Landslide Hazard Assessment by using Neuro-Fuzzy Technique in Vaz Watershed

In this study, Neuro-Fuzzy model was used to prepare the map of Vaz watershed landslide susceptibility in GIS environment. Location of landslides occurred in the study area was determined through interpretation of aerial photographs and the field monitoring. In the next step, factors affecting landslide occurrence such as altitude, lithology, slope, aspect, distance to drainage, distance to road, distance to fault, rainfall, and land use were digitized. Then, landslide susceptible areas were assessed by using ANFIS model based on factors affecting landslide occurrence. For this purpose, two membership functions of triangulate and Gaussian were used to prepare landslide susceptibility map and compare their results. Two statistical criteria of RMSE and MEE were employed to evaluate model performance. The results showed high efficiency of ANFIS model for preparing landslide susceptibility map and also the ANFIS model with Gaussian membership function over-performed the ANFIS model with Triangular membership function in the study area. Landslide hazard map showed that the regions with high risk have the most area indicating high risk-taking of Vaz watershed in occurrence of landslides.