Modelling and Discussion of Landslide Spatial Patterns and Hydrological Controls in the Southern Babor Mountains

Abstract

Landslide susceptibility assessment is critical in the mountainous terrains of northern Algeria, where weak marly formations, active tectonics, and seasonally concentrated rainfall converge to produce recurrent slope instability. This study develops a locally calibrated susceptibility assessment for the Amoucha sector (≈260 km2) in the Babor range, based on a dedicated inventory of 50 landslides compiled from multi-temporal satellite imagery (2010–2025) and field surveys (2024–2025). Nine conditioning factors (elevation, slope, aspect, lithology, rainfall, distance to watercourses, faults, and roads, and land cover) were integrated within a GIS framework at 30 m resolution. Frequency Ratio (FR) and Information Value (IV) models were applied and validated using an independent subset of 15 landslides. The IV model achieved an AUC of 0.835 compared with 0.823 for FR, indicating comparable but slightly superior discrimination. Marls and clays yielded the highest positive association (FR = 1.23; IV = 0.207), whereas limestone and dolomite were underrepresented (FR = 0.34; IV = −1.079). The strongest hydrological signal occurred within 150 m of drainage channels (FR up to 1.54), while annual rainfall showed the weakest association, reflecting the spatial coincidence of the wettest sector with resistant dolomitic formations. Pixel-wise correlation between the two susceptibility surfaces was r = 0.87. The resulting maps identify central and southern sectors as most susceptible and offer a spatial basis for land-use planning, geotechnical investigation, and monitoring priorities within Algeria’s existing risk management framework.