Aims: Landslides are natural hazards that are increasingly influenced by human activities. These events increase sediment input into river systems, changing sediment transport. This study’s main methodological contribution is integrating a bivariate model (Information Value) with an expert-based approach (AHP) to account for the relative importance of conditioning factors in landslide susceptibility mapping.
Material & Methods: First, a landslide distribution map was prepared. Eleven factors, including lithology, slope angle, elevation, aspect, clay content, soil pH, distance to roads, faults, rivers, land-use, and rainfall, were incorporated into the model. The relative importance of these factors was determined using the AHP, and susceptibility maps were prepared using IV and hybrid IV-AHP models. Accuracy was assessed using the ROC curve and area under the curve.
Findings: The results identified distance to roads, slope angle, and lithology as the most conditioning factors. The AUC was 0.770 for the IV model and 0.816 for the IV-AHP model. The IV and IV-AHP models classified approximately 29% and 32% of the Firozkuh Watershed as highly and very highly susceptible, respectively.
Conclusion: The results show that a considerable proportion of the watershed is classified as highly susceptible to landslides. Both the IV and IV-AHP models demonstrated satisfactory predictive performance, supporting their application for landslide susceptibility assessment in the watershed. The main limitation of this study is the unavailability of several important conditioning factors, and future research should evaluate additional statistical and machine learning models.