GIS-based landslide susceptibility mapping using hybrid MCDM models

Abstract

Landslide susceptibility mapping plays an important role in integrated watershed management planning, especially in the field of land-use planning in landslide-prone areas. This study aims to evaluate the performance of widely used hybrid multi-criteria decision-making (MCDM) models including the integrated index method (IIM), AHP-TOPSIS, and AHP-VIKOR to produce landslide susceptibility maps (LSMs) in the Alamut watershed, Iran. Ten causal layers including slope angle, slope aspect, rainfall, lithology, altitude, land use, and the most common distance maps including distance to fault, distance to stream, and distance to road were used to produce LSMs. Meanwhile, the landslide distribution map was used to evaluate the accuracy of the produced LSMs. The weights of the causal factors were also calculated using the analytic hierarchy process (AHP) method based on the viewpoints of 14 experts. Finally, the quality sum (Q_s) index as well as the receiver operating characteristic (ROC) curves were used to validate the performance of the MCDM models. The Q_s values for AHP-TOPSIS, IIM, and AHP-VIKOR were calculated 0.355, 0.903, and 0.703, respectively. Therefore, IIM and AHP-VIKOR methods are more efficient than AHP-TOPSIS to create LSM. The area under the curve (AUC) values of the ROC curves for AHP-TOPSIS, IIM, and AHP-VIKOR were calculated 0.784, 0.853, and 0.844, respectively. Therefore, the accuracy of the MCDM models is acceptable for TOPSIS and excellent for IIM and AHP-VIKOR. In general, results of the performance assessment of MCDM models indicated the good performance of the models to produce LSMs, so that the IIM was introduced as the best model. Also, the performance of IIM and AHP-VIKOR are more than the AHP-TOPSIS model to produce LSMs. Overall, the application of the AHP method in combination with other MCDM models as a hybrid model to produce LSMs is recommended under the principles of group decision-making in AHP for land-use planning and landslide risk management.