Vulnerability of Snowpack to Wildfire and Changing Climate within Western U.S. Ecoregions

ABSTRACT The objective of this study is to estimate the impact of wildfires and climate change on snow water equivalent (SWE) across western U.S. ecoregions. Four measures of snowpack are evaluated including annual peak SWE, annual peak SWE normalized by winter precipitation, date of peak SWE, and date of melt‐out. Random forest (RF) models are developed for estimating post‐wildfire changes relative to pre‐wildfire conditions. The RF model inputs included SWE data from burned and unburned SNOTEL sites and topographic, weather, and land cover predictor variables. The RF models are then applied across the ecoregions. In all ecoregions, the median prediction from the model ranges from a 3.2% decrease to a 3.0% increase in peak SWE due to wildfire. The distribution of changes within each ecoregion includes both decreases and increases. The median prediction also indicates earlier peak SWE and melt‐out dates for all ecoregions except for the Arizona‐New Mexico Mountains. The median peak SWE date difference is approximately 12 days earlier and occurs in the Eastern Cascades Slopes and Foothills while the North Cascades has a median melt‐out day of approximately 14 days earlier. Terrain slope is the most important predictor of changes in peak SWE, and mean temperature is important for changes in the dates of peak SWE and melt‐out. The estimated total changes in SWE due to wildfires that occurred from 2015 to 2020 range from a 1% increase in the North Cascades to a 7% reduction in the Arizona‐New Mexico Mountains.