Probability of Road Interruption due to Landslides under Different Rainfall-Return Periods Using Remote Sensing Techniques

Abstract

Heavy rainfall-induced landslides along mountain routes in Taiwan easily cause road interruption, especially under extreme rainfall conditions. Road interruption due to rainfall-induced landslides is a major problem for Taiwanese authorities. In this study, remote sensing data, including satellite images and aerial photos, were employed to interpret landslides, which were mapped using geographical information systems (GISs). The goal was an event-based landslide inventory. Causative and triggering factors for landslides were adopted in a logistic model to predict landslide occurrence. These factors included lithology, normalized difference vegetation index, slope gradient, slope roughness, maximum elevation, total curvature, total slope height, maximum hourly rainfall, and total accumulated rainfall. With the proposed landslide prediction model, the probability of road interruption due to rainfall-induced landslides was evaluated under different rainfall-return periods. The model was confirmed by careful validation of reported historical events involving road interruption caused by subsequent typhoons. Validation results indicated that the landslide prediction model can be used in predicting road interruption due to rainfall-induced landslides.