Rapid Snowmelt Leads to Greater Streamflow Across the Western United States

Abstract:

Snowmelt is the primary source of surface water in the western United States and for approximately one sixth of the global population. Climate change is altering the magnitude of the mountain snowpack and the timing and rate of snowmelt. We ask how these changes to the mountain snowpack impact how snowmelt is partitioned between evapotranspiration (ET) and streamflow across the western United States. We hypothesize that rapid snowmelt is able to quickly satisfy atmospheric demand for water and bring the soil column to field capacity, inducing infiltration past the root zone and generating streamflow. We use the Variable Infiltration Capacity (VIC) model run at 1/16 of a degree from 1950-2013 using a gridded meteorological data set to simulate snow water equivalent (SWE), ET, potential evapotranspiration (PET), and baseflow (Q_bf). We compute long-term ET/precipitation (P) and PET/P ratios to derive a simulated streamflow anomaly for each model grid cell using the Budyko framework. We use changes in simulated SWE to compute the long-term average snowmelt rate for each model grid cell. Simulation results show a significant positive relationship between snowmelt rate and simulated streamflow anomaly (r²=0.321, p<0.001). There is also a significant, non-linear relationship between snowmelt rate and simulated Q_bf/P (r²=0.73, p<0.001) showing that grid cells with rapid snowmelt partition more water to Q_bf. Furthermore, we see that grid cells with high Q_bf/P have correspondingly high streamflow anomalies (r²=0.52, p<0.001). This shows that rapid snowmelt causes greater infiltration below the rooting zone to produce streamflow and positive streamflow anomalies within the Budyko framework, confirming our hypothesis. Due to these relationships, we can expect early, slower snowmelt driven by climate change to produce less streamflow across the western United States.