Assimilation of High Frequency (HF) Radar Observations in the Chesapeake-Delaware Bay Region Using the Navy Coastal Ocean Model (NCOM) and the Four-Dimensional Variational (4DVAR) Method

Consistent and accurate coastal ocean monitoring necessitates the availability of three components: (1) an observing network that adequately samples the monitored domain, (2) a coastal ocean circulation model with a sufficiently high resolution that takes into account the often complex geometry and dynamics that occur near the coastline, and (3) an analysis system that is able to accurately assimilate the sampled observations to initialize the coastal model for forecasting. Here the data from a network of three SeaSonde HF radars deployed in the mid-Atlantic region of the Eastern United States is assimilated into NCOM using a 4DVAR method during the month of July 2013. HF radars provide high resolution and wide spatial coverage of surface currents. Most assimilation studies using HF radars involve sequential methods whereas 4DVAR inherently allows for the temporal dimension available in the observations. The NCOM-4DVAR assimilation results show that the assimilation cannot accurately fit the observations with small initial conditions and model errors. The biggest challenge for the assimilation system consists of an erroneous wind stress that consistently steers the model in a different direction than the observed velocities. Even-though the assimilation was able to reduce a portion of the model’s discrepancy to the observations; those gains were almost immediately lost in the forecast stage. Reducing model resolution to be closer to that of the observations significantly improved the accuracy of the analysis and forecast, highlighting the importance of constraining high resolution models with a sufficient quantity of observations.