Mapping post-wildfire erosion across California using very high-resolution multitemporal satellite imagery
Loading...
Authors
Suter, Ingrid Helen
Issue Date
2022
Type
Thesis
Language
Keywords
Debris-flow , Fire , Post-fire debris-flow , Probability , Rainfall intensity , USGS M1 likelihood model
Alternative Title
Abstract
Exposure to post-wildfire debris-flow hazards and the need for accurate hazard assessments is increasing for many communities owing to the dramatic increase in the frequency and magnitude of wildfires and the expansion of values at risk into steeplands. Empirical observations to guide predictions of the post-fire landscape response to rainfall have not kept pace with rapidly changing fire regimes have introduced high-severity wildfires into new climates zones, vegetation types and landscapes. Central and northern California are such regions where we lack extensive observations of post-fire erosion and debris-flow responses and hence the extent to which fire affects the susceptibility to runoff-related erosion is unconstrained. In this work, we utilize multitemporal, very high-resolution satellite imagery to map post-wildfire erosion and debris-flow responses, or lack thereof, in a consistent manner across five burn scars from southern California to northern California. We then compare mapped erosion magnitudes to maps of debris-flow probability from the USGS emergency post-fire debris-flow hazard assessment, which is based on an empirical model trained with debris-flow responses observed in southern California. We focus on fires that were predicted to have high debris-flow hazard owing to high values of topographic steepness, burn severity, and soil erodibility, and that have received a stressing rainstorm with intensities greater than the predicted threshold intensity needed to trigger debris flows. To turn mapped erosion magnitude into maps of debris-flow occurrence, we used the database from Staley et al. (2017) which includes 334 field-confirmed debris flows. We determined that debris-flow occurrence was associated with our highest erosion magnitude class in ~90% of cases and our medium class in ~10% of cases. Our mapping across California shows that while metrics currently used to predict debris-flow hazard were approximately uniform across studied fires, erosional responses were notably nonuniform. In the Transverse ranges of southern California, erosion features were ubiquitous and commonly consistent with a debris-flow interpretation, whereas moving into central and northern California erosion features were generally smaller in magnitude, less ubiquitous, and more consistent with features found after fluvial scour in steep channel networks as opposed to catastrophic debris-flows, or simply absent at the mapped scale. Such heterogeneity in the post-fire erosional response highlights the need for more observations to develop and test the next generation of post-fire debris-flow models capable of accurate prediction in different landscapes. These results demonstrate that mapping erosion features from multitemporal, very high-resolution satellite imagery can provide a useful and robust response metric.