Oliver Francis, Hui Tang, Martino Bernard

Runoff-generated debris flows are common hazards in mountainous regions, causing millions of dollars lost and hundreds of casualties yearly. Early warning systems based on rainfall thresholds have been implemented to reduce the impact of these hazards. These thresholds tend to be based on short monitoring periods, which cannot fully capture the varying responses of catchments to rainfall. As a result, the uncertainty of many thresholds is unknown, limiting their usefulness to the general public. We propose a new modelling framework to derive probabilistic rainfall intensity-duration (ID) thresholds from limited observations. We test this framework on a small catchment in the Italian Dolomites to determine probabilistic thresholds for the occurrence of debris flows. Instead of a widely used power-law function, our new rainfall thresholds are a negative exponential function controlled by infiltration capacity. These probabilistic thresholds can help improve early warning system performance by providing additional information to the public.