ABSTRACT
Many landslide susceptibility and hazard zoning methods are currently available. The development of GIS-aided mapping procedures has enhanced the capability of preparation of landslide susceptibility maps, thus allowing the analysis of large areas. Unfortunately, testing of the predictive capability of such maps has received little attention in the literature. In this paper we discuss the influence of the terrain heterogeneity on the validation of shallow landslide susceptibility maps. The susceptibility has been determined by means of multivariate statistical techniques (discriminant analysis). Two methods have been used to validate the maps: an index of relative landslide density and a prediction performance curve. In order to evaluate the influence of the terrain heterogeneity on the susceptibility assessment we have used a pilot area in La Pobla de Lillet (Eastern Pyrenees, Spain). The susceptibility analysis has been carried out with four random samples extracted from sectors of different size and geology within the pilot area. We have followed the same procedure with each sample to produce a shallow landslide susceptibility map of the pilot area. Comparison of the results shows that geological heterogeneity affects susceptibility and reliability of the maps. The best performance is using a random sample extracted from the whole pilot area.
