Improving the characterization of initial condition for ensemble streamflow prediction using data assimilation

Abstract

Within the National Weather Service River Forecast System, water supply forecasting is performed through Ensemble Streamflow Prediction (ESP). ESP relies both on the estimation of initial conditions and historically resampled forcing data to produce seasonal volumetric forecasts. In the western US, the accuracy of initial condition estimation is particularly important due to the large quantities of water stored in mountain snowpack. In order to improve the estimation of snow quantities, this study explores the use of ensemble data assimilation. Rather than relying entirely on the model to create single deterministic initial snow water storage, as currently implemented in operational forecasting, this study incorporates SNOTEL data along with model predictions to create an ensemble based probabilistic estimation of snow water storage. This creates a framework to account for initial condition uncertainty in addition to forcing uncertainty. The results presented in this study suggest that data assimilation has the potential to improve ESP for probabilistic volumetric forecasts but is limited by the available observations.