Fast physically-based probabilistic modelling of rainfall-induced shallow landslide susceptibility at the regional scale considering geotechnical uncertainties and different hydrological conditions

The inherent uncertainty in hydro-geotechnical parameters presents a significant challenge for accurately predicting rainfall-triggered shallow landslides in mountainous regions. In this study, a novel probabilistic framework was developed and implemented in the “ Py.GIS-FSLAM-FORM ” software, designed to address the complexities associated with parameter uncertainty, correlation, and distribution. By combining the Fast Shallow Landslide Assessment Model (FSLAM) with the First-Order Reliability Method (FORM), we have enhanced the traditional probabilistic approach to create more accurate landslide susceptibility maps. This study emphasizes the uncertainly of geotechnical parameters and the critical influence of hydrological conditions on landslide susceptibility, especially focusing on the interaction between antecedent recharge ( qa ) and event rainfall ( Pe ). In our study area (Val d’Aran, Spain), the probabilistically based results revealed that areas of very high susceptibility are significantly affected by event rainfall, particularly on slopes of 30–40 degrees and aspects between 100 and 250 degrees. The variability in geotechnical parameters, especially the coefficient of variation (COV) in cohesion and friction angle, plays a crucial role in landslide susceptibility assessment, with increased COVs leading to greater landslide uncertainty. Additionally, cross-negative correlations and non-normal distributions of geotechnical parameters substantially influence the spatial distribution of landslides, notably when combining antecedent recharge with event rainfall. These results highlight the importance of incorporating parameter variability and hydrological conditions in susceptibility models to improve the precision of regional landslide forecasts. While the study was performed in Val d'Aran, its methodologies and conclusions are relevant to mountainous areas worldwide, offering insights for refining landslide prediction models and susceptibility assessments, contributing to global efforts in landslide disaster prevention.