Parameterization of large scale snow redistribution models using high-resolution information: tests in an alpine catchment (Invited)

Snowcover development in alpine environments is highly variable due to the heterogeneous arrangements of terrain and vegetation cover. The interactions between wind flow and surface aerodynamic characteristics produce complex blowing snow redistribution regimes. The snowcover distribution is also influenced by ablation, which varies with surface energetics over complex terrain. For medium to large scale hydrological and atmospheric calculations it is necessary to estimate blowing snow fluxes over incremental land units no smaller than hydrological response units (HRU) or landscape tiles. Blowing snow process algorithms exist and can be deployed, though a robust method to obtain HRU-scale wind speed forcing does not. In this study, snow redistribution by wind was simulated over HRUs in a mountain tundra catchment in western Canada. The HRUs and their aerodynamic properties were delineated using wind speeds derived from a high-resolution empirical terrain-based wind flow model. The wind flow model, based on Ryan (1977), uses a digital elevation model (DEM), reference wind direction and reference wind speed to calculate wind ratios (the ratio of simulated grid cell wind speed to reference wind speed) at 10 m cell resolution, based on terrain aspect, curvature and slope. A high resolution LiDAR DEM of the catchment was available for this. Three parameters are required by the model: the curvature length scale and the weights that control the influence of curvature and slope on calculated wind ratios. These three parameters were estimated via calibration on approximately 1,000 wind speed measurements from each of three meteorological stations located within the Marmot Creek Research Basin. Snow depths estimated from subtraction of summer from winter LiDAR-derived DEMs were used to analyze the relationships between snow depth, calculated wind ratios and terrain variables such as aspect, curvature, elevation, slope, and vegetation height. Snow depth was most strongly related to the calculated wind ratios; accordingly, HRUs were created by grouping grid cells according to wind ratio and averaging this ratio for each HRU. The upscaled wind ratios were applied to a measured reference wind speed to drive HRU-based snow redistribution calculations using the physically based Prairie Blowing Snow Model (PBSM). PBSM was coupled to an energy balance snowmelt model to account for the effect of midwinter melt events on snow ablation. The modelled results showed good agreement compared to the LiDAR-derived snow depth measurements and to manual snow survey measurements. These results indicate that the proposed method to obtain upscaled HRU-scale wind speed forcing is suitable and that a physically based blowing snow model can be applied over mountainous terrain. Modelled snow accumulation is very sensitive to the complexity of HRU definitions.