On deep convection events and Antarctic Bottom Water formation in ocean reanalysis products

• Published in: Ocean Science
• Main Authors: Aguiar, Wilton, Mata, Mauricio M., Kerr, Rodrigo
• Format: Article in Journal/Newspaper
• Language: English
• Published: Copernicus Publications 2017
• Subjects: article; Verlagsveröffentlichung; Antarctic; Southern Ocean; Weddell; Weddell Sea; Antarc*; Sea ice
• Online Access: https://doi.org/10.5194/os-13-851-2017; https://noa.gwlb.de/receive/cop_mods_00008195; https://noa.gwlb.de/servlets/MCRFileNodeServlet/cop_derivate_00008152/os-13-851-2017.pdf; https://os.copernicus.org/articles/13/851/2017/os-13-851-2017.pdf

Open ocean deep convection is a common source of error in the representation of Antarctic Bottom Water (AABW) formation in ocean general circulation models. Although those events are well described in non-assimilatory ocean simulations, the recent appearance of a massive open ocean polynya in the Estimating the Circulation and Climate of the Ocean Phase II reanalysis product (ECCO2) raises questions on which mechanisms are responsible for those spurious events and whether they are also present in other state-of-the-art assimilatory reanalysis products. To investigate this issue, we evaluate how three recently released high-resolution ocean reanalysis products form AABW in their simulations. We found that two of the products create AABW by open ocean deep convection events in the Weddell Sea that are triggered by the interaction of sea ice with the Warm Deep Water, which shows that the assimilation of sea ice is not enough to avoid the appearance of open ocean polynyas. The third reanalysis, My Ocean University Reading UR025.4, creates AABW using a rather dynamically accurate mechanism. The UR025.4 product depicts both continental shelf convection and the export of Dense Shelf Water to the open ocean. Although the accuracy of the AABW formation in this reanalysis product represents an advancement in the representation of the Southern Ocean dynamics, the differences between the real and simulated processes suggest that substantial improvements in the ocean reanalysis products are still needed to accurately represent AABW formation.