Comparison of an empirical and a process-based model for simulating debris-flow inundation following the 2010 Schultz Fire in Coconino County, Arizona, USA

Abstract

The importance of understanding the extent of areas threatened by post-wildfire debris flows cannot be overstated, as illustrated by the post-Thomas Fire flows through Montecito, California, in January 2018. Methods and models developed by the U.S. Geological Survey to identify burned basins at risk of producing post-wildfire debris flows are well established, effective and commonly used. In contrast, there is no similarly established methodology for delineating debris-flow hazard zones downstream of basins prone to producing post-fire debris flows. Understanding potential inundation zones is critical for protecting human life, property and infrastructure. Recently, some communities and local government agencies have begun assessing potential risks from post-wildfire hazards before an area burns (pre-fire hazard assessments). These assessments utilize modeled burn severity maps and existing methodologies to identify basins likely to generate post-fire debris flows should the basins burn. In most studies, however, there have been no attempts to delineate hazard zones downstream of the basins that could produce post-fire debris flows. This information is critical for identifying mitigation opportunities and for establishing emergency evacuation routes and procedures. Here, we report on work using a newly developed process-based model and an empirical model, Laharz using two different sets of mobility coefficients, to assess debris-flow runout from a recently burned basin. The actual extent of debris-flow runout is known, which allows us to compare model performance. Laharz is efficient for assessing large areas but requires the user to select the location of deposition a priori, and mobility coefficients for post-fire debris flows have not yet been developed. Laharz did not adequately predict the downstream extent of deposition using either set of mobility coefficients. The process-based model using two sets of parameters, friction angle, φ, and ratio of pore fluid pressure to total basal normal stress, λ, provided a range of results. The simulation using parameters = 0.8 and φ = 0.35 provided the best match between mapped and modeled deposits and provided a better estimate of inundation relative to Laharz. This two-model approach is helpful for assessing the shortcomings and benefits of each model, and for identifying the next steps needed for developing a method to identify post-fire debris-flow hazard zones before a fire begins.