HR: 0830h
AN: C41B-0962    [PDF]
TI: Modeling the Spatial Distribution of Snow in a Rugged Alpine Valley
AU: * Erickson, T A
EM: tyler.erickson@colorado.edu
AF: Institute of Arctic and Alpine Research, 1560 30th Street UCB 450, Boulder, CO 80309 United States
AU: Williams, M W
EM: markw@snobear.colorado.edu
AF: Institute of Arctic and Alpine Research, 1560 30th Street UCB 450, Boulder, CO 80309 United States
AB: One of the most challenging problems in snow hydrology is understanding the spatial distribution of snow properties in montane catchments. Natural snow variability is extreme due to complex environmental controls. Many authors have found relationships between topographic parameters and snow distribution, but conclusions are based on datasets that span one or two years and it is not known whether the results are specific to the particular years of the study. Research was conducted in the Green Lakes Valley, an east-facing headwater catchment that abuts the Continental Divide and is part of the Niwot Ridge LTER. Snow depth surveys at maximum accumulation were conducted from 1997 to 2003. Snow distribution was modeled using a combined deterministic and stochastic approach. The deterministic trend component was modeled using three approaches: a constant trend, a linear combination of linear topographic parameters, and a linear combination of non-linear combinations of topographic parameters. The topographic parameters considered included elevation, slope, an index of total radiation, an index of wind exposure, and an index of snowdrift formation. Additionally, a multi-year dataset that includes all the measurements taken from 1997 to 2003 was analyzed, which increased the power of the statistical testing. Wind exposure was found to be a significant predictor of snow depth for every year of the study. Elevation, slope, and the index of radiation were found to be significant predictors for some years, but not for others. All terrain parameters were found to be significant for the multi-year dataset when the non-linear trend model was used. An exponential variogram was used to characterize the spatial correlation of residual errors. The range and variance of spatial correlation were found to be related to commonly measured indices of winter precipitation. Moreover, the correlation (R$^{2}$) of winter precipitation to the variance of spatial correlation increased from 0.745 for the simple model to 0.906 for the non-linear model. Similarly, the correlation of winter precipitation to the length of spatial correlation increased from 0.452 for the simple model to 0.647 for the non-linear model.
DE: 1836 Hydrologic budget (1655)
DE: 1860 Runoff and streamflow
DE: 1863 Snow and ice (1827)
DE: 1884 Water supply
SC: Cryosphere [C]
MN: 2003 Fall Meeting