Potential shifts in deep convection sites for the Atlantic Meridional Overturning Circulation

Recent studies based on the OSNAP observing system in the subpolar North Atlantic have suggested that Labrador Sea Water (LSW) production contributes very little (~ 2 Sv) to the Atlantic Meridional Overturning Circulation (AMOC). This contradicts the conventional view that the upper North Atlantic Deep Water (NADW) is largely formed in the Labrador and Irminger Seas, and in turn contributes roughly half of the deep limb of the AMOC, while the lower NADW originating from the outflow of dense water mass from the Greenland-Iceland-Norwegian (GIN) Seas contributes the other half. Given that these recent studies are based on the OSNAP observation system, which started in July 2014, and that the conventional view is based on pre-ARGO hydrographic datasets spanning multiple decades, there is a possibility that the major deep convection site for the AMOC has shifted from the Labrador/Irminger Seas to the GIN Seas in recent years due to natural variability and/or anthropogenic global warming (AGW). A preliminary analysis of historical hydrographic data and atmospheric reanalysis indicates that during the strong positive North Atlantic Oscillation (NAO) phase from 1975 to 2000, the westerlies were stronger and shifted northward. As a result, the Labrador Sea experienced enhanced deep convection and increased the production of LSW. During the negative NAO phase from 1950-1974, on the other hand, the westerlies were weaker and shifted southward, which in turn suppressed deep convection activity in the Labrador Sea and decreased the production of LSW. During the last 18 years, a weak positive NAO phase has prevailed. However, the volume of LSW is at its minimum similar to the negative NAO phase of 1950-1974. It is possible that the intensification of the Greenland ice sheet melting and other AGW effects are suppressing the deep convection activity in the Labrador Sea. In this study, these hypotheses are tested using a series of surface-forced ocean/sea-ice model simulations constrained by global hydrographic data for six decadal periods (1955-1964, 1965-1974, 1975-1984, 1985-1994, 1995-2004, 2005-2014). The method used, known as a robust diagnostic simulation, has previously been shown to reproduce realistic overturning pathways as compared to other ocean and fully-coupled model simulations.