Relevance of erosion processes when modelling in-channel gravel debris flows for efficient hazard assessment

A storm, composed of two delayed cells, hit the Monte Antelao slopes (Dolomites, North Eastern Italy) in the early morning of July 18th, 2009. The resulting runoff triggered two consecutive debris flows along the Rovina di Cancia channel. The detailed topographic data collected before and just after this event allowed an accurate reconstruction of the morphological changes experienced by the channel bed. These data are here used as benchmark to test the ability of numerical models to reproduce the dynamics of a real event, taking into account the morphology changes of the channel bed. The aim is to provide an efficient model for engineering applica- tions on large scales, such as those required by debris flow hazard assessment. A rainfall-runoff transformationis applied to reconstruct the solid-liquid hydrograph needed for computing the debris flow propagation. Two routing models are used: a GIS-based movable bed model, and a widely used fixed bed model (FLO-2D). Although similar results are obtained in terms of areas subjected to deposition, significant differences emergein terms of mobilized volumes. Only the simulation of both the deposition and entrainment processes allows to reliably reproduce the sediment volumes estimated from the pre- and post-event topographic data. This in- formation is fundamental in any hazard assessment because the volume of sediment mobilized by debris flow events exerts a fundamental control on the extension of areas subjected to inundation and on the thickness of sediment deposits. The capability to reproduce correctly the mobilized volumes also entails a more reliable simulation of the evolution of the peak and volume of the solid-liquid hydrograph as the debris flow propagates downstream, allowing the identification of the channel reach where banks could be overflowed. Conversely, adopting a fixed bed model leads mainly to an underestimation of the both the transported sediments volumes and the area subjected to deposition. As a consequence, the maximum debris flow depth in the portion of the channel subjected to erosion is underestimated and that in the portion of the channel subjected to deposition is overestimated. All these types of information are of great importance for an effective hazard assessment.