Weighting predisposing factors for shallow slides susceptibility assessment on the regional scale

摘要

The quality of landslide prediction models at the regional scale does not automatically increasewith the increasing number of layers assumed to be landslide predisposing factors. Additionally,the significanceof such layers as predisposing factors is frequently not evaluated. In this paper we apply a sensitivity analysis to astatistically-based landslide susceptibility model,performed for shallow translational slides occurred in a test sitelocated north of Lisbon (Portugal). The model is applied individually to each layer (e.g. slope,aspect,transverseslope profile,geomorphology,lithology,superficial deposits and land use) and to different combinations ofoverlapped layers. The computation of success-rates and prediction-rates for such models allows concluding thatthe relationship between number of variables within the prediction model and the quality of predicted resultsis not linear,and it is possible to obtain an accurate landslide susceptibility map using a limited number ofinstability predisposing factors in the prediction model.