Spatial variation in annual actual evapotranspiration of terrestrial ecosystems in China: Results from eddy covariance measurements

Abstract

Understanding the spatial variation in annual actual evapotranspiration (AET) and its influencing factors is crucial for a better understanding of hydrological processes and water resources management. By synthesizing ecosystem-level observations of eddy-covariance flux sites in China (a total of 61 sites), we constructed the most complete AET dataset in China up to now. Based on this dataset, we quantified the statistic characteristics of AET and water budgets (defined as the ratio of AET to annual mean precipitation (MAP), AET/MAP) of terrestrial ecosystems in China. Results showed that AET differed significantly among both different vegetation types and climate types in China, with overall mean AET of 534.7±232.8 mm yr-1. AET/MAP also differed significantly among different climate types, but there were no distinct differences in AET/MAP values across vegetation types, with mean AET/MAP of 0.82±0.28 for non-irrigated ecosystems. We further investigated how the main climatic factors and vegetation attributes control the spatial variation in AET. Our findings revealed that the spatial variation of AET in China was closely correlated with the geographical patterns of climate and vegetation, in which the effects of total annual net radiation (R n), MAP and mean annual air temperature (MAT) were dominant. Thus, we proposed an empirical equation to describe the spatial patterns of AET in China, which could explain about 84% of the spatial variation in AET of terrestrial ecosystems in China. Based on the constructed dataset, we also evaluated the uncertainties of five published global evapotranspiration products in simulating site-specific AET in China. Results showed that large biases in site-specific AET values existed for all five global evapotranspiration products, which indicated that it is necessary to involve more observation data of China in their parameterization or validation, while our AET dataset would provide a data source for it.