Wildfire impacts on western United States snowpack
Date
2022
Authors
Giovando, Jeremy, author
Niemann, Jeffrey, advisor
Arabi, Mazdak, committee member
Fassnacht, Steven, committee member
Stevens-Rumann, Camille, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
Snowpack in the western U.S. is critical for water supply and is threatened by wildfires, which are becoming larger and more common. Numerous studies have examined impacts of wildfire on snow water equivalent (SWE), but many of these studies are limited in the number of observation locations, and they have sometimes produced conflicting results. The objective of this study is to distinguish the net effects of wildfires on snowpack from those of climate. Data from 45 burned sites from the SNOTEL network are used to perform an empirical analysis to determine SWE impacts from wildfire. For each burned site, unburned control sites are identified from the same level III ecoregion. Impacts of climate changes on snowpack are analyzed first by comparing pre-wildfire and post-wildfire snow water equivalent at the unburned sites. Combined climate and wildfire effects are considered by comparing pre-wildfire and post-wildfire SWE at the burned sites. Wildfire impacts are then isolated by taking the difference between the burned and unburned sites. Four separate snow measures are considered in this analysis and include annual maximum SWE, normalized annual maximum SWE, peak SWE date, and melt-out date. Wildfires have on average advanced melt-out (9 days) and maximum SWE dates (6 days) and reduced annual maximum SWE (10%) across all the sites considered in the analysis. The combined effects of climate and wildfire have advanced melt-out and maximum SWE dates approximately 14 days and 10 days, respectively, while decreasing maximum SWE for the combined effects was approximately 10%. The wildfire-induced changes in SWE were compared to several possible controlling variables including burn severity, leaf-area index change, dominant pre-wildfire tree genus, years since the fire, and site elevation. Due to increasing wildfire magnitude, the potential vulnerability of snowpack is an important consideration for water managers. An analysis to quantify the spatial variability of wildfire impacts on snowpack within the western U.S. ecoregions and vulnerabilities of annual maximum SWE was performed. Random forest models were developed for each measure using topographic, climatic, and land cover predictor variables along with snowpack data from wildfire impacted SNOTEL sites. The results indicate terrain slope is an important variable for maximum SWE, while incoming shortwave radiation and aridity are important for peak SWE date and melt-out date changes, respectively. The largest spatial variability amongst all snow measures is maximum SWE with a range of 5% increase to over 10% decrease due to wildfire impacts. Spatial variability for peak SWE and melt-out dates varied between ecoregions with the largest range in the northern and mid-latitude ecoregions. Peak SWE and melt-out dates are expected to be earlier with the exception of the Arizona-New Mexico Mountains where later melt-out dates are possible. South-facing gentle slopes were identified as the most vulnerable for maximum SWE changes. The total snow water volume difference due to wildfires occurring between 2015 through 2020 ranged from a 1% increase in the North Cascades to a 6% reduction in the Arizona-New Mexico Mountains. A consequence of increased wildfire activity in the western U.S. has resulted in increasing post-wildfire risk assessments by federal, state, and local governments. Locations of these assessments include watersheds which have snowmelt as part of the hydrologic regime. The current gap in generalized recommendations for water managers related to parameter adjustments in snow models presents challenges for water managers performing these risk assessments. Data from wildfire impacted SNOTEL sites were again used to estimate changes in two key parameters (the melt-rate function and the snowfall threshold temperature). The observed changes from pre- and post-wildfire periods at each SNOTEL site were used to develop a suite of general linear models to adjust the melt-rate function and threshold temperature. The model inputs include readily available topographic, climatic, and land cover information. The results indicate melt-rates typically increase after a wildfire, especially for periods later in ablation season. The snowfall threshold temperatures were more variable and site dependent, although the statistically significant changes suggest increases in the threshold temperature will occur post-wildfire. The coefficients from the models suggest that changes to the vegetation canopy are most important for estimating melt-rate and threshold temperature differences beginning immediately after the fire event though approximately 10 years post-wildfire. After vegetation canopy, other important input variables include the air temperature and topographic characteristics (i.e., elevation, northness, and eastness).