The influence of the spatial distribution of snow on basin-averaged snowmelt

Spatial variability in snow accumulation and melt owing to topographic effects on solar radiation, snow drifting, air temperature and precipitation is important in determining the timing of snowmelt releases. Precipitation and temperature effects related to topography affect snowpack variability at large scales and are generally included in models of hydrology in mountainous terrain. The effects of spatial variability in drifting and solar input are generally included only in distributed models at small scales. Previous research has demonstrated that snowpack patterns are not well reproduced when topography and drifting are ignored, implying that larger scale representations that ignore drifting could be in error. Detailed measurements of the spatial distribution of snow water equivalence within a small, intensively studied, 26-ha watershed were used to validate a spatially distributed snowmelt model. These observations and model output were then compared to basin-averaged snowmelt rates from a single-point representation of the basin, a two-region representation that captures some of the variability in drifting and aspect and a model with distributed terrain but uniform drift. The model comparisons demonstrate that the lumped, single-point representation and distributed terrain with uniform drift both yielded poor simulations of the basin-averaged surface water input rate. The two-point representation was a slight improvement, but the late season melt required for the observed stream-flow was not simulated because the deepest drifts were not represented. These results imply that representing the effects of subgrid variability of snow drifting is equally or more important than representing subgrid variability in solar radiation. © 1998 John Wiley & Sons, Ltd.