Subseasonal Temperature Variability Drives Nonlinear Snow Loss With Warming

Abstract Documenting and predicting the loss of seasonal snow is vital but complex owing to regional heterogeneity in the response of snow to warming. Such heterogeneity arises because the marginal loss of snow per degree of warming depends strongly on mean winter temperatures: Regions warmer than −8°C experience steep, accelerating snow declines, while colder regions show little change. This robust, generalizable pattern emerges across snow measures, spatial scales, regions, and data sets, but lacks a clear physical explanation. Here, using in situ and gridded observations of Northern Hemisphere snow and temperature, we reveal a simple yet potent driver: how the mean and variability of the daily wintertime temperature distribution influences the number of winter days below freezing. As average wintertime temperatures warm toward freezing, days below 0°C—critical to snow accumulation and persistence—decrease nonlinearly. The loss of each cold day is in turn associated with a linear reduction in snowfall, snowpack persistence, and snow cover duration, explaining the rapid snow declines for regions with mean winter temperatures approaching—but still below—freezing. We further show that this relationship entails a sharp escalation in the likelihood of extremely low‐snow winters, or snow droughts, during warm winters, even in regions with average temperatures several degrees below 0°C. Our findings resolve a key puzzle in the nonlinear temperature sensitivity of snow, offering a framework for seasonal‐scale snow resource monitoring and forecasting and for interpreting future projections of seasonal snow with warming.