Evaluation of MITgcm-based ocean reanalysis for the Southern Ocean

Abstract. Global and basin-scale ocean reanalyses are becoming easily accessible and are utilized widely to study the Southern Ocean. Yet, such ocean reanalyses are optimized to achieve the best model-data agreement for their entire model domains and their ability to simulate the Southern Ocean requires investigations. Here, we compare several Massachusetts Institute of Technology general circulation model (MITgcm)-based ocean reanalyses (ECCOv4r5, ECCO LLC270, B-SOSE, and GECCO3) for the Southern Ocean. For the open ocean, the simulated time-mean hydrography and ocean circulation are consistent with observations. The MITgcm-based ocean reanalyses show Antarctic Circumpolar Currents (ACC) measuring approximately 149±11 Sv. The simulated 2 °C isotherms are located in positions similar to the ACC and roughly represent the southern extent of the current. Simulated Weddell and Ross Gyre strengths are 51±11 Sv and 25±8 Sv, respectively, consistent with observation-based estimates. However, our evaluation finds that the time evolution of the Southern Ocean is not well simulated in these ocean reanalyses. While observations showed little change in open-ocean properties in the Weddell and Ross Gyres, all simulations showed larger trends, most of which are excessive warming. For the continental shelf region, all reanalyses are unable to reproduce observed hydrographic features, suggesting that simulated physics determining on-shelf hydrography and circulation is wrong. Nevertheless, ocean reanalyses are valuable resources and can be used for generating ocean lateral boundary conditions for regional high-resolution simulations. We recommend that future users of these ocean reanalyses pay extra attention if their studies target open-ocean Southern Ocean temporal changes or on-shelf processes.