Climate change-induced shifts in landslide susceptibility in São Sebastião (southeastern Brazil)

Landslides are a pressing natural hazard, particularly in regions prone to extreme weather events, and their frequency is expected to rise due to climate change. This paper investigates landslide susceptibility in São Sebastião, a coastal region in southeastern Brazil, under various climate change scenarios. The study fills a critical gap in understanding how future precipitation changes driven by climate models could affect the area's susceptibility to landslides. Current assessments often overlook the combined effects of environmental variables and land-use dynamics under future climate conditions. To bridge this gap, this research integrates environmental variables, including Soil Moisture Index (SMI), slope degree, saturation, relief dissection, geomorphology, geology, and topographic position index (TPI), with land use and land cover (LULC) data. Scenarios from the Intergovernmental Panel on Climate Change (IPCC) for RCP2.6, RCP4.5, RCP6.0, and RCP8.5 CMIP5 (Climate Models Intercomparison Programme Version 5) models were applied to model the impact of changing precipitation patterns on landslide susceptibility. Using geospatial data and a weighted sum model, susceptibility maps were developed for each climate scenario and validated with a landslide inventory and receiver operating characteristic (ROC) analysis. The findings indicate a notable shift in landslide risk, with scenarios RCP6.0 and RCP8.5 showing significant increases in moderately susceptible areas due to higher precipitation intensities. Frequency Ratio (FR) analysis revealed varying levels of landslide susceptibility across scenarios, with RCP2.6 showing lower probabilities for moderate landslides (FR: 0.007946) compared to higher ratings for RCP4.5, RCP6.0, and RCP8.5 (FR: 1.663156 for high landslides). Slope and TPI emerged as the most influential variables, while land-use types, particularly urban areas and deforestation zones, showed heightened vulnerability in future scenarios.