In: Biogeochemistry of Seasonally Snow-Covered Catchments. IAHS Special Publication No. 228, Proceedings of the international symposium held July 11-12, 1995, Boulder, CO, USA.

DISTRIBUTED SNOWMELT MODELING USING A CLUSTERING ALGORITHM

ROBERT F. HARRINGTON
Department of Hydrology and Water Resources, University of Arizona, Tucson, Arizona, 85721, USA
KELLY ELDER
Center for Remote Sensing and Environmental Optics, University of California, Santa Barbara, California, 93106, USA
ROGER C. BALES
Department of Hydrology and Water Resources, University of Arizona, Tucson, Arizona, 85721, USA

Abstract Methods for spatially distributed physical modeling of snowmelt are presented. Estimation of snow-water equivalence using regression trees, and distributed snowmelt modeling using iterative clustering are described; these methods are applied to Emerald Lake Basin, a 120 ha seasonally snow covered basin in California. Modeled snow melt and basin discharge substantially agree, yet modeled and observed snow covered area agree only over approximately 65% of the basin after five weeks of melt, highlighting the importance of using snow cover patterns to assess model performance.

Fig. 1 Emerald Lake Basin. Transect A-A' refers to snow-w ater equivalences shown in Fig. 4.

Fig. 2 Comparison of lake outflow and modeled snowmel t. (A) Daily discharge; (B) cumulative discharge.

Fig. 3 Comparison of observed and modeled SCA. White area s are snow covered; black are snow free.

Fig. 4 Cross sectional comparison of surveyed and modele d SWE along transect A-A'; zero distance along transect is at southwest end of transect (A)(see Fig. 1). SWE values are ove rlaid on the topographic profile.