Landslide Hazard Zoning Using Bayesian Theory

The aim of present research is landslide hazard zoning using Bayesian theory in a part of Golestan province. For this purpose, landslides inventory map was created by landslide locations of landslide database (392 landslide locations). Then, the maps of effective parameters in landslide such as slope degree, aspect, altitude, slope curvature, geology, land use, distance of drainage, distance of road, distance of fault, stream power index (SPI), sediment transport index (STI), and rainfall were prepared in GIS environment. Relationship between effective factors and landslide locations were considered using Bayesian probability theory. In the next step, parameters classes weights were found and the landslide susceptibility mapping was achieved by fourteen modeling approaches (using whole parameters and deleting parameters one by one). The verification results by ROC curve and 30% landslide locations showed that the Bayesian probability model has 71.37% accuracy for the second approach of modeling in the study area.