(c) Meridional ocean heat transport from ocean reanalyses

We calculate the total meridional ocean heat transport (MHT) at the intergyre region using the following formulation in depth coordinates [33]:

where Cp is specific heat capacity (3994 J kg−1 C−1), ρ is density of seawater (1025 kg m−3), V is meridional velocity (corrected to ensure net zero meridional volume transport) and θ is potential temperature. The integration is done over the entire water column (D) and stretches from the eastern (xe) to the western boundary (xw). Following Hall & Bryden [33], we further partition the MHT into the vertical (MHTvert) and the horizontal gyre (MHTgyre) components of the heat transport using the following expressions:

and

where the overbars in MHTvert indicate zonal averages and L(z) is the corresponding ocean basin width at a given depth. The primes in MHTgyre indicate deviation from the zonal average. This distinction helps to clarify the respective contributions of the vertical or overturning and horizontal gyre circulation to the overall MHT at the intergyre boundary.

The MHT decomposition at the intergyre region includes components driven by the circulation associated with transformation of water from shallow to deep layers (referred to as ‘vertical’ circulation or overturning in depth space) and the circulation associated with the subpolar gyre (referred to as ‘horizontal’ or gyre circulation in depth space). The purpose of this approach in the present study is to elucidate the mechanisms that control the vertical and horizontal MHT variability, which are not likely to be the same processes [34]. This choice may seem counterintuitive given recent emphasis (following Lozier et al. [35]) on the computation of AMOC in density space at the OSNAP array, so we explain further. The transformation of lighter water to denser water in the SPNA is a gradual process with the upper layer steadily losing buoyancy as it circulates along pathways around the gyre (figure 1), leading to isopycnals that slope from east to west [35,36]. This means that, geographically, the subpolar gyre is an integral part of the AMOC pathway and the subpolar AMOC computed in density space is a rate of transformation from light to dense, shallow to deep, and east to west, and includes the gyre. If our goal is to estimate the total diapycnal transformation from light water to dense, the subpolar AMOC in density space provides that and is larger than that computed in depth space. However, while total MHT computed in density space and depth space are by definition identical, in order to examine the importance of the gyre in MHT, the decomposition needs to be in depth space.

The MHT calculations are performed using the monthly mean fields (1993–2020) of the global eddy-resolving ocean reanalysis, GLORYS12, retrieved from the Copernicus Marine Environment Monitoring Service (CMEMS, https://data.marine.copernicus.eu/product/GLOBAL_MULTIYEAR_PHY_001_030/description). The ocean model component of GLORYS12 is the NEMO platform driven by ERA-interim atmospheric forcing until 2018 and ERA5 thereafter [37]. GLORYS12 has a horizontal resolution of 1/12∘ and 50 vertical levels. For the MHT calculations, we use the latitudinally averaged (45∘ N--46∘ N) meridional velocities and potential temperatures stretching from the western to the eastern boundary. We also take advantage of the sea-surface heights in the wider North Atlantic from GLORYS12 to examine their relationship to the MHT variability. Furthermore, we evaluate the GLORYS12 MHT calculations using the ensemble mean and spread of four ocean reanalyses (GLORYS2V4, ORAS5, GloSea5 and C-GLORSv7), referred to as MHTgrepv2. These ocean reanalyses have a horizontal resolution of 1/4∘ and 75 vertical levels, and have been shown to provide reasonable estimates of the overturning and MHT when compared with OSNAP observations [38]. They are also available through CMEMS (https://data.marine.copernicus.eu/product/GLOBAL_REANALYSIS_PHY_001_031/description).