Spatio-Temporal Model Scheme for Shallow Landslide Occurrence: Example of Tidone Catchment

Shallow landslides represent a critical natural hazard. Areas with complex geological, hydrological, and geomorphological conditions are more prone to this natural hazard. This study focuses on the Tidone catchment within the Oltrepò Pavese region of Italy, where shallow landslides have caused serious damages. The objective of this research is to develop a data-driven spatio-temporal model for prediction of shallow landslide occurrence. The model integrates multiple geospatial and temporal variables, consisting of rainfall intensity, soil saturation, slope angle, land use, and geological characteristics, utilising a Multivariate Adaptive Regression Spline (MARS) framework to estimate both spatial and temporal probability of the occurrence of shallow landslides. Spatial analysis was conducted using eleven predisposing factors, nine factors derived from Digital Elevation Models (DEM), two categorical factors, land use and bedrock geology, while temporal factors were developed through rainfall data obtained from meteorological stations and soil saturation degree, incorporating ERA5-Land satellite products. The findings highlight critical areas in a daily time frame prone to landslides, offering valuable results for disaster mitigation and land-use planning. While the model shows a strong predictive capability, limitations include the dependence on high-resolution data and constraints of the timeframes. Future research may focus on refining the model and application of that for risk analysis.