Exploring the Potential of Snow Data Assimilation to Improve Seasonal Streamflow Prediction Across the Western United States

Friday, 18 December 2015
Poster Hall (Moscone South)
Andrew James Newman1, Chengcheng Huang2, Martyn P Clark3, Andrew Wood3, Levi D Brekke4 and J R Arnold5, (1)University Corporation for Atmospheric Research, Boulder, CO, United States, (2)Beijing Normal University, Beijing, China, (3)National Center for Atmospheric Research, Boulder, CO, United States, (4)Bureau of Reclamation Denver, Denver, CO, United States, (5)US Army Corps of Engineers, Jacksonville, FL, United States
Traditional seasonal streamflow forecasting uses observations of snow water equivalent (SWE) as predictors in statistical models to provide seasonal runoff forecasts. Over the past 10-20 years, investigation of the ability to assimilate SWE observations into hydrologic models has shown mixed results. Additionally, these applications have generally been focused on well-known research basins or a limited basin set. Here we use a large-sample watershed scale basin dataset to explore the potential of SWE data assimilation in seasonal streamflow prediction across the entire Western US. Simple linear correlation analysis is used to define the extent to which springtime SWE observations provide information for 1 April to 31 July runoff volume forecasts. Comparing observed and simulated SWE-runoff correlations for over 100+ basins identify basins where assimilating SWE observations will have the most impact on seasonal volume forecasts. Results show that the least potential for SWE assimilation is in basins that are relatively well instrumented, as well as basins at higher altitude with well-defined accumulation and melt seasons. In these basins the precipitation input to the model (primarily snowfall) has similar information content to the SWE observations. A set of data assimilation experiments are conducted to illustrate inter-regional differences in the value of SWE assimilation.