probabilistic landslide risk analysis and mapping (case study: chehel-chai watershed, golestan province, iran)

the efficiency of three statistical models, ahp surface-weighted density bivariate (semi-quantitative models), stepwise multivariate regression and logistic multivariate regression models were compared in chehel-chai watershed in golestan province, iran. in current study the hazard map was prepared according to the top model of landslide hazard map. chehel-chai watershed is located as one of gorganrud river sub basins in golestan province. the distribution map of the area landslide was provided using the stereoscopically interpretation of aerial photos and field observations and nine effective factors including elevation, slope, aspect, lithology, distance from fault, stream and road, land use, and precipitation rate were chosen concerning the expert view and source review. the hazard potential of landslides was also prepared using three models. the differences were compared between models of hazard classes with chi-square test, the agreement rate of risk maps with kappa index and the evaluation of the model accuracy with total quality index (qs). the risk map was provided based on the risk equation by the combination of the risk maps, the element frequency and the element vulnerability degree. the results showed that all models had 99% reliability level and there were a high separation among the risk classes. kappa index was variable between 0.0 to 0.2 representing that the correspondence between them is negligible. the weighted (ahp) bivariate statistical model was selected as the best model for the basin with qs equal to 3.62. 12.13% of the surveyed basins were located in high loss class and very high loss class, respectively. it was concluded, 41% of chehel-chai watersheds were at moderate risk and from it 13% was in high classes risk that must be considered in risk management, landslide risk and land logistics of this mountainous area.