Spatial Prediction of Landslides Using Hybrid Multi-Criteria Decision-Making Methods: A Case Study of the Saqqez-Marivan Mountain Road in Iran

Landslides along the main roads in the mountains cause fatalities, ecosystem damage, and land degradation. This study mapped the susceptibility to landslides along the Saqqez-Marivan main road located in Kurdistan province, Iran, comparing an ensemble fuzzy logic with analytic network process (fuzzy logic-ANP; FLANP) and TOPSIS (fuzzy logic-TOPSIS; FLTOPSIS) in terms of their prediction capacity. First, 100 landslides identified through field surveys were randomly allocated to a 70% dataset and a 30% dataset, respectively, for training and validating the methods. Eleven landslide conditioning factors, including slope, aspect, elevation, lithology, land use, distance to fault, distance to a river, distance to road, soil type, curvature, and precipitation were considered. The performance of the methods was evaluated by inspecting the areas under the receiver operating curve (AUCROC). The prediction accuracies were 0.983 and 0.938, respectively, for the FLTOPSIS and FLANP methods. Our findings demonstrate that although both models are known to be promising, the FLTOPSIS method had a better capacity for predicting the susceptibility of landslides in the study area. Therefore, the susceptibility map developed through the FLTOPSIS method is suitable to inform management and planning of areas prone to landslides for land allocation and development purposes, especially in mountainous areas.

Automated summary

This study compared the predictive capacities of fuzzy logic-ANP (FLANP) and fuzzy logic-TOPSIS (FLTOPSIS) for mapping landslide susceptibility along the Saqqez-Marivan main road in Kurdistan province, Iran. The FLTOPSIS method showed better prediction accuracy with an AUCROC of 0.983 compared to 0.938 for the FLANP method. The susceptibility map developed through the FLTOPSIS method is suitable for informing management and planning in landslide-prone areas in mountainous regions.