مدل سازی و ارزیابی پتانسیل حساسیت اراضی نسبت به لغزش با استفاده از مدل های احتمالاتی آنتروپی شانون و وزن شاهد تئوری بیزین‌ (مطالعه موردی: حوضه سرخون کارون)

Landslide susceptibility mapping is essential for land use planning and decision-making especially in the mountainous areas. The main objective of present research is to produce landslide hazard zonation at Sarkhoun basin in Karoon basin using two statistical models such as an index of Shannon’s entropy and weight of evidence and to assess the obtained results. At the first stage, landslide locations were identified in the study area by interpretation of aerial photographs, image sattellites, and from field investigations and then landslide inventory map was created for study area. The landslide conditioning factors such as slope degree, slope aspect, altitude, lithology, land use,distance of road, distance of fault, distance of drainage, topographic wetness index (TWI), Convergence Index, and precipitation were extracted from the spatial database and they were digitized in GIS environment. With integrated variables, landslides were calculated in each variable class and ‎weighted in index of entropy and weight of evidence model. In the last, landslide hazard zonation map ‎were obtained with both of models. The results of landslide susceptibility mapps of both statistical models were indicated more than 70 percent the occurred landslides were located in very high and high zones that about half of the basin area (over 45 percent) constitute. Also, the results of both models together were revealed that land use, has the greatest impact on the occurred landslides. Resolution of the zones, based on the seed cell area index (SCAI) and frequency ratio (FR) were evaluated suitable for both statistical models. Finally, the ROC (receiver operating characteristic) curves for landslide susceptibility maps were drawn and the areas under the curve (AUC) were calculated. The verification results of both models were evaluated very well and showed that the index of entropy model (AUC=89%) performed slightly better than weight of evidence model (AUC=82%).