Pattern Recognition Analysis of Snowdrifts

Abstract
Terrain features and prevailing wind patterns play a dominant role in how snow is accumulated. Research was conducted to determine the location of snowdrifts on a 0.41 km2 subbasin of the Reynolds Creek Experimental Watershed located in the Owyhee Mountains in southwestern Idaho. Snow depths at maximum accumulation were determined on a 7.6 m grid using aerial photography. The terrain variables used in the analysis were ground slope, ground aspect, elevation, distance from the top of the ridge, in the direction of the prevailing wind and the difference in elevation between the top of the ridge and the point. A pattern recognition technique called cluster analysis was used to determine the best terrain variables for locating snowdrifts. Our analysis indicated that slope and aspect were the best variables. Discriminant functions using these variables and the prevailing wind were developed to predict the location of snowdrifts. The functions were tested on a 16.7 km2 area.