Vegetation and terrain effects on snow accumulation and snow melt

In snow-melt dominated watersheds, snow accumulation and snow melt are the two most important processes in streamflow generation. Both processes show high variation within a watershed and in time. Variation at a stand level is mainly a result of wind redistribution, micro topography, fallen trees, individual plants, etc. At the watershed scale factors such as topography, aspect, and vegetation type and structure become more important. Little is known about the relative importance of small and large-scale variations for the modeling of snow water equivalent (SWE) in a watershed. In Cotton Creek, a snow-melt dominated, 17.4 km2 watershed located in the Kootenay Mountains, in south-eastern British Columbia, Canada, we designed snow surveys specifically to estimate the spatial variation of snow water equivalent (SWE). The watershed was stratified into 19 sites, according to elevation, aspect and forest cover. To capture small scale variability of SWE, 60 snow depth and 12 SWE measurements were taken along two perpendicular transects. The mean snow density of each site was used to convert the snow depth into SWE. To increase the sample size for large scale variation of SWE, 10 additional sites with 20 snow depth and 10 SWE samples at each site were added. Seven snow surveys were carried out in roughly bi-weekly intervals during spring 2005 and 2006. Multiple linear regressions are used to regionalize the SWE mean and the standard deviation of each site. Spatial autocorrelation of SWE is inferred by variograms computed along the 60 m transects. The observed small and large scale variation in space and time is summarised by regression tree models. Elevation, aspect and forest cover (clearcut and forest) explain about 80% of the observed large scale variation (mean SWE for each plot) of snow accumulation within the watershed in both years. Small scale variability, e.g. the variation of SWE within the plots, only depends on forest cover. Geostatistical analysis shows that the autocorrelation ranges of SWE within the 60 m transects are about 3 m and lower. During the snow melt, small scale variability increases, first at the lower elevations and than at the higher elevations. In future, we plan to include small and large scale variations of SWE into hydrological model to estimate the effects of these variabilities on snowmelt runoff and energy fluxes.