This study uses the inventories of landslides during typhoon Mindulle, Morakot, and the 07/19 rainfall event established by Central Geological Survey as landslide data. The elevation, slope, slope aspect, slope high, lithology, terrain roughness, slope roughness, plan curvature, profile curvature, total curvature, distance from the road, and distance from the river are first chosen as the landslide causative factors, based on previous studies.Secondly, calibration and selection procedures are performed to efficiently select the factors. Logistic regression
method is used to establish the landslide susceptibility model. Furthermore, the rainfall intensities of different rainfall durations are used as a landslide triggering factor in different rainfall events. The maps of potential landslides are delineated to discuss the influence of rainfall on landslide susceptibility analysis. The landslide susceptibilities are separated into four levels, including high, medium, low, and steady. According to the results, the model adopting the proposed rainfall factor increased the landslide predictive capability for long-duration and high-intensity rainfall events, such as the Morakot event. For the other two events, similar landslide predictive capabilities are obtained with and without applying the landslide causative factor in the model.
Key Words : Logistic regression, the Southern Cross Island Highway, rainfall intensity.