Abstract: Objective High—resolution precipitation data are a fundamental basis for the fine—scale estimation of rainfall erosivity in complex terrain regions. Owing to the sparse distribution of meteorological stations over the Tibetan Plateau and the coarse spatial resolution of satellite—based precipitation products, the characterization of rainfall erosivity remains insufficient. Therefore, it is necessary to develop high—resolution precipitation data to improve the accuracy of regional rainfall erosivity estimation. Methods Based on the GPM IMERG monthly precipitation product for 2018–2020, multi—source environmental variables were incorporated, and a random forest model was employed to achieve 1 km spatial downscaling. Subsequently, the monthly rainfall erosivity factor (R) was calculated, and its spatiotemporal distribution characteristics as well as differences under different precipitation—driven datasets were analyzed. Results 1) The RF downscaling model effectively reconstructed the spatial distribution of monthly precipitation over the Tibetan Plateau. For the independent test set, the coefficient of determination (R²) ranged from 0.73 to 0.82, the root mean square error ranged from 25.8 to 35.3 mm in each month, the mean absolute error (MAE) ranged from 11.9 to 15.8 mm in each month , and the Kling-Gupta Efficiency (KGE) ranged from 0.78 to 0.85. 2) From 2018 to 2020, the R factor over the Tibetan Plateau exhibited a spatial pattern of decreasing values from the southeast to the northwest. High-value areas were mainly distributed along the southern margin, southeastern margin, and local eastern regions of the plateau, whereas the central—western and northwestern regions were characterized by low values. Temporally, the R factor showed a pattern of high values in summer and low values in winter. The interannual spatial pattern was generally stable, although the extent and intensity of summer high-value areas varied among years. 3) Compared with the original GPM IMERG data, the downscaled results showed generally consistent intra—annual variation trends, but differences were observed in peak intensity, transitional months, and local spatial details. Specifically, the downscaled R factor was generally lower in July and relatively higher in April–May and September, while its spatial structure and gradient expression were more detailed. ConclusionsThe downscaled monthly precipitation data can effectively improve the spatial representation of rainfall erosivity in complex terrain regions and enhance the depiction of local-scale heterogeneity. These results provide reliable data support for the fine—scale estimation of rainfall erosivity over the Tibetan Plateau and for subsequent research in soil and water conservation.