Variabilité de la salinité de surface d'après un modèle global de couche mélangée océanique

Sea surface salinity (SSS) influences numerous oceanic phenomena, for instance surface water ventilation, deep water formation and thermohaline circulation. SSS also controls some ocean-atmosphere coupled processes, such as the intensity of freshwater flux and the penetration of heat flux and turbulence. Salinity is more difficult to measure than temperature from in situ surveys, which results in 20 times less data being currently available. Moreover, sea surface temperature (SST) is routinely estimated from satellites, which is not possible yet for SSS. Two space missions will fill this gap soon: SMOS from the European Space Agency and Aquarius/SAC-D from NASA and CONAE. To contribute to the SMOS project, we propose a method for estimating SSS from current satellite observations and for studying the mechanisms governing its variability. We developed a simplified model of the ocean mixed layer, based on the "slab mixed layer" formulation (Frankignoul et Hasselmann, 1977). This 2D horizontal model is implemented over the global ocean, using a near 100 km resolution, and integrated during a climatological year. Air-sea fluxes are taken from the ECMWF meteorological model (ERA40 reanalysis) and the surface currents are provided by altimeter data (SSALTO-DUACS analysis). The mixed layer depth (MLD) is derived from SST observations, using an original inversion technique. The MLD fields obtained from this inversion are well correlated to in situ estimates. This effective depth represents the air-sea fluxes penetration and ensures consistency between fluxes, les currents and SST. We first validate the simulation through examination of the heat balance in north-eastern Atlantic, by comparing to measurements and models from the POMME experiment. Then we study the salinity balance in the global domain, in terms of its geographical distribution and seasonal evolution. Equilibrium between the various processes appears generally more complex than for temperature. Noteworthy, the role of atmospheric flux is less predominant (22%), while geostrophic advection (33%) and diapycnal mixing (22%) contribute more strongly. Our results indicate this model succeeds in reconstructing SSS variability over most of the oceans. Daily SSS variations are also simulated, whereas they are not represented in current observed data at a global scale. Owing to its simplicity and fast computation, the model will be useful to the SMOS mission. It can help for the measurement calibration/validation and provide a first guess estimate to the sophisticated algorithm required for SSS restitution.