Font Size: a A A

Snow distribution in alpine watersheds

Posted on:1996-03-10Degree:Ph.DType:Dissertation
University:University of California, Santa BarbaraCandidate:Elder, Kevin JamesFull Text:PDF
GTID:1462390014986744Subject:Hydrology
Abstract/Summary:
Regression trees (binary decision trees) have been applied to snow water equivalence (SWE) measured in the field in several alpine basins of the Sierra Nevada of California. The independent variables used in the regression tree models were net solar radiation, elevation, slope angle, and vegetation cover type. Elevations were taken directly from digital elevation models (DEMs), and slope angles were calculated for each basin using the DEMs. Values of spatially-distributed net solar radiation were calculated using the DEMs, interim terrain-based data products derived from the DEMs, and solar ephemeris data. Vegetation data were obtained from related research projects or government agencies. Dependent variables, point values of SWE, were calculated using measured snow depths combined with snow densities which were interpolated over space from a limited number of sample points by several methods. Snow-covered areas were calculated for survey dates where either SPOT multispectral imagery or aerial photographs were available. Model results were tested at 5 m, 30m, 50m and 100m grid scales in basins ranging in size from 69 ha to 1909 ha. The regression tree models proved to be effective in modeling the spatial distribution of SWE in alpine basins. Results explained between 40 and 80 percent of the observed variance in the SWE field data using only two to four independent variables. Net solar radiation was the most important independent variable in almost all model results. Elevation and slope were of second and third order importance, respectively, and vegetation played a reduced role. Results showed that in complex alpine terrain the mean of field measurements is often not a good estimator of mean basin SWE conditions because the irregular terrain causes a strong spatial dependence in SWE distribution. In addition, the spatially-distributed SWE models show that the within basin distribution of SWE varies radically from mean conditions. Results show that 30m to 50m grid sizes may be the most appropriate scales for modeling SWE distribution in alpine basins similar to those studied.
Keywords/Search Tags:SWE, Alpine, Distribution, Snow, Net solar radiation
Related items