Characterizing Wind-Induced Snow Redistribution
with Digital Terrain Analysis to Enhance Spatial Snow Modeling
Adam Winstral1, Kelly Elder1, and Robert Davis2
Snowmelt runoff forecasts form the foundation of agricultural practices, flood forecasts, and reser-
voir management for temperate mountainous regions throughout the world. Accurate geographical
assessment of on-the-ground snow water equivalence (SWE) is the foundation for physically based
runoff modeling.
Previous work has documented a distinct decrease in the strength of the physical relationships be-
tween modeled basin parameters (i.e., elevation, net radiation, and slope) and measured snow accu-
mulations above timberline. This reduction is attributed to the effects of wind-induced redistribution
above timberline, which heretofore have not been accounted for in modeling efforts. We attempt to
characterize the effects of wind-induced snow redistribution through the use of two GIS-derived
terrain parameters in our models of distributed SWE in the Tokopah Basin, a headwater catchment of
the Kaweah River, California. The study basin encompasses an area of 19.1 km2 with elevation
ranging from 2630 to 3495 m. The research area is in the alpine zone; 88 percent is classified as void
of canopy cover, 10 percent as containing 19 percent canopy cover.
The cosine of the angle between the zenith and the local horizon, and a directionally constrained
average slope gradient are focused on the predominant upwind direction in order to define the degree
of exposure or sheltering for a given site. The average slope gradient takes on both positive and
negative values based on the elevation difference weighted by distance of cells located within a user-
defined area, and the cell of interest. These redistribution parameters are individually regressed on
observed SWE to determine their applicability in further modeling. Regression trees that include and
exclude the redistribution parameters are then constructed and compared, based on the observed
accumulation patterns from snow surveys collected in April 1997 and May 1998.
Regression models that include redistribution parameters have a better fit to the observed snow
distribution than those formulated with just the elevation, net radiation, and slope predictors. The
difference between the models was small for the 1997 survey, considerable for the 1998 survey. The
similarity of the 1997 models coincides with that year's climatic factors, which restricted the amount
of snow available for redistribution. Individual parameter analyses found strong agreement between
the redistribution parameters and observed snow accumulation and direct correspondence with sub-
basin specific wind patterns. Though the radiation index and redistribution parameters are correlated
due to the strong southerly component of local winds, the overall evidence indicates that the em-
ployed redistribution parameters do approximate the effects of wind-induced snow redistribution
and did improve basin-wide models of snow distribution.
1Colorado State University, Department of Earth Resources, Fort Collins, Colorado 80523, USA
2 U.S. Army Cold Regions Research and Engineering Laboratory, 72 Lyme Road, Hanover, New Hampshire
03755-1290, USA
92
Previous Page
