Long-term snow distribution observations in a mountain catchment: assessing variability, time stability, and the representativeness of an index site

Authors: Winstral, Adam; Marks, Daniel

Submitted to: Water Resources Research
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/3/2013
Publication Date: 1/15/2014
Citation: Winstral, A.H., Marks, D.G. 2014. Long-term snow distribution observations in a mountain catchment: assessing variability, time stability, and the representativeness of an index site. Water Resources Research. 50:293-305. DOI: 10.1002/2012WR013038.

Interpretive Summary: Alpine snow distributions are highly heterogeneous and this heterogeneity strongly affects earth/atmosphere water and energy exchanges, mountain hydrology, and alpine ecology. These vital differences often occur at scales of tens of meters, which are far too small to capture in continental or global models. In these large-scaled models it is impossible to explicitly account for these differences; instead statistical approximations based on theoretical distributions and characteristic statistics are sought. Yet since it is very difficult and costly to collect volumetric snow data, current approximations are based on a handful of isolated measurement campaigns. While it has been found that differing locations, climates, and topographies combine to produce broad characteristic regional patterns, it has also been shown that these broad characterizations vary in time. Very little research has been conducted assessing the controls on this variation. This research presents an analysis of a decade-plus of manually collected distributed snowcover observations from an intermountain site – the first data from such a site. We demonstrate how wind and melt further influence the “characteristic” heterogeneity at this site. It is also commonly accepted that theoretical lognormal distributions adequately approximate snow distributions; we show that Weibull and gamma distributions better fit these intermountain data. It is anticipated that these findings will help to further large-scaled modeling efforts.

Technical Abstract: This study presents an analysis of snow distribution heterogeneity and the factors affecting this variability. The analysis focuses on manually-sampled data from 21 snow surveys covering 11 years at the drift-dominated Reynolds Mountain East catchment (0.36 km2) in southwestern Idaho, U.S.A. Surveys were conducted mid-winter and in early spring. Inter-season and intra-season trends were examined along with the time-stability of distributions, goodness of fit to theoretical distributions, and the representativeness of an index site as a measure of basin-wide snow water equivalent. The average snow depth coefficient of variation (CV) over the entire time period was 0.71, which is in accordance with broad regional assessments. Higher wind speeds and increased melt led to increased heterogeneity and higher CVs. Consistent wind directions produced accumulation patterns that were very stable from year-to-year. Many previous studies have suggested that vital sub-grid snow heterogeneity in large-scaled models can be approximated with lognormal distributions. In every snow survey, it was found that Weibull and gamma distributions better approximated observations. It was also found that an index site, akin to most mountain weather observation sites, provided a reasonable approximation of catchment-averaged SWE in most years. However, the reliability of this measure decreased in years that deviated from normal patterns.