The size, frequency, and burn severity of wildfires is increasing in forests across the western U.S. Erosion after wildfire, and particularly high severity wildfire, can deliver significant volumes of sediment to rivers, which will eventually make its way downstream to water supply reservoirs. The two core objectives of this project were to develop new computer-based modeling tools to identify potential reservoir vulnerabilities to post-fire erosion and sedimentation, and to apply our framework to evaluate wildfire risks to the water supply reservoirs across Utah, one of the driest states in the U.S.

The first major outcome of this project was a suite of novel geospatial modeling tools that together allow for the identification and quantification of risks posed by post-wildfire erosion and sedimentation to critical, downstream water infrastructure. Accomplishing this objective required the collection of new data sets and the development of numerous geospatial tools and predictive models, most notably: 1) an ArcGIS toolkit for process-based watershed delineation (the USUAL Watershed Tools), 2) a machine-learning model to predict and produce fire severity maps based on local topographic and vegetation conditions under variable fuel aridity conditions, 3) the largest known dataset of post-fire debris flow grain sizes and volumes outside of southern California, and 4) an ArcGIS toolkit that is capable of concurrently running the USGS post-fire debris flow hazard models, post-fire hillslope erosion models, and geometric sediment delivery models (the Wildfire Erosion & Sedimentation Toolkit (WEST)), which all together can provide predictions of the locations, volumes, and sizes of sediment that could enter a watershed's river network after a wildfire. Further, these models and tools integrate seamlessly to inform a river network sediment routing model (the Network Sediment Transporter (NST)) coded in Landlab as part of this project. In doing so, we created a comprehensive modeling framework capable of predicting fire severity and post-fire landscape response under a range of potential future weather conditions and ultimately estimate a potential range in the total magnitude, arrival times, and character of sediment that may reach critical water supply reservoirs downstream after a watershed burns.

The second major outcome of this project was the successful application of our new framework to conduct pre-fire vulnerability assessments of Utah's water supply reservoirs to post-fire sedimentation. As part of this effort, our aim was to address four key questions related to 1) identifying which reservoirs were most vulnerable, 2) quantifying the time lag between wildfire occurrence and loss of reservoir storage capacity, 3) determining potential controls on sediment connectivity, and 4) estimating the modeled increases in sediment yield contributed by wildfire. Through an evaluation of model scenarios from over 125 watersheds that drain to Utah's most critical water supply reservoirs, we found that:

Model results of post-fire sedimentation revealed the greatest reservoir vulnerability in Utah's watershed management areas of northern Utah and in southwest Utah, where in general we find the relative source areas of watersheds are much larger than their respective reservoir capacities compared to other regions of the state. Across Utah, we found that more than 50% of reservoirs were predicted to lose >1% of their storage capacity over a decade due to wildfire-related sedimentation, consistent with the state's estimate for average reservoir sedimentation. However, this rate increased with the modeled severity of wildfire and intensity of post-fire rainfall. In scenarios modeled under high percentile conditions of climate and weather, we found 25% of reservoirs were predicted to lose >5% of storage, more than 10% were predicted to lose >10% of storage, and a handful of reservoirs in the state were predicted to experience catastrophic losses in storage capacity (>50%) due to wildfire erosion and sedimentation. Most watershed models predicted that half of the total sediment inputs after fire would reach the respective downstream reservoir within 2 years, but the majority of sediment (90%) was not predicted to reach the reservoir until post-fire years 3-4 for more than half of the watersheds. For roughly 25% of watersheds, it took closer to or even over a decade. While a number of factors contributed to the observed variations in sediment connectivity (i.e., how much and how quickly sediment enters a reservoir), topography and the relative magnitude of post-fire debris flows appear to be key factors.

This work has to date resulted in the publication of peer-reviewed journal articles, the publication of open-source and open-access software tools, the successful training and degree completion for multiple postdoctoral researchers, graduate students, and undergraduates, the dissemination of knowledge through professional conference presentations, and the development of and active engagement with a stakeholder group that included representatives from a wide range of federal, state, and local government agencies, public utilities, dam operators, water conservancy districts, and more. We are continuing our efforts to translate all project findings and outcomes into open-access and open-source models and peer-reviewed publications.

Last Modified: 08/24/2023
Modified by: Jonathan A Czuba