Modeling wildfire impact on hydrologic processes using the Precipitation Runoff Modeling System

Abstract:

As large magnitude wildfires persist across the western United States, understanding their impact on hydrologic behavior and predicting regional streamflow response is increasingly important. Sediment and debris flows, as well as elevated flood levels in burned watersheds are often addressed, but wildfires also alter the timing and overall volume of both short and long-term runoff, making the prediction of post-fire streamflow critical for water resources management. Watershed models are a powerful tool for both representing wildfire runoff response and discerning the processes that induce that response. In the current study, selected wildfire-impacted basins across the western United States are modeled using the Precipitation Runoff Modeling System (PRMS) in order to develop a generalized approach. This distributed-parameter, physical process based watershed model allows us to target specific processes, while still having the flexibility to account for uncertainty and complex physical interactions that are not explicitly represented in model parameterization. Two change detection modeling approaches are considered. First, models calibrated using pre-fire data are applied to the post-fire period and residuals between simulated and observed flow are examined to quantify the response in each specific watershed. Here an analysis of the model’s ability to detect long-term response is also presented. Second, the post-fire conditions are modeled by adjusting appropriate parameters, and the parameter differences are used to guide process learning. In this latter method, parameters are specifically tailored to represent processes affected by wildfire, and scenarios with different parameter interactions are statistically compared. The results of these analyses are synthesized to provide a framework for predicting wildfire runoff response using PRMS, which will ultimately empower water resource decisions.