PREDICTING WILDFIRE EFFECTS ON STREAMFLOW USING HYDROLOGIC MODELING

The increasing incidence of large and severe wildfires is a growing concern for forested watersheds. Past research has shown that wildfires can cause substantial increases in peak flows and annual water yields, leading to potential water quality concerns and land and water management challenges. However, responses have been variable and there have been few studies at large basin scales, creating uncertainties about post-fire hydrologic responses. To address these uncertainties, we projected the effect of three larges wildfires (>70,000 ha) on streamflow in two important forested source watersheds in the Cascade Range of Oregon, US. Streamflow was predicted using the Soil and Water Assessment Tool (SWAT) model, calibrated on data prior to the wildfires. Wildfire effects were predicted by modifying model parameters to represent the landscape and hydrologic impacts of the wildfires based on burn severity maps. Burned and unburned scenarios were compared using random forest models to identify drivers of increased annual water yields and peak flows. The models predicted an average annual water yield increase of 10.5–14.6 % at the headwater scale and 2.2–8.0 % at the outlet scale. Peak flows had larger increases of 58.8–95.0 % in the headwaters and 17.6–24.2 % at the outlets. Wildfire severity and percent area burned were identified as the most important drivers of post-fire changes. However, annual precipitation, aridity, aspect, soil type, geologic province, and vegetation characteristics were important. This work helps further the goal of better predicting post-fire streamflow changes, which is critical for forest and drinking water management decision making. Even at very large scales, the effect of wildfire is operationally and ecologically significant.