Assimilation of HF radar: Raw or Cooked?

Powell, B.S., Matthews, D., and Janekovic, I.

HF radars cover much of the coastal United States, and many of these data are now being used in modeling activities. Through measuring the doppler shift of the returned signal, the surface ocean current along the radial line from the radar is derived. When two or more radars sample a near-orthogonal patch of ocean, simple statistical inversion techniques are used in an attempt to calculate ocean currents. Most modeling groups working with HF radar assimilate these artificially generated ocean currents. Variational data assimilation, which relies upon the physics of the model to generate the covariances, is ideally suited for working with data close to its source. To this end, we present a method in ROMS for assimilating direct radials generated in real-time by HF radars and compare the results with the assimilation of the artificial current vectors. Due to the temporal and spatial overlapping requirement, assimilation of radials will provide a larger constraint to the assimilation than the vectors, and we explore the results of the two.