| Abstrakt: |
It is important effective factors prioritization, landslide susceptibility zonation, modeling and predicting in environmental decision-making and planning in order to prevent the destruction and protection of natural resources in mountainous areas, specially. The aim of this study is to prioritize effective factors, landslide susceptibility zonation assessment using two-variable probability methods of maximum entropy (MaxEnt) and dempster shafer in Doab Samsami watershed of Chaharmahal and Bakhtiari province. For this purpose, 15 factor maps affecting landslide occurrence as independent variables and landslide distribution map as a dependent variable were prepared using existing maps, ALOS digital elevation model, Landsat 8 satellite images and field survey. After preparing the required geomorphometric maps in SAGA GIS®7.7.1 software, classification of independent variables was performed using frequency ratio index (FR) and landslide distribution map in the environment ArcGIS® 10.8 and they were weighed or quantitated them. In order to implementation and validation of models, landslide distribution data were randomly divided into two categories of training and test data in the proportion of 70 and 30%, respectively. Using 15 effective factors and landslide distribution map, Max Entropy (MAXENT) and Dempster Shaffer models are performed and landslide susceptibility zonation maps are prepared and each model is divided into five very low, low, moderate, high and very high. In order to evaluate the classification accuracy and validation of the models, the frequency ratio and seed cell area index (FR&SCAI) and the area under receiver characteristic curve (AUC-ROC) were used, respectively. According to the results of the maximum entropy model, annual precipitation factors, lithology, distance to road and drainage, topographic position index, topographic wetness index, convergence index, slope gradient, stream power index, distance to fault, slope length, Convexity of land surface, slope direction, surface texture, and land use are important in landslide occurrence, respectively. According to landslide susceptibility zonation maps in both models, more than 50% of landslides occurred in high and very high susceptibility categories. Finally, the validation results of the models showed that the Demester shafer model with AUC-ROC index of 0.95 and classification accuracy with higher FR & SCAI index, greater efficiency and desirability for zoning, modeling and landslide prediction in the study area. [ABSTRACT FROM AUTHOR] |
