The following HTML text is provided to enhance online
readability. Many aspects of typography translate only awkwardly to HTML.
Please use the page image
as the authoritative form to ensure accuracy.
Natural Climate Variability on Decade-to-Century Time Scales
ponding to negligible snow feedback to values representing a twofold amplification. Cloudiness complicates the interpretation of the snow feedback even further. The IPCC concludes that "the snow-albedo-temperature feedback processes in models are somewhat more complex" (IPCC, 1992, p. 117) than the conventional interpretation of the positive snow-albedo-temperature enhancement of a perturbation. Since these models also generally lack realistic surface hydrologies, one must view with caution their simulations and/or projections of atmosphere-snow interactions.
The model experiments of Yeh et al. (1983), Barnett et al. (1989), Yasunari et al. (1989), and Cohen and Rind (1991), all of which were summarized above, are potentially relevant to decade-to-century-scale climatic change even though the simulation periods are subdecadal. This relevance stems from the fact that climate changes over decadal time scales may perturb the large-scale snow distribution by amounts comparable to the changes of snow prescribed in the modeling experiments. However, biases introduced by the parameterizations of the snow and surface physics must be addressed before the results of the experiments can be viewed with confidence. A step in this direction has been made by Washington and Meehl (1986), who found that the inclusion of a simple temperature dependence in the parameterization of snow albedo can change substantially the globally averaged surface temperature increase caused by a doubling of CO2. Ingram et al. (1989) and Covey et al. (1991) also examined the effects of high-latitude surface albedo parameterizations, although their experiments focused on sea-ice-albedo. Using an earlier generation of the National Center for Atmospheric Research's model, Williams (1975) found that changes in surface albedo (and SST) influenced a simulated ice-age circulation more than did the orographic changes caused by the glaciation. Further surface-sensitivity experiments with more current atmospheric models and with more realistic treatments of the surface physics and hydrology are needed.
An issue in need of particular attention is the apparent paradox involving recent observational data (on snow and air temperature) and modeling studies of the feedback between snow, soil moisture, and temperature. The modeling studies cited above indicate that a positive anomaly of snow cover tends to depress surface air temperatures through its enhancement of soil moisture for up to several months after the snow melt. The model results seem inconsistent with the observed tendencies toward higher springtime temperatures and increases of snowfall in northern land areas over the past several decades. However, an earlier retreat of snow during the late winter or early spring may be a consequence of the warming, which is most pronounced in the winter and spring seasons, especially if the increases of snow depth have occurred primarily in the northernmost land areas (as the data suggest they have). The earlier retreat of snow creates the possibility that the albedo effect may offset, or even dominate, the tendency of greater soil moisture to delay the seasonal warming. Since this scenario is especially likely if the upper layers of soil dry rapidly, it is clearly important that models accurately resolve the hydrological and thermal changes in the upper layers of the land surface during the snow-melt period. The potentially high climatic leverage of the snow-melt period should make it a focus of observational and modeling studies, whether the time scales of interest are decadal or century or even longer.
Several conclusions emerge from the results surveyed here. The first, which pertains to the strategy for diagnosing the role of snow cover in climate variability, is that both modeling and data analysis are essential and complementary diagnostic tools. Because snow is involved in a variety of interactions within the climate system, and because the distribution of snow is determined largely by other components of the climate system, controlled experimentation with numerical models is a key element of the diagnostic strategy. However, the biases and other limitations of models are such that numerical experiments yield convincing conclusions only when the observational data give some credibility to the model results.
The diagnostic studies of the past few decades appear, at first glance, to have produced a somewhat inconclusive picture of the climatic role of snow cover. However, several scientific conclusions do emerge when the results are viewed in an aggregate sense:
Anomalies of snow cover are clearly associated with significant local anomalies of air temperature, at least in the lowest 100 to 200 mb. The duration of these local anomalies generally ranges from several days to several weeks and is often limited by the fluctuations of the snow anomalies themselves over weekly and monthly time scales.
Over the past two to three decades, changes of surface air temperature are broadly consistent with changes of snow coverage, especially in winter and spring.
There is little or no evidence that the albedo effect of wintertime snow anomalies produces meaningful signals in the large-scale atmospheric circulation.
The snow-hydrology-soil moisture feedback appears to be capable of producing a meaningful response in the atmosphere during spring and summer. The scale of this response is at least regional and possibly larger. Regions susceptible to snow-hydrology effects include eastern Asia and the northern portions of Eurasia and North America.
The mechanism(s) by which snow influences the large-scale atmosphere are sufficiently complex that rela-