Landslide susceptibility mapping using fuzzy-analytic network process

In recent decades, with the expansion of human activities on the natural environment, a variety of hazards, including landslides, have had serious human and financial damage. As a result, landslide susceptibility assessment and identifying the critical areas for watershed protection seems to be necessary. In this study, landslide susceptibility using combined fuzzy and Analytic Network Process (FANP) methods has been modelled on the Farub Roman basin. To achieve this goal, four clusters; topography, biological, hydro-climate and geological and criteria such as elevation, slope, aspect, curvature, distance from roads, land use, normalized difference vegetation index (NDVI), distance from rivers, drainage density, rainfall, soil moisture index, distance from faults and lithology have been considered. The results showed that the Fuzzy-analytic network process model is appropriate for landslide susceptibility modelling for as much as in model validation through the Relative Operating Characteristic (ROC) curves, the AUC, 0.83 has achieved with the standard error; 0.07, of the P-value equal to zero. For assessing the landslide susceptibility in the Farub Roman basin based on the results of the ROC curve, fuzzy ANP model evaluated very well. In addition, the results showed that, 66% of the known landslides have been found in areas with high and very high sensitivity. Due to the estimation of the high and very high sensitivity of landslides; 51% of the total area, the implementation of studying watershed protection seems to be necessary.