A combined EOF/variational approach for mapping radar-derived sea surface currents

A technique for reconstruction of the 2d surface velocity field from radar observations is proposed. The method consecutively employs two processing techniques: At the first stage raw radial velocity data are subject to EOF analysis, which enables to fill gaps in observations and provides estimates of the noise level and integral parameters characterizing small-scale variability of the sea surface circulation. These parameters are utilized at the second stage, when the cost function for variational interpolation is constructed, and the updated radial velocities are interpolated on the regular grid.Experiments with simulated and real data are used to assess the method's skill and compare it with the conventional 2d variational (2dVar) approach. It is shown that the proposed technique consistently improves performance of the 2dVar algorithm and becomes particularly effective when a radar stops operating for 1–2 days and/or a persistent gap emerges in spatial coverage of a basin by the HFR network.

Research highlights
► Combining variational and EOF-based methods improves quality of HFR data processing.
► The method is particularly effective in reconstructing velocity from a single radar.
► The approach can be used for real-time data processing.