Landslide susceptibility mapping by use of Weight of Evidence (WofE) and Frequency Ratio (FR) and Dempster-Shafer (DSH) models: A case study of Sari-Kiasar region, Northern Iran

Landslide is one of the major natural hazards caused financial losses, in lives and destruction of natural resources each year. The aim of this study was comparisons of three models, namely WofE, FR and DSH to the determination of the landslide prone areas in Sari-Kiasar watershed. In the first, 105 landslides occurred in the study area were collected based on aerial photographs in the 1:25,000 scale and field studies divided into two group haphazardly to generate training 75% and testing 25% dataset. Then, 17 landslide conditioning factors including geological, geomorphological, hydrological and anthropogenic were prepared to spatial relationship with landslide occurrence in the study area. The most important factors in the occurrence of landslides in the study area were rainfall followed by slope and vegetation. The validation results as a percentage of the cumulative area under the curve (AUC) showed that the success rate of WofE, FR and DSH models are 92.05, 92.05 and 91.31 percent respectively and the prediction rate are 92.72, 92.73 and 85.44 percent respectively. The results show that in terms of the accuracy of the model used to base on success rate, three models are placed in excellent group (0.9 to 1), also in terms of the accuracy of the model used to base on prediction rate, WofE, FR models are placed in excellent group (0.9 to 1) and DSH is placed in good group (0.8 to 0.9). The results showed that the WofE and FR model have a higher prediction accuracy than of DSH model.