We use a mix of historical data analysis and controlled regional climate model simulations to test the hypothesis of slower snowmelt in a warmer world. We have four main conclusions. First, analyses of daily snowpack depletion over western North America demonstrate lower ablation rates in locations with less SWE. The positive relationship between ablation rates and SWE occurs because deeper snowpack persists to late spring and summer when energy availability is high. Second, the widespread simulated reduction of annual meltwater volume over western North America is associated with small regional increases in low snowmelt rates, and, critically, a large reduction in high snowmelt rates. This is an important new finding that contradicts the perhaps intuitive notion that snowmelt rates in a warmer climate will exceed historical values, suggesting a tendency for slower snowmelt in a warmer world. Third, the reduction in high melt rates occurs during spring and early summer, and the increase in low melt rates occurs in mid-winter. The reduction in high snowmelt rates is associated with a contraction of the snow season, where reductions in daily snow-cover percentage are skewed toward spring and summer. Fourth, in a warmer climate, the contraction of spring snow-cover reduces by as much as 64% the snow-covered area that is exposed to net energy sufficient to drive moderate to high snowmelt rates, resulting in slower snowmelt in a warmer world.
Large reductions of high spring snowmelt rates in response to climate warming will impact the hydrology of snow-dominated regions, where snowmelt rates are critically linked to streamflow. In particular, slower snowmelt in response to warming may have implications on reduced streamflow and basin water yield. Compared with rainfall, the low-intensity long-duration nature of snowmelt more readily infiltrates soils32, forming hydrologic connections between hillslopes and water tables33 that sustain streamflow34. High snowmelt rates have been shown to be particularly effective at generating streamflow17. Slower snowmelt in a warmer world may decrease the likelihood that wetness thresholds that permit hydrologic connectivity will be exceeded, leading to spring and summer streamflow declines and lower runoff efficiency. For example, slower snowmelt in regions with less SWE may help to explain recent findings35 that US basins with a lower fraction of precipitation as snow have lower mean annual streamflow. Thus, slower snowmelt has implications on ecological processes sensitive to a timing shift to earlier and reduced snowmelt and streamflow including atmospheric carbon uptake by forests12, fish survival rates36, 37 and the risk of wildfire15.
Slower snowmelt in a warmer world will also impact water supply and hydropower production. One hypothesis is that less spring meltwater from reduced snowpack will move less efficiently to downstream reservoirs at a time of increasing seasonal aridity, evaporative demand and socioeconomic needs. Slower snowmelt in a warmer world has unresolved implications on the future risk of spring snowmelt floods. Previous studies have predicted future declines in spring snowmelt floods in the western US4 and British Columbia, Canada38, due to simulated reductions in spring snowpack in response to climate warming. While extreme events are not evaluated in the current study, slower snowmelt may be an important mechanism to explain those projected declines in spring snowmelt floods. Conversely, mid-winter increases in melt rates combined with a greater proportion of precipitation falling as rain could locally increase winter flood risk.
Hydrologic implications of slower snowmelt in a warmer world are likely to vary substantially with regional climate, elevation, soil properties, vegetation, evapotranspirative demand, and climate response to changes in greenhouse gas concentrations. Our study is limited to presenting the processes that explain projected changes in snowmelt rates. We report large general declines in spring snowmelt rates over great spatial extents and averaged over a decade—the impact of warming on individual (for example, extreme) melt events may differ. Additionally, warming effects on snowmelt are likely to have more pronounced spatial variability than can be simulated at 4 km grid spacing. For example, topographic shading occurs at much finer scales than simulated here. Earlier snowmelt coincident with a time of lower seasonal sun angles39 may be slower than we report as a result of terrain shade effects40. Given the critical need to better understand the impact of climate change on water resources, future studies are needed to address the myriad hydrological and ecological consequences of less snowfall, reduced seasonal snowpack and a shift toward slower snowmelt.
We identify a mechanism of change in snowmelt rates that impacts snow water resources over much of western North America. While slower snowmelt in a warmer world may appear paradoxical or contrary to the well-accepted idea of water cycle intensification41, the change is analogous to a downward shift in elevation to a warmer environment where melt rates are historically lower. Such dynamics are evident in historical snowpack observations, where shallower snowpack generally melts earlier and at lower rates than deeper, later-lying snow-cover. The shift to slower snowmelt is caused by warming-induced snow-cover depletion that limits the potential for high snowmelt rates typical of the historical spring and early summer when the snowpack is less likely to freeze overnight, solar insolation is high and snow albedo values are low. The identification of such hydrologic shifts within elevation-driven regional climate regimes is critical as slight perturbations may cause thresholds to be crossed and system behaviour to be indefinitely altered21. The implications on streamflow and ecology of large-scale changes in the magnitude of this critical water flux must be better understood to increase climate resilience.