INTEGRATING PRECIPITATION ESTIMATES FROM AN ATMOSPHERIC MODEL ENSEMBLE WITH DEBRIS-FLOW MODELS TO ASSESS POST-FIRE DEBRIS-FLOW INUNDATION HAZARDS

Intense rainfall on watersheds burned by wildfire can generate debris flows that travel for several kilometers or more. When these fast-moving flows interact with downstream communities and infrastructure, they can have devastating impacts on life and property. Short-duration, high-intensity rainfall is a key driver of post-fire debris flows. Past studies developed models to estimate debris-flow likelihood and volume at the outlets of recently burned watersheds given information about topography, soil erodibility, burn severity, and rainfall intensity averaged over a 15-minute duration (I15). These models are often used following a fire to assess debris-flow hazards associated with design storms, which are characterized by a particular value of I15. While this approach provides valuable information for post-fire hazard planning, it does not account for the impacts of spatially varying rainfall intensities across the burn scar. Here, we utilize existing debris-flow likelihood and volume models in combination with an ensemble of rainfall intensity estimates from a high-resolution atmospheric model to derive probability distributions for debris-flow likelihood and volume at the outlet of a series of watersheds burned by the Thomas Fire near Montecito, California. Based on these distributions of likelihood and volume, we perform a series of debris-flow runout simulations to generate an uncertainty-rated prediction of debris-flow inundation. Specifically, we use Monte Carlo sampling methods together with a recently developed debris flow inundation model, the Progressive Debris-Flow routing and inundation model (ProDF). The 100-member atmospheric model ensemble was configured to provide a 24-hour forecast of peak I15 over the study area during a rainstorm on 9 January 2018 that produced a series of damaging debris flows. This work serves as a proof-of-concept experiment for how real-time weather forecast products could be integrated with land surface models to assess debris-flow impacts in an operational setting.