Landslide susceptibility mapping of southern part of Marsyangdi River basin, West Nepal using logistic regression method

Landslide susceptibility mapping has been carried out around the Sundar Bajar to Besi Shahar area in the southern part of Marsyangdi river basin covering an area of about 221 Km². Logistic regression model has been used to generate the landslide susceptibility map of the area. This method assumes that the future landslides will have the same causal factors as the landslides initiated in the past. The landslide inventory map was prepared using the Google Earth images with the field identification and verification. Thirteen intrinsic causative factors and one extrinsic factor i.e. rainfall were used in analysis. Slope, aspect, curvature, elevation and relief were the Digital Elevation Model (DEM)-derived parameter whereas the thematic layers like lithology, distance to road, distance to river, lineament density, stream density, wetness index and land-use were prepared in the different GIS environment. Receiver operating characteristic (ROC) curves in SPSS was used to validate the model using the training and validation landslide data. The area is classified into five classes i.e. very low, low, medium, high and very high susceptibility using the ROC curve of the validation data. Most of the area belongs to the very low to low susceptibility class. The prepared susceptibility map can be used for the landslide hazard management.