Blowing snow and drifting snow are defined as events in which snow particles have become airborne and been transported by winds. Blowing snow refers to wind borne snow deep and dense enough to impair visibility at or above the heights of heads of observers. Wind borne snow not reaching this level is properly referred to as `drifting snow'. However, the difference between the two is arbitrary and `one of degree rather than kind' (King and Turner, 1997). Since the conditions for generation of blowing snow are strong wind and loose fallen snow on the ground, blowing snow events could happen under a clear sky. Blizzards are severe winter storms which results in very low visibility due to blowing snow and/or snowfalls.

Blowing snow is a common phenomenon in many high latitude regions. Mann et al. (2000) analyzed the British Antarctic Survey's second phase STable Antarctic Boundary Layer Experiment (STABLE2) data between March and November of 1991 at Halley research station, Antarctica and concluded that the occurrence of blowing snow events is 9.9% of the time between May and November (215 days). Dery and Yau (1999a), using the European Center for Medium-Range Weather Forecast Re-Analysis grid data (1979 -- 1993), analyzed the frequency of blowing snow days. They found that blowing snow can happen on one out of three days in many areas in Antarctic and Arctic areas.

The physical processes of blowing snow include the saltation of snow particles bouncing along the surface and the suspension of snow particles controlled by turbulence and the gravitational settling, as well as the sublimation of the particles and its impact on the size spectrum, plus the turbulent diffusion of the water vapor released and heat required during sublimation. Our basic model, PIEKTUK, comes in fetch- or time-dependent forms, simulating the evolution, with time or fetch, of a column of blowing snow lifted from a horizontally homogeneous snow-covered surface, once the threshold wind speed for saltation has been exceeded.

Three spectral, time-dependent blowing snow models, PEIKTUK-T, WINDBLAST and SNOWSTORM, developed independently at York University/CANADA, Leeds University/UK and Utrecht University/The Netherlands, have been inter-compared under standard conditions to validate the physics implementation and numerical aspects. In all models, sublimation leads to an increase in mixing ratio and a decrease in temperature and the vertically integrated sublimation rate generally decreases with time after an initial phase. However, if the feedback of sublimation on the environment is not considered in the model, the predicted sublimation rate can be much higher. The vertically integrated sublimation rates predicted by the three models, plus bulk version of PIEKTUK-T (-BT), are shown in Figure 1.

Comparing models against each other is an excellent test of internal consistency, but the real test of the assumptions built into the models requires comparisons with field measurements. We have also utilized available field data for blowing snow to test the models. The British Antarctic Survey's 2nd STable Antarctic Boundary Layer Experiment (STABLE2) was conducted at Halley in 1991 (Dover, 1993; Mann, 1998). The American Great Plains blowing snow observations were made near Laramie, Wyoming, U. S. A., in 1974 (Schmidt, 1982). In those experiments, wind speed, temperature, relative humidity, particle number density and particle size distribution were observed at several levels. In the basic models, particle settling velocity is assumed to be equal to spherical particle terminal velocity in still air (gamma=1) and the particle diffusion coefficient is set equal to that for momentum (beta_3=1). The results from model comparisons with field data show that the basic model has insufficient capability to diffuse enough particles upwards or to hold the particles in suspension. Although it can be corrected by increasing the particle diffusion coefficient (beta_3>1), our belief is that it is the assumed settling velocity of blowing snow particles that is in error. The settling velocity given by solving the particle motion equation in a turbulent spectrum is smaller than the terminal velocity. Particle irregularity will also reduce the settling velocity. The settling velocity of blowing snow particles has been obtained from observational data by fitting an ideal power law profile for each particle size bin. The settling velocity obtained in this way is generally smaller than the spherical particle terminal velocity (gamma<1). By using the modified settling velocities, model results are greatly improved, as shown in Figures 2 and 3.

Testing the model with field data is important to improve the parameterizations. After the model validation has been completed, it will be used to develop parameterizations to include the sublimation and the transport of blowing snow within land surface schemes (e.g. CLASS (Verseghy, 1993)) coupled to climate, weather forecast and hydrological models.