Multi-source Rainfall Erosivity Fusion based on Extra Tree Regression

The Chinese Mainland annual rainfall erosivity fusion grid dataset is a 38 year long-term rainfall erosivity dataset, which aims to promote the development of soil erosion and hydrological modeling. Its records began in 1983, with a spatial resolution of 0.25 °× 0.25 °. Using data from over 500 sites in China as reference data, several common grid datasets were fused based on machine learning models (Extra Tree Regression), including remote sensing data (CHIRPS, CPC), interpolation data based on sites (CN05.1, CHM), and reanalysis data (ERA5)Annual rainfall erosivity obtained by using ETR model to fuse CHIRPS, CPC, CN05.1, CHM, ERA5 datasets RE_yyyy represents the RE of the year yyyy Moderate RE represents the RE calculated from precipitation data with a daily rainfall of 10-24.9mm Large RE represents the RE calculated from precipitation data with a daily rainfall of 25-49.9mm Heavy RE represents the RE calculated from precipitation data with a daily rainfall of 50mm or more