Investigation of rill initiation mechanisms following the 2017 Thomas fire, CA, using high-resolution topography data, An

Abstract

Wildfires are increasing in both number and severity throughout the mountainous western United States. Fire can alter soil hydraulic properties, including by inducing hydrophobicity and reducing soil wettability. Additionally, rainfall on the newly bare, exposed earth can result in soil sealing and compaction. Together, these effects often generate greater overland flow that can channelize into rills. These small, ephemeral channels can produce substantial erosion and sediment transport after wildfires and have been linked to post-fire hazards such as debris flows. Rills have been extensively studied in field and lab experiments and have been described using mathematical models. Although many studies include observations of rills following wildfires, few have linked these observations with mechanistic insights in the field due to their small footprint and short duration on the landscape. At my field site north of Montecito, CA, widespread rilling was observed after the high-intensity rainfall and consequent debris flows that occurred following the 2017 Thomas Fire. Here, I identified points of rill initiation in the burned landscape using high-resolution imagery and topographic data derived from a photogrammetry dataset collected following a post-wildfire precipitation event. I calculated and extracted local slope and drainage area values associated with each point of channel initiation. I then fit four theoretical channel initiation models presented by Montgomery and Dietrich (1994) and one additional model to the slope and area data. These models included: 1) turbulent Horton overland flow, 2) laminar Horton overland flow, 3) seepage erosion, 4) laminar saturation overland flow, and 5) turbulent saturation overland flow. All parameters other than slope were estimated together as one “black box” coefficient in order to fit the models. I separately calculated model coefficients for each of the five mechanisms using field-collected or estimated values to evaluate the real-world performance of the data-driven model fits. Although the model fit of rill channel initiation by saturation overland flow had the highest relative R-squared of the theoretical initiation mechanisms in this study, the data-driven model coefficient fits did not align with the model coefficients calculated using the field-collected parameters. This discrepancy suggested that none of the theoretical models investigated in this study fully captured the physical initiation mechanism of rills at this field site. Additionally, statistically significant differences in critical drainage areas for channel initiation between northeast- and southwest-facing hillslopes suggest the possibility of aspect-dependent hydrologic differences at the field site.