Global Estimates of Evapotranspiration and Gross Primary Production Based on MODIS and Global Meteorology Data
Wenping Yuan,Shuguang Liu,Guirui Yu,Jean-Marc Bonnefond,Jiquan Chen,Ken Davis,Ankur R. Desai,Allen H. Goldstein,Damiano Gianelle,Federica Rossi,Andrew E. Suyker,Shashi B. Verma
DOI: https://doi.org/10.1016/j.rse.2010.01.022
IF: 13.5
2010-01-01
Remote Sensing of Environment
Abstract:The simulation of gross primary production (GPP) at various spatial and temporal scales remains a major challenge for quantifying the global carbon cycle. We developed a light use efficiency model, called EC-LUE, driven by only four variables: normalized difference vegetation index (NDVI), photosynthetically active radiation (PAR), air temperature, and the Bowen ratio of sensible to latent heat flux. The EC-LUE model may have the most potential to adequately address the spatial and temporal dynamics of GPP because its parameters (i.e., the potential light use efficiency and optimal plant growth temperature) are invariant across the various land cover types. However, the application of the previous EC-LUE model was hampered by poor prediction of Bowen ratio at the large spatial scale. In this study, we substituted the Bowen ratio with the ratio of evapotranspiration (ET) to net radiation, and revised the RS-PM (Remote Sensing-Penman Monteith) model for quantifying ET. Fifty-four eddy covariance towers, including various ecosystem types, were selected to calibrate and validate the revised RS-PM and EC-LUE models. The revised RS-PM model explained 82% and 68% of the observed variations of ET for all the calibration and validation sites, respectively. Using estimated ET as input, the EC-LUE model performed well in calibration and validation sites, explaining 75% and 61% of the observed GPP variation for calibration and validation sites respectively.Global patterns of ET and GPP at a spatial resolution of 0.5° latitude by 0.6° longitude during the years 2000–2003 were determined using the global MERRA dataset (Modern Era Retrospective-Analysis for Research and Applications) and MODIS (Moderate Resolution Imaging Spectroradiometer). The global estimates of ET and GPP agreed well with the other global models from the literature, with the highest ET and GPP over tropical forests and the lowest values in dry and high latitude areas. However, comparisons with observed GPP at eddy flux towers showed significant underestimation of ET and GPP due to lower net radiation of MERRA dataset. Applying a procedure to correct the systematic errors of global meteorological data would improve global estimates of GPP and ET. The revised RS-PM and EC-LUE models will provide the alternative approaches making it possible to map ET and GPP over large areas because (1) the model parameters are invariant across various land cover types and (2) all driving forces of the models may be derived from remote sensing data or existing climate observation networks.