Quantifying impacts of precipitation scenarios projected under climate change on annual probability of rainfall-induced landslides at a specific slope

Because of climate change, the intensity, duration, and frequency of future extreme rainfalls are projected to increase in many regions worldwide according to results from global climate models (GCMs). As rainfall is a major triggering factor for landslides, the frequency and risks of landslides are expected to increase. Consequently, it is crucial to quantify the impacts of climate change on landslide risks at a specific site. This study develops a statistical downscaling method based on the generalized extreme value (GEV) distribution and a probabilistic method for assessing the annual probability of rainfall-induced landslides at a specific site under projected precipitation scenarios from GCMs. The GEV-based statistical downscaling method bridges the gap between site-specific extreme rainfall (e.g., annual maximum 1-day rainfall) projection for a future scenario and GCM outputs with large spatial and temporal scales. A physics-based slope model is also used to properly depict slope failure mechanisms. The results indicate that the projected increase in extreme rainfall due to climate change leads to increases in annual frequency of heavy rainstorms and the annual slope failure probability, which are primarily controlled by the increase in the standard deviation, instead of the mean, of future extreme rainfalls.