Investigating the Relationship Between Peak Snow-Water Equivalent and Snow Timing Indices in the Western United States and Alaska

Understanding the distribution of snow-water equivalent (SWE) is crucial for the prediction of water resources in the western United States. Backward running SWE reconstructions use satellite-observed, binary snow presence imagery to reconstruct SWE mass estimates. This approach relies on the connection between snow timing and peak SWE, yet few studies have directly examined this relationship. Here, we investigate the strength and spatiotemporal variation in this relationship across the western United States and Alaska. Within the SNOTEL network (n = 611 sites), we find that most variance in peak SWE is explained (median R-2 = 0.64, sigma = 0.18) by, in order of explanatory skill, the timing of snow disappearance, onset, and cover duration, with variation in skill primarily related to climate conditions-like winter storm size and storm frequency-rather than topographical setting. We expand this analysis with a diagnostic model of peak SWE driven by remotely sensed snow timing indices applied to five hydrologically important regions in the western United States and Alaska. Uncertainties arising between blending point and 500 m grid-scale observations were found to influence model SWE bias, but a robust correlation (median R = 0.88) with observations persisted across all tested thresholds. Overall, this supports the viability of snow timing information for quantifying spatial patterns of peak SWE (mean R-2 = 0.76, percent bias = 3.6%) over the past two decades. These findings carry important implications for the development of SWE reanalysis products and for the evaluation of climate and hydrologic models.

Automated summary

Understanding the distribution of snow-water equivalent (SWE) is crucial for predicting water resources in the western United States. This study found that most variance in peak SWE can be explained by the timing of snow disappearance, onset, and cover duration, with variations primarily related to climate conditions. The use of remotely sensed snow timing information was shown to be effective in quantifying spatial patterns of peak SWE over the past two decades.