Soil erosion represents a severe risk to global food security and ecosystem services, with rainfall erosivity serving as the key driver of hydrological erosion. Under climate change, rainfall erosivity exhibits high spatiotemporal heterogeneity and dynamics, making it crucial to accurately capture its long-term spatiotemporal dynamics globally. Therefore, this study integrates high-temporal-resolution satellite data with station observation data to establish a global, long-term, climate-zone-specific framework for monthly rainfall erosivity simulation. Monthly rainfall erosivity (2001-2020) was calculated using GPM IMERG 30-minute data and calibrated. Subsequently, using rainfall characteristics from the MSWEP dataset, the XGBoost algorithm was employed to construct monthly rainfall erosivity models for each month and climate zone. Finally, the model was used to reconstruct a 40-year global monthly rainfall erosivity (1981-2020). Results indicate: (1) The model exhibits good agreement with reference data globally and demonstrates robust cross-validation performance. (2) Between 1981 and 2020, rainfall erosivity significantly changed across 23.6% of global areas, exhibiting an annual weakening trend dominated by spring-driven reductions. Specifically, 36.92% of regions showed significant increases concentrated in low latitudes, while 63.08% experienced significant decreases primarily distributed in mid-latitude zones. (3) Rainfall erosivity exhibits a seasonal inverse trend globally, weakening in spring and strengthening in summer. The research framework proposed in this paper provides an effective approach for accurately simulating and capturing the seasonal dynamics and long-term trends of global rainfall erosivity, offering critical data to support the development of dynamic soil erosion prevention and control strategies.
Chen et al. (Mon,) studied this question.