Aims The generalized complementary principle of evapotranspiration is one of the important methods to estimate evapotranspiration when the observed data are scarce. In implementing this method, an accurate estimation of parameter α_e is critical. The temporal and spatial variation of α_e and the applicability of different methods for calculating α_e were investigated at eight flux stations under different climatic conditions and ecosystem types in China.
Methods Firstly, the annual and monthly values of α_e were calibrated based on the measured data. The spatiotemporal variability of α_e was investigated and the influence of α_e with different temporal scales on the calculation accuracy of the generalized complementarity principle model were compared. Considering that α_e can not be calibrated without measured evapotranspiration data, the applicability of two statistical models of annual α_e values based on aridity index (AI)(Liu method and Brutsaert method) were evaluated to determine whether α_e can be determined using AI. Finally, the error sources of each calculation method were analyzed.
Important findings α_e value varies with season, and the monthly variations of α_e differ among different flux stations. In terms of spatial variation, the annual values of α_e at humid sites were larger than those at arid sites. The α_e calculated by Liu method and Brutsaert method were close to the calibrated values. In applying the generalized complementary principle model, high simulation accuracy can be achieved by using the calibrated annual α_e, and the accuracy can be further improved by using the monthly α_e. Two AI-based methods also achieved accurate simulation results, which further confirmed the potential of predicting α_e based on AI in the absence of observed data. The generalized complementary principle model can simulate the annual variation trend of evapotranspiration when using annual α_e, but the estimated value were biased in some months. The evapotranspiration calculated by Liu method and Brutsaert method were underestimated in summer months of the drought sites, which may be caused by the fact that the AI was overestimated in summer months when rainfall was concentrated. The results further demonstrate the potential of the generalized complementary principle in estimating evapotranspiration in a wide range of natural environments.