Back calculation of the 2005 Le Dar debris flow with EDDA 2.0 model: Initiation, entrainment, and deposition of a pro-/periglacial debris flow

Debris flows represent one of the major natural hazards in mountainous regions, and due to climate change, the hazard is expected to increase. With the retreat of glaciers and thawing of permafrost, new areas covered with loose material are left behind. Considering that several of these new areas have few or no recorded past events and that widely used methodological approaches are based on data from past events, the debris flow hazard assessment for these areas remains a significant challenge. Therefore, to reduce the risk related to debris flow, new methodologies and physically based models that couple the precipitation event to the initiation processes and, consequently, with the entrainment, deposition, and posterior flood, are required. 

A promising open-source, free, physically based, and depth-averaged model that targets this gap is the EDDA 2.0. This model couples precipitation intensity with infiltration and surface runoff, models slope instability and erosion by surface runoff as debris flow initiation processes, as well as the entrainment along the transition zone, deposition, and solid concentration evolution. This study evaluates the applicability of the EDDA 2.0 model for a rainfall-triggered debris flow in the Dar catchment, Switzerland. First, a comprehensive local sensitivity analysis is conducted through a one-at-a-time approach to identify the model parameters (soil and debris flow simulation) that control depositional extent, maximum flow depth at two cross-sections, and the eroded sediment volume. The sensitivity of each parameter is qualified and quantified by the screening (K1) and variance (K3) indices, respectively. From the sensitivity analysis, the model is calibrated for the 24th June 2005 event using the selected most relevant parameters. 

Our results show that the dominant parameters of the EDDA 2.0 model are the erodibility and the Manning coefficients, while the average grain size, deposition coefficient, and soil permeability play a secondary role in the analyzed outputs. The calibration process shows a good fit with the data observed after the event of the 24th of June 2005; for most of the analyzed metrics, the EDDA 2.0 model performs better than the RAMMS::DF, a widely used debris flow model in hazard assessment. While precipitation scenarios for hazard assessment are not yet included, they are part of a currently ongoing project. Preliminary modeling-with some limitations-provides us with the first insight into the challenges that must be addressed in the integration of precipitation with infiltration and erosion due to surface runoff.