Linking High Latitude Convection with the Meridional Overturning Circulation in a Hierarchy of GFDL climate Models

Climate models demonstrate strong natural variability on interannual to multi-decadal time scales of deep convection in the North Atlantic subpolar regions with some preferred modes of oscillation. Most CMIP5 (the fifth phase of the Coupled Model Intercomparison Project) models, under preindustrial forcing, also show periodic Southern Ocean (SO) open sea convection with a wide distribution in the spatial extent, periodicity and intensity of SO convection (de Lavergne et al., 2014).

Here we compare and contrast (a) the mechanisms and time scales of variability of deep-water convection in the North Atlantic and in the SO and (b) the oceanic teleconnections resulting from these convection spots. For part (b), the focus is on the multi-decadal to centennial variability of the upper and lower branches of the meridional overturning circulation represented by the Atlantic Meridional Overturning Circulation (AMOC) and Antarctic Bottom Water (AABW), as well as the potential see-saw mechanisms between these two.

The focus is on a hierarchy of coupled models from the GFDL (Geophysical Fluid Dynamics, NOAA) family including the coarse-resolution 3 o CM2MC, the 1 o ESM2G model, and the 1/4 o CM4 model. The latter two were part of the CMIP5 and CMIP6 inter-comparison projects, respectively.

Complex interplays between the SO convection, Labrador Sea convection, Antarctic Circumpolar Current (ACC) strength, Antarctic Intermediate Water (AAIW), AMOC are discussed and contrasted across models.

These models demonstrate regular out-of-phase multi-decadal variability (80, 50 or 100 years) of Southern Ocean open-sea convection and Labrador deep convection in long control simulations with no anthropogenic CO2 forcings. In the lowest resolution model, AABW and ACC strength increase, while AAIW formation and AMOC strength decrease in decades with strong WS Convection. AMOC responds strongly to the WS convection and resulting decreases in SO Westerlies, with AMOC signals propagating fast northward (on a 20- to 30-year timescale) all the way to the North Atlantic, affecting in turn the Labrador Sea density properties and convection.

The strong relevance of SO convection for AMOC (and the resulting see-saw mechanism between AMOC and AABW) in the lowest resolution model is due to a strong response of SO westerlies to WeddellSea convection and impacts on AAIW. The role of SO westerlies in directly driving AMOC decreases in higher resolution models, increasing the relative impact of the Labrador convective events on AMOC. The see-saw mechanisms between AAIW and AMOC, and between AMOC and AABW on multi-decadal time scales are also identified and contrasted between these varying resolution models.

Potential implications for both future climate and paleoclimate are discussed, and suggestions for research directions are proposed.