Comparison Between Selected Experimental Methods And Statistical And Artificial Neural Network For Landslide Hazard Zonation Case Study: Behesht Abad Dam Reservoir

Abstract

In order to decrease the destructions due to landslides, it’s important and unavoidable to recognize and to map the hazard zonations. For this, different mehods are utilized by researchers in other countries with specific conditions. In this paper, landslide inventory map has been prepared and then the effective parameters on the landslides in the study area have been investigated. Finally, some empirical methods such as Mora-Varson and Nilson methods with bivariate Statistical and Artificial Neural Network(ANN) methods are selected by using comparison of various methods between original locations and this study area in Behesht Abad Dam reservoir. <br />In consequence of landslide hazard zonation mapping by above mentioned methods, some relations including empirical Probability Factor(P), Landslide Index(Li) and Reciever Operating Characteristic(ROC) Curves are used to evaluate the accuracy of each method. Finally, the results of ROC curves and calculation of Area Under ROC Curve(AUC) are based for evaluation of accuracy. Therefore, Artificial Neural Network and Statistical methods are selected to provide suitable maps in this area.