An Investigation of Liquid Water Percolation and Model Transferability in Multiple Snow Climates (Invited Presentation)

Differences in how snow processes are represented between hydrologic models cause disparities in the modeled snowpack evolution. Model performance is also degraded when applying a model to a new location or application for which it was not developed. Here, we investigate model sensitivity to both a parameter-based and physically-based liquid water percolation routine within the SnowModel snow evolution framework. Liquid percolation was specifically investigated since this process is sensitive to snow temperature, melt, and rain precipitation, which vary dramatically amongst snow regimes. We compared the two routines against continuous point measurements in Washington state’s maritime climate and Colorado’s continental climate. The physically-based liquid percolation routine improved peak snow depth by approximately 3 meters (~1.4m of snow water equivalent) in the Olympic Mountains of Western Washington as compared to the parameter-based routine. The physically-based routine also exhibited the best overall model fit and transferability between climates and snow seasons for all domains. Additional tests indicated that the physically-based routine identified sporadic runoff events better than the parameter-based routine, therefore resolving snowpack runoff throughout both the accumulation and ablation season.