Sear surface salinity variability in subtropical North Atlantic from Aquarius satellite mission and historical in-situ data

Sea surface salinity (SSS) variability in the North Atlantic is studied using the first year of Aquarius observations and historical thermosalinograph (TSG) data, collected by the Global Ocean Surface Underway Data Pilot Project (GOSUD). The particular focus of this study is on the subtropical maximum of SSS, where the Salinity Processes in the Upper Ocean Regional Study (SPURS) experiment is about to start. Spatial maps of SSS, based on the Aquarius remote-sensing, are constructed using the optimal interpolation technique. A novel approach utilizes the long-wavelength error correlation, specified for each Aquarius track (beam) and for a given cycle. This results in much reduced inter-beam biases (which, if not treated, produce characteristic stripes in the SSS maps) and ascending-descending differences. The analysis demonstrates a consistent pattern of seasonal variability which is most pronounced in the tropical band and off the north-east coast of South America, extending as far as 25N. Other kinds of variability, characterized by strong spatial gradients, include rapid temporal changes on time scales as short as one month in some cases. Over the SPURS domain, the seasonal cycle, measured by Aquarius, is relatively weak with the amplitude of about 0.4 psu, consistent with the climatological data. The mesoscale variability of SSS is investigated using high-resolution TSG data. Wavenumber power density spectra of SSS are computed to determine the spatial decorrelation scales as well as other statistical parameters required (e.g.) for optimal interpolation of the high-resolution SSS data from Aquarius, and to assess errors due to the components of variability not resolved by the Aquarius measurements. In the wavelength range from 15 km to 400 km, the spectral energy density of SSS follows a power law with the slope n=-1.9. This slope is much steeper than the theoretical arguments of geostrophic turbulence predict under the assumption that SSS is a passive tracer. The observed spectral slope indicates that the small-scale structures are less energetic than expected and that spatial gradients of SSS reflect primarily the scale of large coherent vortices. The first zero crossing of the SSS autocorrelation function occurs near 100 km, consistent with the length scale of the energy containing eddies in this region.