Mapping Evapotranspiration using MODIS and Meteorological Model Data over the Northeast Asia

Evapotranspiration (ET) is one of major variables in the hydrologic cycle of terrestrial ecosystems and in assessments of the water resources of any region. Satellite remote sensing, especially the MODerate resolution Imaging Spectroradiometer (MODIS) sensor, offers a promising technique that has been used to provide reasonable estimates of ET for wide area in daily time scale, but is hampered by frequent cloud contamination or data gaps. The objectives of this study are 1) to apply and test a stand-alone ET algorithm during clear or partial cloudy sky conditions using MODIS land surface and atmospheric products, and 2) to use the regional meteorological model such as the Regional Data Assimilation and Prediction System (RDAPS) data in cloudy conditions to facilitate continuous daily ET estimates. The MODIS ET algorithm based on the Penman-Monteith equation was applied to predict ET at some flux measurement sites over the Northeast Asia. This algorithm considers both the effects of surface energy partitioning processes and environmental constraints on ET. We devised gap-filling approaches for MODIS aerosol and albedo data that were identified as bottlenecks to determine the retrieval rates of shortwave radiation and ET. The input variables for estimating ET derived from MODIS showed a good agreement with flux tower observations. In spite of the high accuracy of MODIS-derived input variables, MODIS ET showed meaningful errors (ME= -0.65 to -0.59 mm day-1, RMSE= 0.50 to 1.10 mm day-1). These errors were mainly associated with errors in the estimated canopy conductance. MODIS was used to calculate ET during clear sky conditions, while the regional meteorological model data provided input variables for the calculation of ET under cloudy sky conditions. The ET produced by incorporating with the regional meteorological model was evaluated by comparing ET based on MODIS, which showed a good agreement with the MODIS ET. Our preliminary results indicate that MODIS can be applied to monitor the land surface energy budget and ET with reasonable accuracy and that MODIS incorporated with meteorological model has the potential to provide reasonable input data of ET estimation under cloudy conditions.