The initiation of runoff-generated debris flow in steep carbonate catchments

Debris flows are a common hazard in Alpine headwater catchments during intense convective rainstorms. These debris flows are commonly triggered by runoff entraining previously deposited sediment within the catchment. A debris flow will be initiated if rainfall exceeds the given rainfall intensity threshold. We usually define the rainfall intensity threshold as a function of storm duration (rainfall intensity-duration threshold). Above this empirically recorded threshold, the resulting surface runoff can mobilise sediment from the hillslopes and within the channel network. Thresholds are usually defined empirically for a given geographic region via monitoring of debris flow occurrence and the triggering rainfall intensity. However, direct field observations and rainfall data are sparse and noisy, and it is impossible to define rainfall thresholds when historical data are unavailable. An alternative methodology to derive rainfall ID thresholds is to use simplified physics-based model simulations. In this case, a greater understanding of the controlling factors for debris-flow activities could enable better threshold estimation in unmonitored catchments.

Here we present the initial simulation results of three different monitored catchments in the Dolomite mountains of Northeast Italy. These catchments are dominated by steep dolomite bedrock walls, which can provide large volumes of surface runoff to the catchment during rainfall. To simulate the response to rainfall in these catchments, we use the SWEHR (Shallow Water Equation & Harsine Rose) debris flow model, which we calibrate using a combination of field data and a correlation maximising framework. By focusing on the runoff response to rainfall in the catchments, we identified several key factors in the calibration of the model. The timing and magnitude of the runoff is controlled by the hydrological characteristics of the bedrock, the roughness of the catchment, the availability of sediment in the catchment, and the characteristics of the rainfall. By running multiple rainfall simulations for the catchments, we show how these factors impact rainfall ID thresholds