Objectives: To locate areas most susceptible to landslides within the Mamit district of Mizoram, India, by combining Geographic Information System workflows with the Analytic Hierarchy Process. Methods: Twelve predictors were assembled and processed in ArcGIS 10.8, then weighted via Saaty’s pairwise comparisons. Factor sensitivity was tested by using map removal method. Findings: Integrating the weights produced a susceptibility map in which approximately 33 percent of the district was classed into ‘high’ or ‘very-high’ susceptibility classes. To validate this, we compared the map with 315 past landslides using ROC analysis, which provides an AUC of 0.903, which is generally considered excellent. Despite the AHP expert-based model having identified natural variables such as rainfall as the main influential factor; a map removal sensitivity study revealed that the anthropogenic factor of distance to roads was the most influential predictor of landslide locations. The landslide inventory was highly supportive of this data-driven finding, as 68 percent of the past landslides happened within 200m of the road. Novelty: This study is the first validated AHP-based susceptibility map for the Mamit District, with accuracy suitable for district-scale planning. Keywords: Analytical Hierarchy Process, Geographic Information Systems, Landslide Susceptibility, Mamit District, Mizoram
Lalhruaitluanga et al. (Fri,) studied this question.