Skip to main content
ARS Home » Pacific West Area » Boise, Idaho » Northwest Watershed Research Center » Research » Publications at this Location » Publication #199575
Title: Sampling designs for heterogeneous snow distributions

Author
item Winstral, Adam
item Marks, Daniel

Submitted to: Geophysical Union Canadian
Publication Type: Abstract Only
Publication Acceptance Date: 3/15/2005
Publication Date: 12/15/2005
Citation: Winstral, A., and Marks, D. 2005. Sampling designs for heterogeneous snow distributions. abstract volume, 31st Annual Meeting of the Canadian Geophysical Union, May 8-11, 2005. Abstract 147

Interpretive Summary:

Technical Abstract: Intensive snow surveys of mountain basins are the most accurate means of characterizing the heterogeneous mosaic of snow distribution typically present. The collection of survey data is however costly and time-consuming and important decisions are required to adequately sample larger basins. In this research, four years of survey data collected on a regularly spaced 30 m grid (n = 420) in the Reynolds Mountain East basin (0.36 km2) in southwest Idaho were used to analyze the ability of random, stratified random, and grid-based sub-sampling designs (n = 47) to reproduce the snow surfaces from the full population of samples. First, data from a previous survey were used to establish strata for the stratified design as well as weight the distribution of all the sub-samples. It was found that the reduction in sample size did not substantially affect estimates of basin-averaged snow-water-equivalence (SWE) or output from a snowmelt model initialized with the sub-sampled surfaces. Only subtle differences were observed between sampling designs indicating that accurate pre-determination of distribution patterns could substantially reduce sampling costs. The three sub-sampling designs were then applied assuming snow distribution data were unavailable. For the stratified random design strata were determined using terrain variables and modeled energy fluxes. Wide run-to-run variances were observed in the random and grid designs dependent on the degree of sampling of the drift areas. The stratified random design exhibited the least run-to-run variance and best-simulated late season runoff due largely to adequate representation of the drift areas.