Long-term snow depth simulations using a modified atmosphere-land exchange model

Abstract

Significant areas of agricultural lands are subject to seasonal, relatively thin snow covers. This cover affects temperature and moisture in the soil beneath, watershed hydrology, and energy budgets. The depth of snow impacts soil freezing with implications for soil hydraulic properties and over-winter survival of certain crops. The objective of this study was to incorporate a sophisticated snow cover routine into the atmosphere-land exchange (ALEX) model to simulate snow depths and dynamics of the relatively thin snowpacks of the US Upper Midwest. The ALEX model is used in several agricultural modeling projects, and as the land-surface parameterization in a mesoscale forecast model, but with only crude snow cover treatment. We combined parameterizations and empiricisms from the literature with the ALEX structure. Only three parameters were adjusted to find a set that worked well for 48 station years from three sites in Wisconsin. These were a correction for gauge catch deficiency, the air temperature that differentiates rain from snow, and a parameter related to drainage of liquid water from a melting snowpack. A further independent test included 13 years from one site in Minnesota. The air temperature differentiating rain from snow was also determined by analysis of weather observations, independently of the snow model. Both this analysis and the model revealed 0°C to be the best choice for this temperature in our region. Results showed that with a minimum of calibration the model gives good predictions of continuous snow depth, capturing critical processes of accumulation, ablation and melt in a wide variety of situations. Discrepancies between model and measurements generally originated from a single event and were mainly attributed to processes of blowing snow, misclassification of precipitation type, and anomalous new snow densities. Our results demonstrated the robustness of existing parameterizations and empiricisms for translating environmental observations into snow depth in dynamic simulation models. (C) 2000 Elsevier Science B.V.