A survey on rock fall hazard mapping by logistic reglession in mountainous road of Kurdistan province.

Rock falls are one of the speed movements in dry and continuous material. They are natural process of activities geomorphic on steep slope in mountainous terrains. At this research، Rock fall Hazard Mapping was generated by Logistic Regression، Frequency Ratio and Overlay Index models at a long 11 kilometer of the Salavat Abad road in eastern Sanandaj، Kurdistan، Iran. Dependent variables is occurrence and non-occurrence rock falls and independent variables including; slope degree، slope aspect، slope curvature، elevation of sea، lithology، distance from road، distance from fault، land use. The maps which were generated with three models، compared and verified with rockfalls situation using Success Rate Curve (SRC). Results showed that the logistic regression model، Frequency Ratio model and Overlay Index model have an accuracy of 85. 09، 76. 76 and 81. 44 percent of the Area under Curve (AUC)، respectively. So، Logistic Regression is more proper than the other models for determining regions prone to rock fall in the case study.