Spatiotemporal modeling and projection framework of rainfall-induced landslide risk under climate change.

Global warming is expected to exacerbate extreme rainfall events, potentially increasing the risk of landslides. While landslides have been extensively studied, much of the focus has been on developing static frameworks for landslide susceptibility, with relatively little exploration of spatiotemporal modeling. Furthermore, previous studies have often overlooked the spatiotemporal impacts of climate change and dynamic socio-economic factors on landslide susceptibility and risk. This has resulted in a lack of understanding of how landslide risk will evolve in the future. Consequently, this study proposes a modeling approach to simulate the dynamics of landslide susceptibility and exposure population over the next 80 years. The approach involves a series of novel modeling experiments using rainfall-induced landslide data collected over the past decade in Jiangxi Province, using a GAMs that considers the effects of spatial relationships and spatio-temporal cross-validation, with an AUC of 0.885 and an error rate of 15.32%, and combining with the CMIP6 precipitation data to the direct effect of climate change on landslide susceptibility. In addition, a dynamic risk prediction model combining static and dynamic populations was developed to provide a more comprehensive understanding of future landslide impacts. This research framework serves as a scientific foundation and valuable reference for comprehending the effects of climate change on landslide hazards and their management in urban planning and risk mitigation. It aims to inform the development of more effective strategies to mitigate potential losses from future landslide risks.