Presented at the 2013 SMOS-Aquarius Science WorkshopAn important issue for ocean forecasting systems is the occurrence of sea surface salinity (SSS) bias at various scales which comes from ocean model errors and forcing uncertainties, mainly precipitations. It is not yet possible to fully remove SSS biases with the Argo network data near the surface (depth < 5m) due to a poor sampling. That is one of the reason why Mercator Ocean investigates the best way to deal with SSS data from space (SMOS and Aquarius) by taking into account different errors associated with different scales and possibly fill this data gap. Since several years, Mercator Ocean has designed and built a hierarchy of ocean analysis and forecasting operational systems delivering weekly and daily services in real time. The ocean and sea ice models are based on the NEMO/LIM codes. The data assimilation algorithm relies on a reduced-order Kalman filter with a 3D multivariate modal decomposition of the forecast error. Altimeter data, satellite sea surface temperature (SST) data and in situ temperature and salinity vertical profiles are jointly assimilated. In this study, we show that by assimilating Aquarius SSS data, a complementary information is brought. First results with the global 1/4Â° system show to what extent it should be possible to improve the meso-scale prediction and to correct sea surface salinity (SSS) bias.