Comparison of environmental controls on daily actual evapotranspiration dynamics among different terrestrial ecosystems in China

Abstract

Terrestrial actual evapotranspiration (ETa) is a key state variable modulating land surface processes; however, it remains challenging to recognize how daily ETa responses to environmental variables vary with ecosystems, which is relevant for understanding various ecosystem processes. To this end, comprehensive datasets were synthesized for 22 field sites equipped with eddy covariance towers from the ChinaFLUX network and the National Tibetan Plateau Data Center, which included four different ecosystems across China (e.g., barren or sparsely vegetated-BAR, grassland, forest, and cropland ecosystems). The impacts of environmental variables on daily ETa variations at each field site were assessed by the boosted regression tree (BRT) method. Overall, the BRT results showed that invariant of ecosystem types, net radiation (Rn) was the primary control on daily ETa dynamics with an average contribution of 62.9 % for all the sites, followed by leaf area index-LAI (18.4 %), vapor pressure deficit (8.2 %), soil water content-SWC (7.4 %), and wind speed (3.1 %). More importantly, it was revealed that the interactions of daily ETa with energy (Rn) and water (SWC) supplies could be significantly modified by vegetation through various regulation mechanisms. Specifically, for the BAR and grassland ecosystems, the influences of vegetation on daily ETa highly depended on local land cover and hydrological conditions (e.g., groundwater). For the forest ecosystem, the contrasting impacts of LAI between evergreen and temperate forests were largely attributed to the differences in LAI dynamics over growing seasons, while physiological differences in crops affected the daily ETa responses to LAI variations at cropland sites. This study underscored the varying interactions of daily ETa with Rn and SWC, which could be further shaped by vegetation through a variety of regulation mechanisms across different ecosystems.