²ÊÃñ±¦µä

Schneider, Dominik ¹ ; Molotch, Noah P. ²

¹ INSTAAR
² INSTAAR

Snowmelt is the primary water source in the Western United States and mountainous regions globally. Forecasts of streamflow and water supply rely heavily on snow measurements from sparse observation networks that may not provide adequate information during abnormal climatic conditions. To address this issue we developed a real-time spatially distributed snow water equivalent (SWE) product for the Upper Colorado River Basin that merges previous years SWE patterns derived from a reconstruction model, with interpolations of real-time situ measurements. The approach uses a multiple linear regression to model SWE in which physiography and reconstructed SWE are treated as independent variables and observed SNOTEL SWE the dependent variable. Using a drop-1 approach, the years 2000 - 2011 (Mar 1, Apr 1, and May 1) consistently find reconstructed SWE to be a significant predictor and the explained variability of the model is improved between 0.1 and 0.5 (mean= 0.23) compared to a model just based on physiographics. Independent validation in the Front Range, CO produces mean absolute errors (MAE) between 0.13 and 0.18 m, with significant improvements between 0.03 and 0.09 m over both reconstructed SWE and physiographic SWE (p<0.05). Geostatistical interpolation techniques (IDW, kriging) are used to blend the regression residuals onto the regression surface to incorporate regional effects within the modeling domain. However, MAE is only marginally reduced (~0.01) with blending. Improved analysis of past SWE distribution can provide valuable information for modeling efforts to predict, e.g. hydrologic impacts due to climate change and disturbances. Future validation is planned in additional locations within the modeling domain and a real-time product is in development that uses this ensemble of past patterns of SWE to estimate SWE in the current water year.