Comparison of two landslide-prone area determination methods in Ziarat Watershed, Golestan Province

Landslide has caused great financial losses and casualties. Northern basins of Iran are prone to landslide occurrence and human interference also increased the risk of its large losses. In this study landslide-prone areas were evaluated using Bivariate Weighted AHP model and Non-wighted LRNF model in Ziarat Watershed, Gloestan Province. The landslide distribution was prepared through field survey. According to literature review and watershed conditions four factors of distance from fault, distance from drainage, rainfall and geology, were selected as factors affecting landslide and hazard maps for five classes were prepared. The results showed that the bivariate statistical model assigned 4.4% of the area prone to severe landslide hazard, and 14.5% was located in very low hazard class. While, the non-weighted LNRF model explicate 14.6% of the area in high hazard, and 8.8% of the area assigned as very low hazard class. In order to compare the hazard classes the Density Ratio (DR) was calculated in both landslide hazard maps. According to the results, the hazard classes were increased in high hazard levels. Thus, the employed models can determine the hazard classes in an acceptable manner. The validity of the bivariate weighted AHP model results were 0.81 and 0.96 based on the quality sum (Qs) and the precision index (P) indices, respectively. Also the value of Qs and P were 0.27 and 0.89 for non-weighted LRNF model. Based on the results, the weighted AHP model model is introduced as superior.