Surface circulation and upwelling in the Sardinia Sea: A numerical study.

Abstract: The surface circulation of the Sardinian Sea (the shelf-slope region west of Sardinia, western Mediterranean sea) and the coastal upwelling were studied through the analysis of a 4-years interannual simulation performed with a hydrodynamic 3D numerical model. The model (an implementation of the Princeton Ocean Model) was forced with realistic atmospheric and oceanic fields (analyses) for the quadrennium 2008–2011. The model assimilates sea level data using a 3D-variational assimilation software. Simulated velocities were decomposed in their mean and turbulent part. Eddy kinetic energy and eddy momentum flux, able to describe synthetically in terms of kinetics the fluctuating part of the flow, have been calculated. The EOF decomposition was used to get further insight on the simulated dataset and shed light on the variability of the main dynamical features, as well as to identify and separate the coastal upwelling signature. At surface the mean circulation is characterized by a southward current flow getting closer to the coast in correspondence of the southern corner of the Island where if flows over the shelf edge. Eddy momentum flux field suggests that this southward stream is accelerated by a transfer of momentum from the eddy to the mean field in the area where it reaches maximum velocity. The presence of such a stable stream, having also a consistent fluctuating part, is argued to precondition the coastal upwelling in the southern area. The phenomenology of such a coastal upwelling along the western coast of Sardinia is then described for the first time. The upwelling, especially evident in the southern part of Sardinia, constitutes the main surface temperature signal of the modeled SST anomalies. A significant correlation was found with both wind directions and current intensity, suggesting that both mechanisms (current and wind driven upwelling) participate to precondition and enhance (respectively) upwelling. SST satellite imagery support such a pattern found in the model results. [Copyright &y& Elsevier]