Landslide hazard prediction under an extreme rainfall scenario by considering multiple timescale rainfalls and effective recharge
Zizheng Guo, Mengchen Cheng, Yunge Wang, Gang Xu, Yuhua Zhang, Chong Xu
Abstract
This study proposes a novel framework for landslide prediction considering both antecedent rainfall and critical rainfall (“event rainfall”). We focused on a landslide episode induced by a Typhoon event in Wenzhou City of SE China. A physically-based model was employed to calibrate regional soil properties and land use parameters. The temporal probability of rainfall events was analysed with the help of the Gumbel distribution. The effective recharge from antecedent rainfall was computed using the Easy_Bal software and multi-environmental variables. We computed the response of the probability of failure (PoF) on each pixel to two rainfall conditions. Landslide hazards for different future time periods were estimated by integrating the hazard matrix and topographic effects. The results showed that the variation in PoF subjected to antecedent effective recharge was larger than that to critical rainfall. However, the areas where the stability was mainly controlled by hydrological conditions were rather fixed. Overall stability in the region decreased with the increase of rainfall return periods, but visible differences were seldom observed when the rainfall went beyond a 20-year return period. Finally, the current approach to predict landslide hazard was verified by comparing two scenarios with different rainfall inputs (higher and lower than the Typhoon event) and the landslide episode.