An Assessment of the Landslide Susceptibility Prediction Models in the Bar Watershed- Neyshabur

Landslide is one of the most destructive types of erosion of slopes, causing substantial financial losses.  Identification of causative factors in the landslide occurrence and providing a zoning map of the susceptible areas is one of the basic tools for minimizing the possible damages. In this research, the landslide susceptibility map in the Neyshabur Watershed was prepared using three algorithm including, Support Vector Machine, Maximum Entropy, and Random Forest Algorithm. A map of landslide distribution was prepared along with 12 thematic layers including slope, aspect, plan curvature, profile curvature, elevation, land use, geology, distance from the road, distance from the rivers, distance from the fault, topographic wetness index, and drainage density in the GIS environment. The landslide susceptibility map of the studied area was prepared using three methods of random forest algorithm, maximum entropy, and support vector machine algorithm, and using Receiver Operating Characteristics and 30% of unused landslide points in the modeling process. The results of an assessment of the models indicated that the accuracy of the estimated maps prepared by the Support Vector Machine, Maximum Entropy, and Random Forest Algorithm were 86, 75, and 82 percent, respectively. Therefore, it can be stated that the presented maps can play an important role in identifying the slide-prone areas as well as in the implementation of development plans, especially road construction in the studied area. Given the potential of tourism in the catchment area, it is necessary to pay attention to the landslide potential in the catchment.