Abstract
A statistical analysis of annual extreme water-equivalents of ground snow (reported as inches of water) measured up through the winter of 1979–80 at 76 weather stations in the northeast quadrant of the United States is presented. The analysis suggests that probability distributions with longer upper tails than the Type I distribution of extreme values are preferable for describing the annual extremes at a majority of sites. Sampling errors and the selection of water-equivalents for planning and design purposes also are described.
