Landslide spatial modeling using a bivariate statistical method in Kermanshah Province, Iran

Abstract Landslides are dangerous natural hazards in Iran. The purpose of this study is to map landslide susceptibility using a bivariate statistical model, known as the frequency ratio (FR) model. Kermanshah Province, located in western Iran, was chosen as the study area. In the primary section, an inventory map was prepared using 115 landslide positions. Seventy percent of landslide points were selected for the training model and the remaining 30% were applied for validation purposes. In the next section, 10 conditioning factors such as altitude, slope degree, aspect, profile curvature, distance to roads, distance to fault, distance to river, lithology, land use, and annual mean rainfall were considered because of the study site features, data access, and previous studies. The FR model was employed to create a landslide susceptibility map (LSM). Finally, the efficacy of the model was assessed using the “ROC” curve. The “AUC” for the susceptibility map was reported as 90.3%. The results illustrated that 24.1%, 33.7%, 22.5%, 13.6%, and 6.1% of the study area in the susceptibility classes of the “very low, low, moderate, high, and very high,” respectively. Landslide spatial modeling can be applied by authorities to manage and decrease landslide occurrence.