This study presents recent developments of the OceanVar oceanographic three-dimensional variational data assimilation scheme to create OceanVar2.0. The code has been extensively revised to integrate past developments into a single, consistent, fully parallelized framework. In OceanVar, the background error covariance matrix is decomposed into a sequence of physically based linear operators, allowing for individual analysis of specific error matrix components. We focus on the sea level operator, which provides correlation between Sea Level Anomaly, temperature and salinity increments. OceanVar2.0 offers the flexibility to use either a dynamic height or a barotropic model for closed domains as sea level operators. A diffusive operator to model the horizontal error correlations, replacing the previously used recursive filter, has been implemented. The new code was tested in the Mediterranean Sea and the quality of the analysis assessed by comparing background estimates with observations for the period January–December 2021. The results highlight the better skill of the barotropic model operator with respect to the dynamic height one due to the assumptions required for the level-of-no-motion. Furthermore, we present a method to assimilate along track satellite altimetry considering a forecasting model with tides.