Spatially Distributed Snowmelt Inputs to a Semi-Arid Mountain Watershed
Spatial variability in snow accumulation and melt due to topographic effects on solar radiation, drifting, air temperature, and precipitation is important in detennining the timing of snowmelt releases. Precipitation and temperature effects related to topography tend to 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 only included in distributed models at small scales. Previous research has demonstrated that snowpack patterns are not well reproduced when topography and drifting are ignored. These observations imply that larger scale representations that ignore drifting could be greatly 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. This model was then compared to basin-averaged snowmelt rates for a fully distributed model, a single point representation of the basin, a two point representation that captures some of the variability in drifting and aspect, and a model with distributed terrain and uniform drift. The model comparisons demonstrate that the lumped, single point representation and distributed terrain with uniform drift both yielded very poor simulations of the basin-averaged surface water input rate. The two-point representation was an improvement but the late season melt required for the observed streamflow was still 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.