The reason is that there are several energy sources and sinks for the snowpack besides the direct exchange of heat with the atmosphere. The direct exchange of heat with the atmosphere is known as sensible heat flux. When the atmosphere is warmer than the snow surface, the sensible heat flux is typically positive, which by itself would yield either a warming snow surface or melting snow if the snow surface was at 32˚F, but there are other sources and sinks that can also play a role, as depicted below.
Latent heat flux is produced by the exchange of moisture with the atmosphere. If the air is dry, and ice is sublimating (or water evaporating if the snow surface consists of both water and ice), this has a cooling effect on the snow surface. Alternatively, if the air is moist, it is possible for water vapor to condense on the snowpack and release heat, which has a warming effect on the snow surface.
Sensible and latent heat fluxes are related to the difference in temperature and moisture content, respectively, of the snow surface and atmosphere, but are also affected by wind speed and turbulence. For example, a dry, windy day typically yields stronger sublimation and cooling of the snowpack than a calm day.
But there's more to the story than sensible and latent heat fluxes. Absolutely critical are the roles of longwave and shortwave radiation, the former emitted by the snowpack, clouds, atmosphere, and other "terrestrial" objects and the latter emitted primarily by the sun.
Downwelling longwave radiation from the atmosphere and clouds is absorbed by the snowpack, which also emits radiation at a rate dependent on the snow-surface temperature. If the snow absorbs more longwave radiation than it is emitting, the net longwave radiation is positive. By itself, this would warm the snowpack or melt it if the snow temperature were 32˚F. Conversely, if the snow absorbs less long-wave radiation than it is emitting, the net longwave radiation is negative. By itself, this would cool (or freeze) the snowpack. A snow surface can freeze when atmospheric temperatures are above 32˚F if the the longwave energy loss exceeds sensible heat flux.
On Saturday morning, my son and I ski toured in White Pine Canyon. In most areas, the snow surface was initially frozen, despite air temperatures in the mid-to-high 30s. At that time, the sensible heating of the snowpack by the atmosphere would, by itself, act to warm and melt the snow surface, albeit slowly since temperatures were just above the melting point. However, overnight skies were clear or covered by just a thin veil of high clouds. Under such conditions, the incoming longwave radiation was small and the net longwave radiation was strongly negative. The longwave energy loss exceeded the gain from sensible heat fluxes, enabling the snow surface to freeze. Also assisting was the dry atmosphere, which resulted in a small amount of cooling from sublimation.
We noticed that the snow was sometimes not frozen but soft near the trees. In those areas, the snow surface received significant longwave radiation from the trees and the net longwave energy loss was likely much smaller and unable to offset the sensible heating sufficiently to freeze the snowpack.
During the day, there is an additional energy source, shortwave radiation. The amount of solar radiation incident on a slope depends on the time of day and the slope aspect. At our latitude during the winter, south aspects receive more incident solar radiation than north aspects. East aspects see a morning peak in incident radiation, west aspects an afternoon peak. Total incident solar radiation from sunrise to sunset is largest on south aspects.
There are, however, some important seasonal differences. At our latitude, after the spring equinox, the sun rises and sets north of east and west, respectively. For a time after sunrise and before sunset, the north aspects actually receive more incident solar radiation than the south aspects. This effect is largest on the summer solstice (around June 21). In addition, the difference in incident solar radiation at solar noon between a north and south aspect is largest on the winter solstice and smallest on the summer solstice. For example, at solar noon on December 21st, there is about a 1200 W/m2 difference in incident solar radiation between a 30º south aspect and a 30º north aspect at our latitude. On June 21st, the difference is only about 400 W/m2.
One of the reasons why I like ski touring in White Pine in the spring is that there are many aspects and slope angles and you can chose your route depending on what the sun and snowpack are doing on any given morning. I am a big fan of choice when ski touring for reasons related to safety, creativity, and intellectual stimulation.
Finally, it is worth noting that the amount of radiation absorbed by the snowpack, and thus available to warm or melt the snow, is dependent on the ice crystal types, whether the snow is dry or wet (the latter consisting of both frozen and liquid water), and the presence of impurities like dust. New snow absorbs less shortwave radiation than old snow. Clean old snow absorbs less radiation than dusty or dirty old snow.
There is another source or sink of energy for the snowpack that I've ignored here and that is the transfer of heat with the ground. In the midlatitudes, the ground beneath the snowpack is typically very near 32˚F. In a deep snowpack during the spring, this is not a major player. In a shallow snowpack during winter, however, the transfer of heat from the ground can create a strong temperature contrast between the base of the snowpack and the snow surface that leads to the development of faceted crystals known as depth hoar.
To conclude, there's a lot going on to affect the snow than just the air temperature. Often it is the radiation, either the net longwave or shortwave, that is playing a dominant role in the snowpack evolution.
