Monitoring daily evapotranspiration in Northeast Asia using MODIS and a regional Land Data Assimilation System

We applied an approach for daily estimation and monitoring of evapotranspiration (ET) over the Northeast Asia monsoon region using satellite remote sensing observations from the Moderate Resolution Imaging Spectroradiometer (MODIS). Frequent cloud cover results in a substantial loss of remote sensing information, limiting the capability of continuous ET monitoring for the monsoon region. Accordingly, we applied and evaluated a stand-alone MODIS ET algorithm for representative regional ecosystem types and an alternative algorithm to facilitate continuous regional ET estimates using surface meteorological inputs from the Korea Land Data Assimilation System (KLDAS) in addition to MODIS land products. The resulting ET calculations showed generally favorable agreement (root-mean-square error <1.3mmd(-1)) with respect to in situ measurements from eight regional flux tower sites. The estimated mean annual ET for 3 years (2006 to 2008) was approximately 362.0161.5mmyr(-1) over the Northeast Asia domain. In general, the MODIS and KLDAS-based ET (MODIS-KLDAS ET) results showed favorable performance when compared to tower observations, though the results were overestimated for a forest site by approximately 39.5% and underestimated for a cropland site in South Korea by 0.8%. The MODIS-KLDAS ET data were generally underestimated relative to the MODIS (MOD16) operational global terrestrial ET product for various biome types, excluding cropland; however, MODIS-KLDAS ET showed better agreement than MOD16 ET for forest and cropland sites in South Korea. Our results indicate that MODIS ET estimates are feasible but are limited by satellite optical-infrared remote sensing constraints over cloudy regions, whereas alternative ET estimates using continuous meteorological inputs from operational regional climate systems (e.g., KLDAS) provide accurate ET results and continuous monitoring capability under all-sky conditions.