| Home > Newsevents > Training > Rcourse_notes > PARAMETRIZATION > SURFACE_ASSIMILATION > |
A review of parametrization schemes
for land surface processes
April 2001
European Centre for Medium-Range Weather
Forecasts
Table of contents >>
Next Section >>
Previous Section >>
8 . Snow modelling
The effects of the snow mantle have to be taken into account for appropriate consideration of the surface thermal balance on high latitudes and mountainous regions in GCMs. The presence of snow reduces the energy available at the surface; the albedo of fresh snow is 0.85, while the albedo of a natural surface is in the range 0.12 to 0.25 (Dickinson, 1988). Snow melting is the most important source of soil moisture in spring in high latitudes. In general, due to its thermal properties, snow acts as an insulator between the air above and the soil underneath (Peixoto and Oort, 1992; Walsh et al., 1985). The thermal and mass budgets of a layer of snow lying on the ground is relatively easy to establish on a local scale (e.g. Anderson, 1976), but more complex at scales of a typical GCM grid box, due to the heterogeneity of snow cover. In spite of its importance, there is very little observational evidence of relevance to GCMs on typical melting rates, albedo, snow cover and metamorphic changes followed by the snow mantle.
Most, if not all, current GCMs carry a prognostic equation for snow mass (Manabe, 1969)
|
(33) |
where
is the snow depth (m of water equivalent), 
is the snowfall (Kg m-2s-1) and 
is the rate of melting. The snowfall is either given as an
independent amount from other physical parametrizations in the model, or
is the total amount of precipitation if the surface air temperature is below
a certain threshold. The role of snow parametrization schemes is to specify
the melting rate,
, and the snow albedo entering the thermal budget equation.Melting conditions are met when the equilibrium ground temperature is above 0 C. In that case, an adjustment down to 0 C is made by melting the necessary amount of snow, while the upper soil reservoir collects the melted water. The melted water exceeding the maximum capacity of that reservoir is lost into runoff. Albedo in snow-covered areas is modelled as a background value plus a correction dependent on the snow amount. Some models take into account snow `masking' by the vegetation in the computation of the albedo, so as to reduce the snow-covered area in the presence of tall vegetation, when compared with bare ground terrain (Blondin, 1991). The snow contribution to the albedo can also be made dependent on temperature; snow under melting conditions is made darker to simulate the effect of surface ponding (Dickinson et al., 1986).
Snow-cover fraction (the fraction of the grid box covered by snow) is important for its effects on the albedo and melting. Inspection of any satellite image reveals that snow-cover fraction is essentially dependent on vegetation, and on topographic details such as slope and aspect. In GCMs, the snow-cover fraction normally depends linearly on the amount of snow up to athreshold value, beyond which it is taken as 1. The threshold value can depend on roughness length, a crude way of parametrizing orography and vegetation effects.
More complex snow models are normally introduced by carrying extra predictive variables: snow density (Pitman et al, 1991; Verseghy, 1991; Douville et al, 1995), snow temperature (Verseghy, 1991; Dümenil, private communication),and snow albedo (Douville et al, 1995). Details of the snow-pack metamorphism, e.g. distinguishing between coarse and fine grain, or between old dark snow and finer fresh snow, can be considered by introducing a snow age (time elapsed since last snowfall, Verseghy, 1991) dependency on the density and on the albedo. For its independent thermal budget, the snow pack is considered as an additional variable-depth layer, with thermal conductivity and heat capacity dependent on snow density.
Melting of the snow pack can occur in these more complex models in two different ways: surface melting and deep melting. If the surface energy-balance equation gives a temperature above 0 C, melting of snow occurs at the expense of the excess of energy obtained by cooling the surface to 0 C. The resulting amount of water percolates into the snow layer and might refreeze within the snow pack at some unspecified depth, in a process called ripening of the snow pack. The remaining water wets the upper layer of soil. Deep melting occurs by heat conduction from underneath the snow pack if the soil is above 0 C. There is a adjustment of temperature, as in the surface melting, but no ripening is allowed, the water being immediately available to the soil layer. Note that a separate thermal budget of the snow layer is necessary for proper separation of the two melting mechanisms and the ripening of the snow pack.
As referred earlier in Section 3, phase change of the water in the soil is another important mechanism in high latitudes (Black and Tice, 1988, Williams and Smith, 1992, Miller, 1980). A parametric inclusion of the effects of the solid phase of water, although essential for modelling the soil water and energy transfer in high latitudes, is not considered in most GCM models. Its is possible to write additional equations for the conservation of frozen water at different soil layers (Verseghy, 1991, Pitman et al., 1991). Modifications to the traditional treatment include, in order of importance: i) The thermal effects related to the latent heat of fusion/freezing; ii) Substantial reduction in transpiration in the presence of a frozen ground; iii) Soil-water transfer dependent on a soil-water potential that includes the effect of frozen water. There are indications that these effects are very important for characterising the role of boreal forests in the climate system (Sellers et al., 1995).
Training Course Notes Front Page >>
Table of contents >>
Next Section >>
Previous Section >>
 |
11.06.2002 |
 |
© ECMWF |
