Interannual variability of terrestrial evapotranspiration in Northeast Asia

Abstract:

Terrestrial evapotranspiration (ET) is a major component for the land surface water cycle and the energy interaction between land surface and atmosphere, and for the improvement of understandings such as the terrestrial water management as well as vegetation growth. Satellite remote sensing provides a promising opportunity to quantify the magnitude and variability of ET at the regional scale. This study investigated the interannual variability of ET in Northeast Asian regions containing Korea Peninsula, China, Mongolia, and Japan. The regional daily ET was estimated using various satellite remote sensing data from 2003 to 2010. Satellite-based daily ET calculations showed generally favorable agreement (RMSE < 1.06 mm day^-1) with eight flux tower measurements. Annual ET showed large range in study domain from 466 in 2006 to 498 mm yr^-1 in 2008. The mean annual ET for study period was 481.1±224.6 mm yr^-1 over the Northeast Asia. ET rates were generally higher at the cropland in China, while it was lower at the grassland in Mongolia. In general, large variability of annual ET was detected in central and northern China and eastern parts of Mongolia and Russia, which are dominated by Grassland, Savanna and Shrubland classifications as well as complex terrain. The range of coefficient of variation (CV) on annual ET was from 15 to 30% at those regions. The spatial pattern of CV was similar to the ratio of ET to TRMM precipitation (ET2PRCP) for the arid and semi-arid regions represented to grassland in this domain. CV on ET was generally high when the ET2PRCP was within the range from 0.7 to 1.5. It indicates that precipitation may affect to variation of annual ET at for the arid and semi-arid regions. The results generated by this study indicated that the satellite remote sensing provides the potentials to estimate and monitor ET at the regional scale, and offers a good chance to improve our knowledge on the land surface water balance.