The validation of a snow parameterization designed for use in general circulation models

A snow model designed for use in general circulation models is tested using atmospheric forcing derived from observations. Using data for four sites from the former Soviet Union, simulations of snow mass, snow density, snow depth, albedo and net radiation are validated. While we find the model works well in capturing the general patterns of the seasonal cycles and interannual variability in these quantities, there are systematic problems in the modelling of snow density and snow albedo towards the end of each snow season. An analysis of the snow densification parameterization shows that the model is very sensitive to the value of a poorly known parameter describing the temperature dependence of the snow compactive viscosity. Many problems in the simulation of snow stem from the simplification of complex processes occurring over time, such as layering and ripening, by a single bulk snow layer which prevents the model responding dynamically to changes in the forcing. In the simulation of the four sites, we used various methods for downward longwave radiation, and we find that the simulation of snow by the model is very sensitive to this quantity. Climate models do not provide this flux accurately in transitional seasons at high latitudes which suggests that it is probably premature to include very complex snow schemes into climate models. In this paper, we were able to obtain far better simulations by choosing the ‘right’ method for downward longwave radiation than by tuning parameters or increasing the complexity of the snow parameterization. However, the ability of the model to capture seasonal and interannual variability irrespective of which downward longwave radiation was used is important, and provides some support for the inclusion of complex snow schemes into climate models. © 1998 Royal Meteorological Society