Landslide susceptibility mapping by integrating analytical hierarchy process, frequency ratio, and fuzzy gamma operator models, case study: North of Lorestan Province, Iran

Identifying landslide-prone areas is an essential step in assessing landslide risk and reducing landslide damage. In this paper, GIS-based spatial analysis has been used to prepare the landslide susceptibility (LS) map in the north of Lorestan province in western Iran. For this purpose, three main criteria and their sub-criteria were identified as causative factors including geology and topography (i.e., distance from the fault, lithology, slope, aspect, and elevation), climate (i.e., rainfall and distance from the river), and environmental parameters (i.e., distance from the road, land-cover, NDVI). One hundred thirty-six known landslides were randomly divided into training ([Formula: see text] 70%) and validation ([Formula: see text] 30%) datasets. This study is based on the integration of popular analytic hierarchy process (AHP), frequency ratio (FR), and the fuzzy gamma operator (FGO) techniques. AHP was utilized to prioritize causal factors and fuzzy technique was applied in two stages of factor map fuzzification and calculation of sub-criteria maps and then overlap of fuzzified map layers. The fuzzy membership (FM) values were determined based on the FR method, which was normalized between the ranges of 0 and 1. Finally, LS zoning maps were estimated in five susceptibility classes (very low, low, moderate, high, and very high). Validation processes by comparing the three output maps with the layer of validation landslides in the study area and area under receiver operating characteristic curve confirm that the gamma value of 0.9 (AUC = 0.88) offers a more accurate LS map compared to other gamma values. The results of this study will be reliable for landslide risk reduction strategies.