Statistical and Physically-Based Rainfall Triggered Landslides Susceptibility Assessment in the Tagus River Left Margin Basins, Almada County (Portugal)

Abstract

The Tagus River left margin slopes (Almada County, Portugal) are rainfall triggered landslide-prone areas, which are a serious threat to life and property. To improve knowledge on this topic, a study on landslide susceptibility assessment and mapping was carried out using simple statistical and physically based methods but using improved databases on geology and land use mapping. The statistically-based information value was used to relate an inventory of landslides with the landslides predisposing factors slope angle, lithology, aspect, curvatures, Topographic Wetness Index and land use. The physically-based methods were the SHALSTAB hydrological module coupled with an infinite slope stability model complete solution. The model results were validated against the landslides inventory using ROC curves and provided values of the Area Under the Curve for the Information value (0.90) higher than those of the infinite slope solution (0.80). These differences raise questions on the influence of database and model selection on the reliability of results. The used approaches, which require a careful analysis of the input data, while providing different results, had a satisfactory global performance, being adequate for the landslide susceptibility assessment at regional scale and relevant tools for hazard prevention.