Deriving a high-quality daily dataset of large-pan evaporation over China using a hybrid model

Global warming is expected to increase the atmospheric evaporative demand and make more surface water for evapotranspiration, aggerating water sources' social and ecological shortage. Pan evaporation, as a routine observation worldwide, is an excellent metric to indicate the response of terrestrial evaporation to global warming. However, several non-climatic effects, such as instrument upgrades, have destroyed the homogenization of pan evaporation and limited its applications. In China, 2400s meteorological stations have observed daily pan evaporation since 1951. The observed records became discontinuous and inconsistent due to the instrument upgrade from micro-pan D20 to large-pan E601. Here, combining the Penpan model (PM) and random forest model (RFM), we developed a hybrid model to assimilate different types of pan evaporation into a consistent dataset. Based on the cross-validation test, on a daily scale, the hybrid model has a lower bias (RMSE=0.41 mm day^-1) and better stability (NSE=0.94) than the two sub-models and the conversion coefficient method. Finally, we produced a homogenized daily dataset of E601 across China from 1961 to 2018. Based on this dataset, we analyzed the long-term trend of pan evaporation. Pan evaporation showed a -1.23±0.57 mm a^-2 downward trend from 1961-1993, primarily caused by decreased pan evaporation in warm seasons over North China. After 1993, the pan evaporation in South China increased significantly, resulting in a 1.83±0.87 mm a^-2 upward trend across China. With better homogeneity and higher temporal resolution, the new dataset is expected to promote drought monitoring, hydrological modeling, and water resources management. Free access to the dataset can be found at https://figshare.com/s/0cdbd6b1dbf1e22d757e.