Abstract
Aerial photographs and GIS were applied for assessing landslide hazard, producing landslide hazard maps, and evaluating these maps, in the Kulekhani watershed (124 km2), located in the central region of Nepal. A landslide distribution map was produced from aerial photo interpretation and field checking. Eight factors, slope gradient, slope aspect, elevation, drainage basin order, distance from ridge, distance from valley, geology, and land use/cover, were analyzed by a multivariate statistical (discriminant) method, to determine the factors and classes influencing landsliding. Using a GIS, three different sets of samples from non-landslide group, and one set of samples from landslide group were generated from the unaligned stratified random and simple random sampling methods, respectively. Three different combinations were then used to examine the effects of the sampling methods on the critical factors and classes, and on the hazard maps produced from the results of the analysis. Each hazard map was produced with four levels of hazard classes: high; moderate; less; and least. The geology was found to be the most important factor to landsliding. Other important factors include elevation, land use/cover and slope aspect. Combinations of different samples did not yield substantial difference in the critical factors and classes, implying that the unaligned stratified random sampling is the appropriate sampling technique for generating the non-landslide group. The accuracies of three hazard maps were found to be acceptable for their practical use in hazard mitigation and watershed management planning.