This paper discusses the development and testing of snow algorithms with specific reference to their use and application in land surface models. New algorithms, developed by the authors, for estimating snow interception in forest canopies, blowing snow transport and sublimation, snow cover depletion and open environment snowmelt are compared with field measurements. Existing algorithms are discussed and compared with field observations. Recommendations are made with respect to: (a) density of new and aged snow in open and forest environments; (b) interception of snow by evergreen canopies; (c) redistribution and sublimation of snow water equivalent by blowing snow; (d) depletion in snow-covered area during snowmelt; (e) albedo decay during snowmelt; (f) turbulent transfer during snowmelt; and (g) soil heat flux during meltwater infiltration into frozen soils.
Preliminary evidence is presented, suggesting that one relatively advanced land surface model, CLASS, significantly underestimates the timing of snowmelt and snowmelt rates in open environments despite overestimating radiation and turbulent contributions to melt. The cause(s) may be due to overestimation of ground heat loss and other factors. It is recommended that further studies of snow energetics and soil heat transfer in frozen soils be undertaken to provide improvements for land surface models such as CLASS, with particular attention paid to establishing the reliability of the models in invoking closure of the energy equation. Copyright © 1998 John Wiley & Sons, Ltd.

snoe hidrology; general circulation models: class; land surface schemes; energy balance