Multi-Model Comparison of Computed Debris Flow Runout for the 9 January 2018 Montecito, California Post-Wildfire Event

Hazard assessment for post-wildfire debris flows, which are common in the steep terrain of the western United States, has focused on the susceptibility of upstream basins to generate debris flows. However, reducing public exposure to this hazard also requires an assessment of hazards in downstream areas that might be inundated during debris flow runout. Debris flow runout models are widely available, but their application to hazard assessment for post-wildfire debris flows has not been extensively tested. Necessary inputs to these models include the total volume of the mobilized flow, flow properties (either inherent material properties or calibration coefficients), and site topography. Estimates of volume are possible in post-event (back calculation) studies, yet before an event, volume is an uncertain quantity. We simulated debris flow runout for the well-constrained 9 January 2018 Montecito event using three models (RAMMS, FLO2D, and D-Claw) to determine the relative importance of volume and flow properties. We broke the impacted area into three domains, and for each model-domain combination, we performed a numerical sampling study in which volume and flow properties varied within a wide, but plausible range. We assessed model performance based on inundation patterns and peak flow depths. We found all models could simulate the event with comparable results. Simulation performance was most sensitive to flow volume and less sensitive to flow properties. Our results emphasize the importance of reducing uncertainty in pre-event estimates of flow volume for hazard assessment. Plain Language Summary Changes in soil properties after fire mean rain can more easily erode the surface and initiate debris flows-mixtures of water, soil, and rock that rapidly move from steep source areas to downstream regions. The combination of frequent fire, steep landscapes, dense population, and intense rain makes southern California prone to debris flows. The scientific community has established methods linking debris flow likelihood and volume to rainfall intensity and burned area characteristics. We presently lack a basis for understanding how flow volume, flow material properties, representation of flow physics, and site topography all combine to produce an inundation hazard. We begin to address this by using observations from the 9 January 2018 Montecito, California, debris flow event to test three candidate models for debris flow inundation. We found that all models can simulate the event and that the largest source of uncertainty in inundated area is the flow volume. Key Points The 2018 Montecito event was successfully simulated with three debris flow inundation models that vary in model physics Model performance was more sensitive to flow volume and site morphology than to flow properties Estimated volumes were lower than rainfall-based predictions, and observed inundation areas were larger than volume-based predictions

Automated summary

The study focused on hazard assessment for post-wildfire debris flows in steep terrain, emphasizing the importance of reducing uncertainty in pre-event estimates of flow volume. Three debris flow runout models were tested using observations from the 2018 Montecito event, showing that model performance was most sensitive to flow volume and less sensitive to flow properties. This highlights the significance of accurately estimating flow volume for effective hazard assessment.