Downscaling Snow Deposition Using Historic Snow Depth Patterns: Diagnosing Limitations From Snowfall Biases, Winter Snow Losses, and Interannual Snow Pattern Repeatability

Repeatable snow depth patterns have been identified in many regions between years with similar meteorological characteristics. This suggests that snow patterns from previous years could adjust snow deposition in space as a substitution for unmodeled snow processes. Here, we tested a pattern‐based snow deposition downscaling routine which assumes (a) a spatially consistent relationship between snow deposition and snow depth, (b) interannually repeatable snow patterns, and (c) unbiased mean snowfall. We investigated these assumptions, and future avenues for improvement, in water‐year 2014 over the California Tuolumne River Watershed. 6 km snowfall from an atmospheric model was downscaled to 25 m resolution using snow depth patterns from seven different years, and was compared to a more common terrain‐based downscaling method. Snow depth patterns were influenced not only by snow accumulation, but also snowmelt, snow sublimation, and snow density, resulting in pattern‐based snow deposition downscaling that was too spatially heterogeneous. However, snow depth simulated using terrain‐based downscaling was too spatially homogeneous, and less spatially correlated with observations ( r  = 0.27), than simulations with pattern‐based downscaling using snow depth patterns from the simulation season ( r  = 0.76), or from a different year ( r  = 0.52). Overall, modeled snow depth errors at peak‐snowpack timing were driven more by atmospheric model snowfall biases than different downscaling methods. In order of most‐ to least‐importance, future research should focus on bias‐correcting coarse‐scale snowfall estimates, correcting snow deposition patterns for winter snow losses and snow density spatial variability, and identifying the historic periods of most‐similar snow accumulation.