How can we account for antecedent conditions in rainfall thresholds for shallow landslides?
Abstract
Landslides are one of the most damaging natural hazards generating losses all over the world. They occur as the result of the increased stress or reduced strength in the soil, typically attributed to either earthquakes or prolonged/intense rainfall. The focus here is on the latter. While shallow landslides are recognized to be triggered by the increase in soil pore water pressure caused by rainfall, antecedent conditions, i.e., the wetness of the soil prior to the rainfall event, has also been found to play a key role. Here, we take advantage of the unique rainfall and landslide records available in Switzerland to define rainfall thresholds and study how we can improve upon their performances by considering antecedent conditions. We first look at antecedent rainfall, defined as cumulated precipitation over N-days prior to the rainfall events. We then compare the predictive power of two different estimates of soil saturation: one provided by a European physically based hydrological forecasting system (TerrSysMP) and the other by a Swiss conceptual hydrological model (PREVAH). We find the latter to be more useful in the context of landslide predictions and recognize that the spatial resolution (500m for PREVAH and 12.5km for TerrSysMP) is a key aspect leading to this result. After demonstrating the usefulness of saturation as estimated by PREVAH in improving upon classical rainfall thresholds, we explore how to best utilize the added information. We find hydrometereological thresholds (where the x axis represents an antecedent metric and y axis a rainfall metric), to be less successful than the classical total rainfall duration threshold at the regional scale. We propose instead a sequential thresholds system, where first a soil saturation state is defined and then two (or more) different rainfall thresholds are utilized, depending on the soil wetness status.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFMNH33A..02L