Landslides Susceptibility Zoning Using Baye's Theorem-ANP Hybrid Model (Case Study: Heyran Defile)

Landslide is one of the morphodynamic processes including significant hazards in terms of fatalities, financial casualties and the number of happening. In this research, Zoning of potential landslide occurrence is studied in Heyran Defile region. To do this the combination of multi-criteria (Analytical Network Process) and statistically (Baye's theorem) models and accompanied by 12 natural and human parameters including, Slope, aspect, land use, lithology, precipitation, vegetation density index (NDVI), slope length (LS), topographic wetness index (TWI), stream power index (SPI), distance to road, distance to fault and distance to river were used. The layer of occurred landslides in the study area have been used to the obtaining weight of each landslide susceptibility parameters classes and validation of the final map which seventy percent of the landslide for running the model and another 30 percent is used to the model validation. The result is a map classified in five categories that respectively to be included Zones with very low, Low, Moderate, High and very High potential. According to the result map 26.3 percent of the area case study has been predicted as a region with high and very high potential for the landslide occurrence. These areas primarily to be included marginal areas of the Ardebil- Astara road. Most landslides also occurred in these areas because of the high construction in bordering the road, disrupt the natural slope of the land for the road construction and broaden it. Land use in these areas is mainly sparse forest, rangelands and agriculture which is located on the slopes. Evaluation of zoning map was done using 30 percent of the occurred landslide. According to the results of this evaluation and placement of a considerable percentage of landslides in the high and very high sensitivity classes (77.6 %), it can be concluded that the accuracy of used models in the landslides susceptibility zoning is acceptable.