Application of a Physics‐Based Snow Model to Assess Impacts of a Changing Climate on Snowpack and Snowmelt in the Northeastern United States

Abstract

In the temperate, humid mountains of the northeastern United States (U.S.), seasonal snowmelt plays an important role in the annual water balance, and midwinter rain‐on‐snow events and spring snowmelt are common drivers of flooding. Frequent near‐freezing winter temperatures make the region particularly susceptible to midwinter rain and ongoing climate warming. However, models used to evaluate seasonal snowpack evolution and inform snowmelt‐derived streamflow forecasting remain limited in their ability to represent these conditions. Here, we evaluate northeastern U.S. snowpack under a changing climate using a distributed, physics‐based snowpack model. SnowModel was applied to a domain in the state of Vermont, surrounding Mount Mansfield (1,340 m). Testing various model configurations showed that incorporating a wet‐bulb temperature threshold (0.5°C) for precipitation phase partitioning and a gravity‐driven liquid water percolation scheme most effectively reduced model error. These modifications reduced mean percent bias by 33 and 75 percentage points for snow depth and snow water equivalent (SWE), respectively. Using the improved model, we assessed the effects of climate‐driven changes through a range of perturbations in temperature and precipitation. Middle elevations (700–900 m) were the most sensitive to these changes, exhibiting up to a 60% decrease in peak SWE with modest warming (+1.1°C). Incremental warming caused increased midwinter melt, including an event in which a +3.3°C temperature increase produced 40 mm of additional snowmelt (+700%) in 1 day at mid‐elevations. These findings demonstrate the sensitivity of the northeastern U.S. snowpack to climate warming, and highlight the importance of improved snow modeling for predicting hydrologic changes.