Ensemble Optimal Interpolation: multivariate properties in the Gulf of Mexico

High-resolution models can reproduce mesoscale dynamics and the variability in the Gulf of Mexico (GOM), but cannot provide accurate locations of currents without data assimilation (DA). We use the computationally cheap Ensemble Optimal Interpolation (EnOI) in conjunction with the Hybrid Coordinate Ocean Model (HYCOM) model for assimilating altimetry data. The covariance matrix extracted from a historical ensemble, is three-dimensional and multivariate. This study shows that the multivariate correlations with sea level anomaly are coherent with the known dynamics of the area at two locations: the central part of the GOM and the upper slope of the northern shelf. The correlations in the first location are suitable for an eddy forecasting system, but the correlations in the second location show some limitations due to seasonal variability. The multivariate relationships between variables are reasonably linear, as assumed by the EnOI. Our DA set-up produces little noise that is dampened within 2 d, when the model is pulled strongly towards observations. Part of it is caused by density perturbations in the isopycnal layers, or artificial caballing. The DA system is demonstrated for a realistic case of Loop Current eddy shedding, namely Eddy Yankee.