Spatiotemporal Variability of Twenty‐First‐Century Changes in Site‐Specific Snowfall Frequency Over the Northwest United States

In the Northwest United States, warming temperatures threaten mountain snowpacks. Reliable projections of snowfall changes are therefore critical to anticipate the timeline of change. However, producing such projections is challenging, as most state‐of‐the‐art climate models are limited in sufficiently resolving influential topography. Here we leverage atmospheric freezing level to estimate precipitation phase and project twenty‐first‐century snowfall frequency change at Snowpack Telemetry Network stations across the Northwest. Under “moderate” and “business‐as‐usual” emission pathways in Coupled Model Intercomparison Project phase 5 models, snowfall frequency is projected to decline at all stations. Business‐as‐usual declines accelerate after midcentury at most locations, whereas moderate declines decelerate. A “critical year” analysis identifies when decadal‐mean snowfall frequency is projected to fall below 50%, 25%, and 10% of cold‐season wet days. Results highlight regions particularly vulnerable to relatively near‐term change, such as the Cascade Range. Considerable station‐to‐station spatial variability emphasizes the value of this site‐specific approach. Plain Language Summary In the Northwest United States, warming temperatures threaten mountain snow resources, which supply freshwater in watersheds throughout the region. Reliable estimates of future snowfall changes are therefore crucial to determine the timeline over which these changes may occur. However, the tools generally used to estimate future snowfall, climate models, have difficulty calculating local changes across mountainous landscapes. Towards addressing this challenge, we use the height where temperature equals freezing to estimate snowfall versus rainfall occurrence over this century, from which snowfall frequency changes in climate models are calculated at point locations across the Northwest. Under “business‐as‐usual” and “moderate” greenhouse gas emissions, average snowfall frequency is estimated to decline at all sites by 2100. The rate of decline under business‐as‐usual emissions increases in the latter half of this century at most locations, whereas moderate rates decrease. A “critical year” identifies when the number of snow days averaged over 10 years falls below 50%, 25%, and 10% of all days receiving rain or snow. Results highlight regions that may experience critical snowfall frequency declines sooner, such as the Cascade Range. Differences among locations are considerable