融合多元不确定性的中国降雨型滑坡暴露性时空预测OA

Rainfall-induced landslides are characterized by wide spatial distribution,complex triggering mecha-nisms,and high suddenness,making accurate prediction of future landslide risk particularly challenging.The inte-gration of climate-model-derived rainfall fields—refined through spatial and temporal downscaling—with landslide susceptibility assessment and rainfall-threshold modeling provides an important technical pathway for forecasting landslide hazards.However,this process involves multiple interconnected components,including climate model simulations,bias correction and downscaling,environmental factor characterization,and socioeco-nomic data integration,all of which introduce substantial uncertainty into predictions of future landslide exposure.To address these challenges,this study develops a spatiotemporal prediction and uncertainty-analysis framework for rainfall-induced landslide exposure in China based on a CMIP6(Coupled Model Intercomparison Project Phase 6)multi-model ensemble.First,static environmental factors,such as topography,geological conditions,and land cover,are integrated to assess regional landslide susceptibility.Second,a rainfall-threshold model based on three-day accumulated precipitation is constructed to simulate the spatiotemporal evolution of landslide hazards under different future climate scenarios.Finally,by coupling projected population and economic datasets under the Shared Socioeconomic Pathways(SSPs),the framework quantifies temporal trends and spatial heterogeneity in landslide exposure.The results indicate that under SSP1-2.6,SSP2-4.5,and SSP5-8.5,population exposure to rain-fall-induced landslide hazards in China is projected to increase to 22.1%,22.5%,and 23.3%,respectively,while GDP exposure is expected to rise to 16.6%,18.6%,and 18.9%.Pronounced differences among bias-correction methods and global climate models(GCMs)are observed in the simulated magnitude of hazard and exposure changes,as well as in the identification of regional hotspot areas,highlighting the critical role of methodological and model uncertainty in landslide risk assessment.Spatially,high-hazard areas exhibit a clear expansion toward southwestern mountainous regions and the hilly areas of South China.Both population and GDP(Gross Demestic Product)exposure display a distinct"high in the southeast and low in the northwest"pattern,with strong cluste-ring along major urban and economic corridors.By explicitly incorporating uncertainty from climate models,bias-correction techniques,and socioeconomic scenarios,this study quantitatively reveals the spatiotemporal evolution and hotspot migration of rainfall-induced landslide exposure in China.The proposed framework provides a scien-tific basis for regional landslide risk management and supports the development of targeted disaster-prevention and mitigation strategies under future climate change.