Real-Time Assimilation of Streamflow into the National Water Model Channel Routing Using Coupled WRF-Hydro and DART

The National Water Model (NWM) generates streamflow analysis and forecast for 2.7 million river reaches and other hydrologic information on a 1-km and 250-m grid over the continental US. Assimilation of streamflow observations has received growing attention in recent years as a cost-effective means to improve prediction accuracy. Currently, the NWM employs deterministic nudging to assimilate over 6,000 USGS streamflow observations and provide initial conditions for its forecasts.

In this work, we describe research on a real-time, ensemble streamflow data assimilation (DA) system that may be considered for operational use of the NWM. The system, referred to as “Hydro-DART”, couples the the NWM’s channel model to NCAR’s Data Assimilation Research Testbed ( “DART”). DART provides state-of-the-art assimilation tools, such as localization and spatiotemporally-varying adaptive inflation, which are both adaptable and scalable to the goals of streamflow assimilation with the NWM. Our research approach centers on updating the model state of channel discharge via ensemble Kalman filter using along-stream-channel localization. In this presentation, we present comparisons with the current NWM nudging scheme at small and regional domains, including large-scale extreme flooding events such as Hurricane Matthew. We discuss the potential pathway to operations and how the work relates to the Joint Effort for Data assimilation Integration.