Discussion

Air temperatures in late winter and spring have a key role in determining the sensitivity of snowpack in mountain basins to warming (Lettenmaier and Gan, 1990; Stewart et al., 2004; McCabe and Clark, 2005). The results presented above demonstrate and quantify the sensitivity of annual peak snowpack and its timing, and of annual runoff, to air temperature and precipitation changes and their interaction in the three basins. Sensitivity of annual peak snowpack timing to air temperature and precipitation changes in the three basins shows that the sensitivity of peak SWE timing to precipitation changes is greater in the colder climate conditions: Reynolds Mountain responds to warming only (Figure 5), Marmot Creek responds to warming and to a lesser extent to precipitation (Figure 4); and Wolf Creek responds to a complex interaction of warming and precipitation change (Figure 3). The potential for precipitation to counteract the effect of warming on the magnitude of the annual peak snowpack, becomes smaller moving from northern latitudes to southern latitudes. Therefore, regional responses to warming and changes to precipitation must be considered (Bower et al., 2004), particularly when evaluating future mountain hydrology (Roche et al., 2018; Sultana and Choi, 2018). This is because the snowpack is shallow and warm at the beginning and end of the season; shallow warm snow ripens and melts faster than deep cold snow as it requires less energy input to overcome cold content and fill its liquid water holding capacity (Colbeck, 1976).
Simulations of future conditions for snow regimes in Reynolds Mountain are in accord with the SWE magnitude and timing trajectories of the past 50 years (Nayak et al., 2010). Higher rates of warming and increased precipitation are projected by RCMs in the northern latitudes (Mearns et al., 2007). Latitudinal change in the role of precipitation increase in offsetting the effect of warming on cold regions hydrology implies that, even though northern latitudes will warm more (Graversen et al., 2008), they will also have more precipitation. Therefore, the precipitation increase may offset the impact of warming on snow and hydrological regimes in northern basins. It is also expected that the response of hydrological processes in different latitudes to the same climatic change will differ.
Although the snow regime in Marmot Creek (Figure 4) is as sensitive as in Wolf Creek (Figure 3) to warming and a decrease in precipitation, its runoff regime is less sensitive than the runoff regime in Wolf Creek (Figure 9). These results are consistent with findings and projections for other mountain areas (Sultana and Choi, 2018; Roche et al., 2018; Jennings and Moltoch, 2019; López-Moreno et al., 2020). The relatively lower sensitivity of the forest peak snow in Wolf Creek (Figure 3e) is because of the increased unloading of intercepted snow at warmer air temperatures that counteracts the reduced snowfall (Pomeroy et al., 2015). The higher resiliency of the Marmot Creek snowpack is due to smaller changes at high elevations and in the blowing snow sink zone of the treeline forest in which a deep snowpack is deposited that remains until mid-summer (MacDonald et al., 2010; Harder et al., 2015; Rasouli et al., 2019a). High elevation and high latitude basins are more resilient to warming because their temperatures are currently well below that required to shift precipitation phase (Bavay et al., 2013; Jennings and Moltoch, 2019). The snowpack lasts longer on the ground at high elevations in Marmot Creek, which moderates the impact of snow loss at low elevations on runoff (Rasouli et al., 2019a; López-Moreno et al., 2020). A high elevation band with air temperatures similar to that in low elevations in Wolf Creek and a rainy environment in the spring and early summer peak runoff period (Pomeroy et al., 2016) explain why the drop in peak snow accumulation is not reflected by a proportional drop in annual runoff in Marmot Creek. This highlights the role of the spatial redistribution of snow on heterogeneous hydrological responses at different elevations in Marmot Creek. The snow and runoff regimes are the most sensitive to warming in Reynolds Mountain because of the (i) higher annual mean air temperature, (ii) near-freezing air temperatures in winter, and (iii) fewer number of days with freezing temperatures (120 days a year, Rasouli et al., 2019a). Rasouli et al. (2019a) found that under a moderate warming and increased precipitation of 7% and 2% in Marmot Creek and Reynolds Mountain, respectively, the annual runoff remained unchanged due to the offsetting effect of increased precipitation on increased evapotranspiration and offsetting effect of decreased sublimation on reduced snowfall (Rasouli et al., 2019a). Less sensitivity of annual runoff to warming relative to snowpacks suggests that warming mountain snowpacks can be decoupled from hydrological regimes (López-Moreno et al., 2020). Snowpack regimes in Reynolds Mountain are more sensitive to warming than to changes in precipitation, similar to the Cascade Mountains of Oregon, USA (Sproles et al., 2013).
Estimated snowpack reduction per degree increase of temperature is 8% in Wolf Creek, 10% in Marmot Creek, and 17% in Reynolds Mountain (Table 2). In Wolf Creek these results are similar to reductions observed in the Svalbard Archipelago (\(\sim 79\)° N, López-Moreno et al., 2016). Snowpack loss in Marmot Creek is in the range of 11–20% reduction reported for the Pyrenees (López-Moreno et al., 2013, 2014) and comparable to a 15% reduction reported for the Swiss Alps (Beniston et al., 2003). Snow loss per degree of warming in Reynolds Mountain is similar to a 20% reduction reported for the Washington Cascades (Casola et al., 2009). The results here are consistent with other basins with similar climates and that climate change affects snowpack in mountain basins across the globe with large reductions at mid-latitudes and relatively small reductions at high latitudes (Roche et al., 2018; Sultana and Choi, 2018).
Under a severe warming of 5°C and a 20% increased precipitation, annual runoff increases in all three basins (Table 2) because of the increasing importance of rain in warmer climates, suggesting that precipitation increase has a primary role in changing annual total runoff and there is a large shift in the runoff mechanism from being snowmelt-driven to rainfall-driven. This shift may result in reduced streamflow (Berghuis et al., 2014) if precipitation does not increase (Table 2). It might also alter forest vegetation over time by making it more prone to wildfire and disease.