Landslide susceptibility modeling in West Shoa Zone, Central Ethiopia: An integrated geospatial and statistical analysis

This study is focused on mapping landslide susceptibility in the Abuna Gindeberet area, West Shoa Zone, Central Ethiopia. The research aimed to develop susceptibility and causative factor maps using integrated geospatial and statistical analysis within a GIS framework. A total of 1222 landslides were documented, with 70% allocated for model training and 30% for validation, based on data from Google Earth images (GEIs) and field surveys. This study evaluates the susceptibility of landslides based on causative elements like slope, aspect, curvature, lithology, land use-land cover (LULC), distance to lineament, distance to stream, and rainfall. The landslide susceptibility maps (LSMs) were made by superimposing the weights of each landslide causative factor using the raster calculator of the spatial analyst tool in ArcGIS 10.8. The final LSMs were categorized into very low, low, moderate, high, and very-high susceptibility classes based on statistical approaches such as Frequency Ratio (FR) and Logistic Regression (LR) models. The LSMs from the FR model reveal that very low, low, moderate, high, and very high susceptibility classes cover 9%, 21%, 24%, 34%, and 12% of the area, respectively. The LR model shows very low, low, moderate, high, and very high classes covering 7%, 33%, 9%, 28%, and 23% of the region, respectively. The study showed that landslides are primarily associated with moderately sloped terrains, largely constituted of sandstone, and found in regions with agricultural activities and sparse forests near lineaments. Validation through the ROC-AUC indicated success rates of 0.857 for the FR model and 0.849 for the LR model.