A Comparative Assessment between Weights-of-Evidence and Frequency Ratio Models for Landslide Hazard Zonation in Vanak Basin

In this study, Vanak catchment because of high sensitivity to landslide was selected. Then with geological, topographic maps and field survey, Landslide hazard map was prepared using GPS as dependent variables. A total of 110 landslides were mapped in GIS out of which 77 (70%) locations were chosen for the modeling purpose and the remaining 33 (30%) points were used for the model validation. Then layers of the landslide conditioning factors including slope degree, slope aspect, plan curvature, altitude, lithology, land use, distance of road, distance of fault, distance of drainage, drainage density, topographic wetness index (TWI) and Normalized Difference Vegetation Index (NDVI) Calculated. The relationship between the predisposing factors and the landslides were calculated using weights-of-evidence and Frequency Ratio Models. Finally, the susceptibility map was classified into five susceptibility classes: very low, low, moderate, high, and very high. In order to verification, the results were compared with landslides which were not used during the training of the models. Subsequently, the Receiver Operating Characteristic (ROC) curves were drawn and the area under curves (AUC) were calculated for landslide susceptibility maps. Results obtained from validation showed that AUC for Frequency Ratio and weights-of-evidence models are 0.917 (91.7%) and 0.890 (89.0%), Therefore, the results revealed that the Frequency Ratio model is more suitable than the weights-of-evidence model. Finally, verification indicates satisfactory agreement between resulted susceptibility map and existing data on landslide location.