Scale Effects in a Physically Based Distributed Snow Model

Highly heterogeneous mountain snow distributions strongly affect soil moisture patterns, local ecology, and ultimately the timing, magnitude, and chemistry of stream runoff. Capturing these vital heterogeneities in a physically based distributed snow model requires appropriately scaled model structures. While the scaling properties of snow distributions have been examined in some detail, the scaling characteristics of the forcing processes that ultimately control distribution and melt patterns have received little attention. In this study, we varied the scale of the main forcing processes - wind speed, snow accumulation, and solar radiation - and examined the consequent effects on simulated patterns of snow accumulation and melt. The effects of vegetation and terrain structure on these sensitivities were also assessed. We found that whereas the snowpack needs to be adequately resolved to a characteristic length scale, these main forcing processes do not always require the same level of detail. Considerable reductions in computational costs were obtainable with minimal effects on accuracy. It was also shown that when snow accumulation and energy fluxes are autocorrelated, scale degradation produces characteristic temporal biases in runoff simulations.