Inferring watershed-scale mean snowfall magnitude and distribution using multidecadal snow reanalysis patterns and snow pillow observations

The magnitude and spatial heterogeneity of snow are difficult to model in mountainous terrain. We investigated how historic snow patterns from a 32-year (1985-2016) snow reanalysis, which uses the timing and rate of remotely-sensed snow disappearance to reconstruct snow water equivalent in past years, could be used to improve current year simulations of snowfall in the Upper Tuolumne, Upper Kings, and Sagehen Creek watersheds in California's Sierra Nevada. Building on past research linking observations of fractional snow-covered area (fSCA) to spatial patterns of snow water equivalent, fSCA maps were used to identify pairs of dates from historic years in the reanalysis with similar snow accumulation and depletion patterns. Historic snow accumulation patterns were then used to extrapolate snow accumulation observed by snow pillows to 90 m snowfall fields. These 90 m snowfall fields were used as input for snow simulations, the accuracy of which were evaluated versus airborne lidar snow depth observations. Except for water-year 2015, which had record low snow in the Sierra Nevada and unique snow patterns, normalized snow accumulation and depletion patterns identified from historic dates with spatially correlated fSCA agreed with each other well, with mean absolute differences of approximately 10%. Domain mean winter snowfall inferred from the relationship between historic snow accumulation patterns and current year snow pillow observations had a +/- 13% interquartile range of biases, relative to the snow reanalysis. Finally, snow depth simulations using snow pillow observed snowfall extrapolated with historic snow accumulation patterns had 70% better coefficients of correlations, and 27% better mean absolute errors, as compared to simulations using more common snowfall forcing. This work demonstrates the benefits of repeatable snowfall patterns and satellite-era snow reanalyses in mountainous regions with uncertain snowfall magnitude and spatial heterogeneity.