Adaptation of SnowModel for vehicle mobility in snow

Vehicle mobility in snow is of particular interest to the U.S. Army. To better predict snow cover impacts on vehicle mobility at fine spatial scales, a numerical model that can simulate the complex physical processes that determine snow strength is an essential tool. We investigated Glen Liston’s SnowModel, a spatially distributed, physically-based snow evolution modeling system, as a predictive downscaling tool for snow strength metrics by assessing its ability to accurately simulate snow depth and snow density at different model resolutions over a collection of Army training sites located in the United States with distinct terrain and land cover properties. Meteorological forcing for the model is generated from weather station data archived in the Integrated Surface Database and the Global Historical Climatology Network. In each domain, SnowModel was run on 10-meter, 100 meter, and 1 kilometer grids and the impact of model resolution on the simulated snow was assessed. The results clearly demonstrate the influence that land cover, in particular tree canopy, has on the simulated snow depth and density. For example, at all stations, snow depth was underestimated in forested areas. Resolution degradation had minimal impact on the results except where it resulted in a change from forested to non-forested land cover type. Efforts to better evaluate the model downscaling and forest canopy parameterizations, as well as efforts to incorporate snow strength as it relates to vehicle mobility using SnowModel are ongoing.