Comparisons of Spatially Distributed Hydrologic Model Outputs for the Dungeness Watershed to Field-Based Measures of Snow Water Equivalent

Many spatially distributed estimates of snow water equivalent (SWE) rely on modeling based in part on remotely sensed data. The Hybrid model is a spatially distributed streamflow forecasting model developed recently for the Dungeness Watershed of western Washington State that incorporates estimates of snow cover using NASA's MODIS. Prior to generating streamflow estimates, the Hybrid model estimates SWE across the watershed in 500m grid cells. Another NASA product, SNODAS, also estimates SWE through a combination of remote sensing and modeling. Over three winters, we monitored SWE on the ground at 12 field sites, and compared field results with those estimated by both the Hybrid and SNODAS models at two different scales (25 ha and 225 ha). Taking into account the size of the confidence intervals, only one snow course had a reasonable correlation of ground SWE with the model outputs at either scale. Both models consistently and slightly overestimated actual SWE for sites below 1400 m. Above1400 m, SNODAS consistently overestimated ground SWE, and the Hybrid model usually underestimated ground SWE, though to a lesser extent. Although the current Hybrid model performs reasonably well, more accurate point (cell) estimates of SWE within the Dungeness Watershed could provide more precise estimates of watershed-scale SWE, and, as a result, subsequent estimates of streamflow in the Dungeness River.