The assimilation scheme is OI-Nudging and the Kalman gain is determined by a prediction error that is fixed in space and time. The model was corrected for a region bounded by the solid line in the figure below. In this region the prediction error was increased linearly from the solid boundary to the dotted boundary and kept constant in the interior. The surface SLA was mapped into the model interior using statistical correlation.
The data from two TOPEX tracks (tracks 10 and 253) that cut through this region was used in this test. The data was nudged into the model for 30 days.
The detail of the eddy initialization is described in Waseda et al. (2000).
The inserted eddy has a horizontal radius of 180 km and the thermocline displacement is about 150 m. Shown below are the snapshots of the SSH and the vertical temperature section at the end of the 30-day assimilation period.
