Evaluation of evapotranspiration methods for model validation in a semi-arid watershed in northern China

This study evaluates the performance of four evapotranspiration methods (Priestley-Taylor, Penman-Monteith, Hargreaves and Makkink) of differing complexity in a semiarid environment in north China. The results are compared to observed water vapour fluxes derived from eddy flux measurements. The analysis became necessary after discharge simulations using an automatically calibrated version of the Soil and Water Assessment Tool (SWAT) failed to reproduce runoff measurements. Although the study area receives most of the annual rainfall during the vegetation period, high temperatures can cause water scarcity. We investigate which evapotranspiration method is most suitable for this environment and whether the model performance of SWAT can be improved with the most adequate evapotranspiration method. The evapotranspiration models were tested in two consecutive years with different rainfall amounts. In general, the simple Hargreaves and Makkink equations outmatch the more complex Priestley-Taylor and Penman-Monteith methods, although their performance depended on water availability. Effects on the quality of SWAT runoff simulations, however, remained minor. Although evapotranspiration is an important process in the hydrology of this steppe environment, our analysis indicates that other driving factors still need to be identified to improve SWAT simulations.