Elsevier

Remote Sensing of Environment

Volume 115, Issue 8, 15 August 2011, Pages 1781-1800
Remote Sensing of Environment

Improvements to a MODIS global terrestrial evapotranspiration algorithm

https://doi.org/10.1016/j.rse.2011.02.019Get rights and content

Abstract

MODIS global evapotranspiration (ET) products by Mu et al. [Mu, Q., Heinsch, F. A., Zhao, M., Running, S. W. (2007). Development of a global evapotranspiration algorithm based on MODIS and global meteorology data. Remote Sensing of Environment, 111, 519–536. doi: 10.1016/j.rse.2007.04.015] are the first regular 1-km2 land surface ET dataset for the 109.03 Million km2 global vegetated land areas at an 8-day interval. In this study, we have further improved the ET algorithm in Mu et al. (2007a, hereafter called old algorithm) by 1) simplifying the calculation of vegetation cover fraction; 2) calculating ET as the sum of daytime and nighttime components; 3) adding soil heat flux calculation; 4) improving estimates of stomatal conductance, aerodynamic resistance and boundary layer resistance; 5) separating dry canopy surface from the wet; and 6) dividing soil surface into saturated wet surface and moist surface. We compared the improved algorithm with the old one both globally and locally at 46 eddy flux towers. The global annual total ET over the vegetated land surface is 62.8 × 103 km3, agrees very well with other reported estimates of 65.5 × 103 km3 over the terrestrial land surface, which is much higher than 45.8 × 103 km3 estimated with the old algorithm. For ET evaluation at eddy flux towers, the improved algorithm reduces mean absolute bias (MAE) of daily ET from 0.39 mm day−1 to 0.33 mm day−1 driven by tower meteorological data, and from 0.40 mm day−1 to 0.31 mm day−1 driven by GMAO data, a global meteorological reanalysis dataset. MAE values by the improved ET algorithm are 24.6% and 24.1% of the ET measured from towers, within the range (10–30%) of the reported uncertainties in ET measurements, implying an enhanced accuracy of the improved algorithm. Compared to the old algorithm, the improved algorithm increases the skill score with tower-driven ET estimates from 0.50 to 0.55, and from 0.46 to 0.53 with GMAO-driven ET. Based on these results, the improved ET algorithm has a better performance in generating global ET data products, providing critical information on global terrestrial water and energy cycles and environmental changes.

Research highlights

► Improving the MODIS ET algorithm (Mu et al., 2007a, old algorithm). ► Global terrestrial annual total ET (62.8 × 103 km3) agrees with reported 65.5 × 103 km3. ► MAE of 24.6% and 24.1% are in the 10–30% range of the accuracy of ET measurements.

Introduction

All organisms require water for their survival (Oki & Kanae, 2006). Unlike most other natural resources, water circulates and forms closed hydrological cycles. The terrestrial water cycle is of critical importance to a wide array of Earth system processes. It plays a central role in climate and meteorology, plant community dynamics, and carbon and nutrient biogeochemistry (Vörösmarty et al., 1998, Zhao and Running, 2010). Demand for the world's increasingly scarce water supply is rising rapidly, challenging its availability for food production and putting global food security at risk. Agriculture, upon which a burgeoning population depends for food, is competing with industrial, household, and environmental uses for this scarce water supply (Rosegrant et al., 2003, Vörösmarty et al., 2010). The water withdrawals from the renewable freshwater resources include blue water from the surface and groundwater as water resources, and green water from the beneficial evapotranspiration (ET) as a loss from the precipitated water over non-irrigated croplands (Oki & Kanae, 2006). Global climate change will affect precipitation and ET, and hence influence the renewable freshwater resources. ET is the second largest component (after precipitation) of the terrestrial water cycle at the global scale, since ET returns more than 60% of precipitation on land back to the atmosphere (Korzoun et al., 1978, L'vovich and White, 1990) and thereby conveys an important constraint on water availability at the land surface. In addition, ET is an important energy flux since land ET uses up more than half of the total solar energy absorbed by land surfaces (Trenberth et al., 2009). Accurate estimation of ET not only meets the growing competition for the limited water supplies and the need to reduce the cost of the irrigation projects, but also it is essential to projecting potential changes in the global hydrological cycle under different climate change scenarios (Teuling et al., 2009).

Remote sensing has long been recognized as the most feasible means to provide spatially distributed regional ET information on land surface. Remotely sensed data, especially those from polar-orbiting satellites, provide temporally and spatially continuous information over vegetated surfaces useful for regional measurement and monitoring of surface biophysical variables affecting ET, including albedo, biome type and leaf area index (LAI) (Los et al., 2000). The MODerate Resolution Imaging Spectroradiometer (MODIS) onboard NASA's Terra and Aqua satellites, provide unprecedented information regarding vegetation and surface energy (Justice et al., 2002), which can be used for regional and global scale ET estimation in near real-time. Three types of methods have been developed to estimate ET from remote sensing data: (1) empirical/statistical methods which link measured ET or estimated ET to large scales with remotely sensed vegetation indices (Glenn, Huete, et al., 2008a, Glenn, Morino, et al., 2008b, Jung et al., 2010, Nagler et al., 2005); (2) physical models that calculate ET as the residual of surface energy balance (SEB) through remotely sensed thermal infrared data (Allen et al., 2007, Bastiaanssen et al., 2005, Kustas and Anderson, 2009, Overgaard et al., 2006); (3) and other physical models such as using the Penman–Monteith logic (Monteith, 1965) to calculate ET (Cleugh et al., 2007, Mu et al., 2007a). We only describe the physical models in this article since we focus on the dynamics of ET process.

For SEB-based physical ET models, thermal-IR based land surface temperature (LST) is a critical remote sensing variable used in these satellite based SEB models (Bastiaanssen et al., 1998a, Bastiaanssen et al., 1998b, Nishida et al., 2003, Su, 2002), yet there are some disadvantages when applying LST to ET estimations at the global scale. Firstly, Hope et al. (2005) found that the relationship between thermal-IR based LST and NDVI at high-latitudes is opposite to that of mid-latitude regions because arctic tundra ecosystems characterized by permafrost provide a large sink for energy below the ground surface. Secondly, sensible heat flux (H) is estimated using the aerodynamic surface–air temperature gradient (or combination of gradients) and aerodynamic resistance, where generally LST has been used as a surrogate for aerodynamic temperature, which is the main reason that accurate estimates of H are very difficult to achieve (Gowda et al., 2008). Thirdly, among other complications, ET can often exceed net incoming radiation at a given time or place, due to advection of H from the surrounding landscape (Glenn et al., 2007). Therefore, it is common for estimated ET to incur 46% (Su, 2002) or greater than 50% error owing to the use of LST in classical sensible heat flux formulation with an aerodynamic resistance (Stewart et al., 1994).

To deal with the problems in the SEB models, Cleugh et al. (2007) developed a remotely sensed ET model using a Penman–Monteith approach driven by MODIS derived vegetation data and daily surface meteorological inputs including incoming solar radiation, surface air temperature and Vapor Pressure Deficit (VPD, the difference between saturated air vapor pressure of a given air temperature and air vapor pressure). Mu et al. (2007a) further modified Cleugh et al.'s model to estimate the global ET (RS-ET). The RS-ET algorithm uses MODIS land cover, albedo, leaf area index (LAI), and Enhanced Vegetation Index (EVI) and a daily meteorological reanalysis data set from NASA's Global Modeling and Assimilation Office (GMAO, v. 4.0.0, 2004) as inputs for regional and global ET mapping and monitoring. Fisher et al. (2008) used Priestley-Taylor method (Priestley and Taylor, 1972) to estimate global ET using AVHRR/NOAA data. Based on Mu et al., 2007a RS-ET model, Zhang et al. (2009) developed a model to estimate ET using remotely sensed Normalized Difference Vegetation Index (NDVI) data; Yuan et al. (2010) modified Mu et al., 2007a RS-ET model by adding the constraint of air temperature to stomatal conductance and calculating the vegetation cover fraction using LAI instead of EVI.

In this paper, we identified problems in the ET algorithm in Mu et al., 2007a paper (hereafter called old algorithm) and solved the problems by improving the old algorithm. In the old algorithm, ET was calculated as the sum of the evaporation from moist soil and the transpiration from the vegetation during daytime. Nighttime ET was assumed to be small and negligible. Soil heat flux (G) was assumed to be zero. For daily calculations, G might be ignored (Gavilána et al., 2007). G is a relatively small component of the surface energy budget relative to sensible and latent energy fluxes for most forest and grassland biomes (da Rocha et al., 2004, Ogée et al., 2001, Tanaka et al., 2008) and is generally less than 20% of net incoming radiation for the forest and grassland sites from this investigation (e.g. Weber et al., 2007; Granger, http://www.taiga.net/wolfcreek/Proceedings_04.pdf). However, the assumption of negligible G in the old algorithm is a significant concern for tundra. In the Arctic–Boreal regions, G can be a substantial amount of net radiation, especially early in the growing season. The assumption of a negligible G may be valid in mid-latitude regions on a daily basis, however in these areas a substantial portion of net radiation melts ice in the active layer, especially early in the growing season (Engstrom et al., 2006, Harazono et al., 1995). The old algorithm neglected the evaporation from the intercepted precipitation from plant canopy. After the event of precipitation, part of the vegetation and soil surface is covered by water. The evaporation from the saturated soil surface is much higher than the evaporation from the unsaturated soil surface, and the evaporation from the intercepted water by canopy is different from canopy transpiration. In this study, we have improved the old ET algorithm by 1) simplifying the calculation of vegetation cover fraction; 2) calculating ET as the sum of daytime and nighttime components; 3) calculating soil heat flux; 4) improving the methods to estimate stomatal conductance, aerodynamic resistance and boundary layer resistance; 5) separating dry canopy surface from the wet, and hence canopy water loss includes evaporation from the wet canopy surface and transpiration from the dry surface; and 6) dividing soil surface into saturated wet surface and moist surface, and thus soil evaporation includes potential evaporation from the saturated wet surface and actual evaporation from the moist surface. Description of the improvements is detailed in Section 2. We parameterized the improved ET algorithm by using the tower GPP, ET data, the global MODIS GPP and Chen et al.'s global precipitation data (Chen et al., 2002), which is described in Section 5. To examine the performances of the improved ET algorithm, we compared the global ET estimated by the improved ET algorithm with that by the old algorithm and other published studies; we also compared both the old and the improved ET estimates with level 4 ET measured at 46 AmeriFlux sites.

Section snippets

Improvements on the MODIS ET algorithm

Terrestrial ET includes evaporation from wet and moist soil, evaporation from rain water intercepted by the canopy before it reaches the ground, the sublimation of water vapor from ice and snow and the transpiration through stomata on plant leaves and stems. Both the old and improved ET algorithms are based on the Penman–Monteith (P–M) equation (Monteith, 1965):λE=s×A+ρ×Cp×esate/ras+γ×1+rs/rawhere λE is the latent heat flux and λ is the latent heat of evaporation; s = d(esat)/dT, the slope of

Eddy covariance flux towers

The eddy covariance technique is a widely used and accepted method to measure ecosystem-scale mass and energy fluxes. The AmeriFlux network was established in 1996, providing continuous measurements of ecosystem level exchanges of CO2, water, energy and momentum spanning diurnal, synoptic, seasonal, and interannual time scales and is currently composed of sites from North America, Central America, and South America (http://public.ornl.gov/ameriflux/). AmeriFlux is part of a “network of regional

Input datasets

In Mu et al., 2007a paper, the performance of the old algorithm was tested at 19 AmeriFlux tower sites using two different meteorological datasets in 2001: (1) aggregated daily meteorological data from the half-hour measurements at flux tower sites and (2) the global GMAO meteorological data at 1.00° × 1.25° resolution. The input albedo was version 4 0.05-degree CMG albedo. In this study, both old and improved algorithms were driven by the two sets of meteorological data and the ET estimates were

Parameterization of the improved ET algorithm

For parameterization of the improved ET algorithm, we largely follow the method for calibrating parameters of MODIS GPP/NPP algorithm (Zhao et al., 2005). Both MODIS GPP/NPP and MODIS ET algorithms use the same controlling factors from VPD and minimum temperature (Tmin) on stomatal conductance. We first adopt the parameters of VPD and Tmin setting from those for MODIS GPP/NPP algorithm (Table 1), then calibrate other parameters for each biome. Below we detail the procedure to parameterize MODIS

Implementing ET algorithm at the global scale

The old and improved ET algorithms were implemented globally over 2000–2006 at a resolution of 0.05° using the preprocessed MODIS remote sensing data and the GMAO meteorological data as detailed in Section 4.2. Fig. 3a and b show that both algorithms have the highest ET over the tropical forests, whereas dry areas and areas with short growing seasons have the lowest estimates of ET. The ET for temperate and boreal forests lies between the two extremes (Fig. 3a and b). The difference in the

Conclusions

We have improved the old ET algorithm by 1) simplifying the calculation of vegetation cover fraction (FC); 2) calculating ET as the sum of daytime and nighttime components; 3) calculating soil heat flux; 4) improving the methods to estimate stomatal conductance, aerodynamic resistance and boundary layer resistance; 5) separating dry canopy surface from the wet, and hence canopy water loss includes evaporation from the wet canopy surface and transpiration from the dry surface; and 6) dividing

Acknowledgements

This research was financially supported by the NASA Earth Observing System MODIS project (grant NNX08AG87A). Eddy covariance flux tower sites are part of both the AmeriFlux and Fluxnet networks. We gratefully acknowledge the efforts of researchers at these sites and thank them for making their data available. Sites are funded through grants from the U.S. Department of Energy (DOE) Office of Biological and Environmental Research (BER) unless otherwise noted. Data collection at the ARM SGP Burn,

References (120)

  • A.R. Desai et al.

    Comparing net ecosystem exchange of carbon dioxide between an old-growth and mature forest in the upper Midwest, USA

    Agricultural and Forest Meteorology

    (2005)
  • C. Feigenwinter et al.

    Comparison of horizontal and vertical advective CO2 fluxes at three forest sites

    Agricultural and Forest Meteorology

    (2008)
  • J.B. Fisher et al.

    Evapotranspiration models compared on a Sierra Nevada forest ecosystem

    Environmental Modelling and Software

    (2005)
  • J.B. Fisher et al.

    Global estimates of the land atmosphere water flux based on monthly AVHRR and ISLSCP-II data, validated at FLUXNET sites

    Remote Sensing of Environment

    (2008)
  • M.A. Friedl et al.

    Global land cover mapping from MODIS: Algorithms and early results

    Remote Sensing of Environment

    (2002)
  • H.L. Gholz et al.

    Energy exchange across a chronosequence of slash pine forests in Florida

    Agricultural and Forest Meteorology

    (2002)
  • M.J. Hill et al.

    Assessment of the MODIS LAI product for Australian ecosystems

    Remote Sensing of Environment

    (2006)
  • A. Huete et al.

    Overview of the radiometric and biophysical performance of the MODIS vegetation indices

    Remote Sensing of Environment

    (2002)
  • J.P. Jenkins et al.

    Refining light-use efficiency calculations for a deciduous forest canopy using simultaneous tower-based carbon flux and radiometric measurements

    Agricultural and Forest Meteorology

    (2007)
  • C.O. Justice et al.

    An overview of MODIS Land data processing and product status

    Remote Sensing of Environment

    (2002)
  • F.M. Kelliher et al.

    Maximum conductances for evaporation from global vegetation types

    Agricultural and Forest Meteorology

    (1995)
  • W. Kustas et al.

    Advances in thermal infrared remote sensing for land surface modeling

    Agricultural and Forest Meteorology

    (2009)
  • W.P. Kustas et al.

    Estimation of the soil heat flux/net radiation ratio from spectral data

    Agricultural and Forest Meteorology

    (1990)
  • B.E. Law et al.

    Environmental controls over carbon dioxide and water vapor exchange of terrestrial vegetation

    Agricultural and Forest Meteorology

    (2002)
  • H.J. Mellouli et al.

    Evaporation losses from bare soils as influenced by cultivation techniques in semi-arid regions

    Agricultural Water Management

    (2000)
  • Q. Mu et al.

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

    Remote Sensing of Environment

    (2007)
  • R.C. Musselman et al.

    Nocturnal stomatal conductance and ambient air quality standards for ozone

    Atmospheric Environment

    (2000)
  • 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. Nagler et al.

    Predicting riparian evapotranspiration from MODIS vegetation indices and meteorological data

    Remote Sensing of Environment

    (2005)
  • J. Ogée et al.

    A long-term study of soil heat flux under a forest canopy

    Agricultural and Forest Meteorology

    (2001)
  • C.B. Schaaf et al.

    First operational BRDF, albedo and nadir reflectance products from MODIS

    Remote Sensing of Environment

    (2002)
  • H.P. Schmid

    Experimental design for flux measurements: Matching scales of observations and fluxes

    Agricultural and Forest Meteorology

    (1997)
  • R.L. Scott

    Using watershed water balance to evaluate the accuracy of eddy covariance evaporation measurements for three semiarid ecosystems

    Using watershed water balance to evaluate the accuracy of eddy covariance evaporation measurements for three semiarid ecosystems. Agricultural and Forest Meteorology

    (2010)
  • N. Tanaka et al.

    A review of evapotranspiration estimates from tropical forests in Thailand and adjacent regions

    Agricultural and Forest Meteorology

    (2008)
  • D.P. Turner et al.

    Scaling gross primary production (GPP) over boreal and deciduous forest landscapes in support of MODIS GPP product validation

    Remote Sensing of Environment

    (2003)
  • R.G. Allen et al.

    Satellite-based energy balance for Mapping Evapotranspiration with Internalized Calibration (METRIC)—Applications

    Journal of Irrigation and Drainage Engineering

    (2007)
  • M. Aubinet

    Estimates of the annual net carbon and water exchange of forests: The EUROFLUX methodology

    Advances in Ecological Research

    (2000)
  • M. Aubinet et al.

    Comparing CO2 storage and advection conditions at night at different CARBOEUROFLUX sites

    Boundary-Layer Meteorology

    (2005)
  • D.D. Baldocchi

    “Breathing” of the terrestrial biosphere: Lessons learned from a global network of carbon dioxide flux measurement systems

    Australian Journal of Botany

    (2008)
  • W.G.M. Bastiaanssen et al.

    SEBAL model with remotely sensed data to improve water-resources management under actual field conditions

    J Irrig Drain

    (2005)
  • R.J. Bouchet

    Evapotranspiration re'elle evapotranspiration potentielle, signification climatique

  • M.A. Caird et al.

    Nighttime stomatal conductance and transpiration in C3 and C4 plants

    Plant Physiology

    (2007)
  • M. Chen et al.

    Global land precipitation: A 50-yr monthly analysis based on gauge observations

    Journal of Hydrometeorology

    (2002)
  • K.L. Clark et al.

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

    Ecological Applications

    (2004)
  • D. Courault et al.

    Review on estimation of evapotranspiration from remote sensing data: From empirical to numerical modeling approaches

    Irrigation and Drainage Systems

    (2005)
  • H.R. da Rocha et al.

    Seasonality of water and heat fluxes over a tropical forest in eastern Amazonia

    Ecological Applications

    (2004)
  • M.J. Daley et al.

    Interspecific variation in nighttime transpiration and stomatal conductance in a mixed New England deciduous forest

    Tree Physiology

    (2006)
  • T.E. Dawson et al.

    Nighttime transpiration in woody plants from contrasting ecosystems

    Tree Physiology

    (2007)
  • J. Demarty et al.

    Assimilation of global MODIS leaf area index retrievals within a terrestrial biosphere model

    Geophysical Research Letters

    (2007)
  • R. Engstrom et al.

    Modeling evapotranspiration in Arctic coastal plain ecosystems using a modified BIOME-BGC model

    Journal of Geophysical Research

    (2006)
  • Cited by (0)

    View full text