SIMULATING POST-WILDFIRE RUNOFF: HOW IMPORTANT ARE SPATIAL VARIATIONS IN INFILTRATION CAPACITY AND MEASUREMENT SCALE?

Debris flows occur with increased frequency following wildfire in steep terrain relative to similar unburned areas and can pose serious threats to human life and infrastructure. In contrast to debris flows that mobilize from shallow landslides, which are commonly initiated on hillslopes due to a rise in pore fluid pressure within the soil, post-wildfire debris flows are regularly initiated by water runoff in steep channels. In post-wildfire settings, predicting the timing and magnitude of runoff can be complicated by strong spatial heterogeneities in saturated hydraulic conductivity caused by the patchiness of fire-induced soil-water repellency. While it may be feasible to collect infiltration measurements at a small scale throughout a recently burned area, it would be prohibitively time-consuming to explicitly map the spatial variations in soil-water repellency over an entire catchment. In this study, we used a spatially distributed hydrologic model to examine the relationship between spatially variable infiltration and a basin scale, effective saturated hydraulic conductivity by performing numerical experiments on synthetic topography. We then tested several methods for estimating basin-effective saturated hydraulic conductivity using field measurements of infiltration at a recently burned catchment (0.6 km²) in the Santa Ana Mountains, CA. Infiltration measurements were incorporated into the numerical model via several different methods and results were compared with observed runoff for different rainstorms. Preliminary results suggest that the basin-effective saturated hydraulic conductivity decreases slightly when variations in saturated hydraulic conductivity are spatially correlated relative to cases where they are uncorrelated. However, more generally, results indicate that runoff in response to high intensity rainfall can be adequately simulated by using the geometric mean of the distribution of point-based infiltration measurements. Improved methods for predicting runoff at the watershed scale within post-wildfire landscapes are crucial for developing process-based models that can be used to inform site-specific mitigation plans and hazard assessments associated with flooding and debris flows.