In the Northern Plains wintertime precipitation plays a vital role in the hydrological cycle. Understanding the spatial and temporal variations of snowfall are key in modeling the hydrological response of a watershed during springtime melt.

Modeling of the spring snowmelt often starts late in the winter season using estimates of the snowpack derived from National Weather Service measurements and cooperative observer data collection points. Typically there are few point estimates within a basin, and due to the nature of spatial variability of snow, the data collected is not representative of the snow distribution. Radar is an alternate tool for measuring the dynamic variability of snowfall.

Snowfall and snow redistribution are driven by many factors including the atmospheric conditions; of wind speed, wind direction, temperature, humidity, and surface features of vegetation, topography, slope and aspect of the land. The spatial and temporal variability of these factors makes the modeling of snow pack complex.

The use of a Geographic Information System (GIS) allows for correlating atmospheric conditions and spatial features to quantify how much snow will accumulate under various conditions. Under this framework it is possible to determine the snow depth and snow water equivalent at a location while taking into consideration the inherent variability of snow distribution factors, i.e. landcover, landuse, and topography.

During the winters of 1999-2000 and 2000-2001 a field site located near Grand Forks, North Dakota, was instrumented to collect atmospheric conditions, soil temperature and moisture. A snow course was established to collect snow depth and snow water equivalent measurements. Data collected was used to validate the results of a three-dimensional blowing snow model. The effect of using radar-estimated snowfall as data input verses in situ measured point values was evaluated as well as the importance of vegetation, particularly a living snow fence, on snow distribution.