Global water balance evapotranspiration (1991-2020 average) dataset for bias correction of remote sensing-based evapotranspiration

A benchmark global water balance evapotranspiration (WBET) dataset was generated using machine learning for large-area bias correction of remote sensing-based evapotranspiration (RSET). The dataset contains WBET (1991-2020 average) of over 15,000 Level 6 HydroBASINS worldwide. Additionally, a comparison of this dataset with three RSET and a machine learning-based global datasets is included. Global analysis using the Budyko curve and regional analysis with observed WBET in the conterminous United States are also included. This data release is part of a study in which we developed a method named WABE-AI (artificial intelligence-based water balance equivalence) for large-scale bias correction of RSET and implemented it on a global scale.