT-JPL model and introduce a new formulation for soil moisture based on the thermal inertia concept. The aim is to minimize the need for climatic reanalyses data by incorporating thermal remote sensing information in order to facilitate future model regionalization. The PT-JPL model in its original formulation has proven to be successful over 36 Fluxnet sites at monthly time scales, ranging from boreal to temperate and tropical ecosystems. However, none of those included semiarid vegetation with annual rainfall below 400 mm (Fisher et al., 2008, Fisher et al., 2009). Model performance using in-situ and satellite data was compared with field data from Eddy Covariance systems at two semiarid sites: an open woody savannah in the Sahel (Mali) and Mediterranean tussock grassland (Spain). Finally, to place the results in the context of global drylands, model results were compared to published results from similar models using remote sensing at dryland savanna and grasslands sites across the globe.
Section snippets
Field sites and data
Two field sites (Fig. 1) have been used to test the model in semiarid conditions: an open woody savannah in Mali and tussock grassland in Spain. A general description of the sites is included in Table 1.
PT-JPL-daily model description
The daily model proposed here (hereafter PT-JPL-daily) is a modified version of the algorithm described in Fisher et al. (2008) where “λE” is partitioned into canopy transpiration (λEc) and soil evaporation (λEs) (Eq. 1). In this paper, we did not consider interception evaporation (λEi), or evaporation from a wet canopy surface, as in low LAI ecosystems it accounts for a limited amount of the total water flux (Mu et al., 2011) and in turn using it requires observations of relative humidity at
Global sensitivity analyses (EFAST) approach
Considering the variability around mean annual conditions, the contribution to uncertainty was less than 20% for most parameters and variables in the Sahelian savanna. The greatest uncertainty was due to two of the biophysical constraints: fSM and fT with 22.19% and 17.68% respectively (total effect). Five other variables involved in LAI estimation and energy partition between soil and canopy contributed around 12% to model uncertainty (Fig. 3). However, the relative importance of each variable
Conclusions
The Priestley Taylor-Jet Propulsion Laboratory (PT-JPL) evapotranspiration λE model, developed by Fisher et al. (2008) is based on the Priestley–Taylor equation downscaled according to multiple stresses. The PT-JPL is attractive for its simplicity and potential for regionalization using satellite data. In this study, a daily version of the model was evaluated in some of the most extreme conditions from the point of water availability: an open woody savanna in the Sahel and a Mediterranean
Acknowledgments
This study was funded by the Danish Council for Independent Research and Technology and Production Sciences (FTP) Grant 09-070382, the research project RNM-6685 financed by the Regional Andalucian Government (Spain), and the CARBORAD project CGL2011-27493 (Plan Nacional MICINN, Spain). MODIS data were obtained through the online Data Pool at the NASA Land Processes Distributed Active Archive Center (LP DAAC), USGS/Earth Resources Observation and Science Center, Sioux Falls, South Dakota (//lpdaac.usgs.gov/get_data
References (101)
- et al.
Mapping canopy conductance and CWSI in olive orchards using high resolution thermal remote sensing imagery
Remote Sensing of Environment
(2009) - et al.
Evaluating evapotranspiration rates and surface conditions using Landsat TM to estimate atmospheric resistance and surface resistance
Remote Sensing of Environment
(2002) - et al.
Primary production of Inner Mongolia, China, between 1982 and 1999 estimated by a satellite data-driven light use efficiency model
Global and Planetary Change
(2005) - et al.
Designing a spectral index to estimate vegetation water content from remote sensing data: Part 1—Theoretical approach
Remote Sensing of Environment
(2002) - et al.
Estimation of sensible heat flux over sparsely vegetated surfaces
Journal of Hydrology
(1997) - et al.
Regional evaporation estimates from flux tower and MODIS satellite data
Remote Sensing of Environment
(2007) - et al.
Nonlinear sensitivity analysis of multiparameter model systems
Journal of Computational Physics
(1978) - et al.
Multi-scale soil moisture measurements at the Gourma meso-scale site in Mali
Journal of Hydrology
(2009) - et al.
Global estimates of the land–atmosphere water flux based on monthly AVHRR and ISLSCP-II data, validated at 16 FLUXNET sites
Remote Sensing of Environment
(2008) - et al.
Monitoring land degradation risk using ASTER data: The non-evaporative fraction as an indicator of ecosystem function
Remote Sensing of Environment
(2008)
Climate change projections for the Mediterranean region
Global and Planetary Change
Surface thermodynamics and radiative budget in the Sahelian Gourma: Seasonal and diurnal cycles
Journal of Hydrology
Intercomparison of a 'bottom-up' and 'top-down' modeling paradigm for estimating carbon and energy fluxes over a variety of vegetative regimes across the U.S
Agricultural and Forest Meteorology
Overview of the radiometric and biophysical performance of the MODIS vegetation indices
Remote Sensing of Environment
Evidence for intensification of the global water cycle: Review and synthesis
Journal of Hydrology
Maximum conductances for evaporation from global vegetation types
Agricultural and Forest Meteorology
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
A general approach to estimate soil water content from thermal inertia
Agricultural and Forest Meteorology
Estimation of regional net primary productivity (NPP) using a process-based ecosystem model: How important is the accuracy of climate data?
Ecological Modelling
High resolution remote estimation of soil surface water content by a thermal inertia approach
Journal of Hydrology
Operational estimates of areal evapotranspiration and their significance to the science and practice of hydrology
Journal of Hydrology
The AMMA-CATCH Gourma observatory site in Mali: Relating climatic variations to changes in vegetation, surface hydrology, fluxes and natural resources
Journal of Hydrology
Development of a global evapotranspiration algorithm based on MODIS and global meteorology data
Remote Sensing of Environment
Improvements to a MODIS global terrestrial evapotranspiration algorithm
Remote Sensing of Environment
Global products of vegetation leaf area and fraction absorbed PAR from year one of MODIS data
Remote Sensing of Environment
On the relationship between FAPAR and NDVI
Remote Sensing of Environment
Coupling diffusion and maximum entropy models to estimate thermal inertia
Remote Sensing of Environment
Source approach for estimating soil and vegetation energy fluxes in observations of directional radiometric surface-temperature
Agricultural and Forest Meteorology
Theoretical analysis of a remotely measurable soil evaporation transfer coefficient
Remote Sensing of Environment
Calibrating a soil–vegetation–atmosphere transfer model with remote sensing estimates of surface temperature and soil surface moisture in a semi arid environment
Journal of Hydrology
A simple interpretation of the surface temperature/vegetation index space for assessment of surface moisture status
Remote Sensing of Environment
SEtHyS_Savannah: A multiple source land surface model applied to Sahelian landscapes
Agricultural and Forest Meteorology
Thermal inertia mapping from NOAA-AVHRR data
Synergistic use of multisensor data for land processes
A remote sensing driven distributed hydrological model of the Senegal River basin
Journal of Hydrology
Response of surface energy balance to water regime and vegetation development in a Sahelian landscape
Journal of Hydrology
Correcting eddy-covariance flux underestimates over a grassland
Agricultural and Forest Meteorology
The potential of multitemporal Aqua and Terra MODIS apparent thermal inertia as a soil moisture indicator
International Journal of Applied Earth Observation and Geoinformation
The relative importance of surface and aerodynamic resistances in a multi-source energy-CO2 model
Physics and Chemistry of the Earth
On temperature and water limitation of net ecosystem productivity: Implementation in the C-Fix model
Ecological Modelling
Soil moisture retrieval using thermal inertia, determined with visible and thermal spaceborne data, validated for European forests
Remote Sensing of Environment
Global estimates of evapotranspiration for climate studies using multi-sensor remote sensing data: Evaluation of three process-based approaches
Remote Sensing of Environment
Global estimates of evapotranspiration and gross primary production based on MODIS and global meteorology data
Remote Sensing of Environment
Water content estimation in vegetation with MODIS reflectance data and model inversion methods
Remote Sensing of Environment
Satellite based analysis of northern ET trends and associated changes in the regional water balance from 1983 to 2005
Journal of Hydrology
A climatological study of evapotranspiration and moisture stress across the continental United States based on thermal remote sensing: 2. Surface moisture climatology
Journal of Geophysical Research-Atmospheres
Breathing of the terrestrial biosphere: Lessons learned from a global network of carbon dioxide flux measurement systems
Australian Journal of Botany
Evapotranspiration reelle et potentielle, signification climatique
Monitoring water stress using time series of observed to unstressed surface temperature difference
Agricultural and Forest Meteorology
New light-scattering photogoniometer for accurate measurements
Journal of Physics E-Scientific Instruments
Cited by (108)
Integration of microwave satellite soil moisture products in the contextual surface temperature-vegetation index models for spatially continuous evapotranspiration estimation
2023, ISPRS Journal of Photogrammetry and Remote SensingParameter sensitivity analysis for a biochemically-based photosynthesis model
2023, Research in Cold and Arid RegionsIntegration of carbon dioxide concentration in a simplified process-based model for evapotranspiration estimation in an old-growth forest
2022, Forest Ecology and Management