Interactive comment on “ Impact of the oceanic geothermal heat flux on a glacial ocean state ”

My main concern with this particular methodology is as follows: In my view, a comparison to the impact of the geothermal heat flux on the present-day state in the same ocean model and a comparable configuration (without salinity restoring) is missing. A second pair of “control” experiments "GH_control" vs. "REF_control" is needed for present-day conditions to allow for a meaningful assessment, minimizing the influence of the “model error” that arises from using different ocean models. The present-day simulations referred to in the Discussion section (p. 3603) were either carried out un-


Introduction
The oceanic geothermal heating (OGH) is the heat flux through the sea floor which is generated by the internal heat content of the lithosphere.This flux is maximum near the oceanic ridges or underwater volcanic regions and is minimum (∼50 mW.m −2 ) in the abyssal plains (see e.g.Stein and Stein, 1992;Davies and Davies, 2010;Goutorbe et al., 2011).
The importance of the OGH as a heat source for the ocean system has long been controversial.Although the ocean is largely heated and thermally driven at the surface, several recent studies suggest that the OGH can also affect the ocean's dynamics and heat budget (Scott et al., 2001;Adcroft et al., 2001;Emile-Geay and Madec, 2009;Hofmann and Maqueda, 2009;Urakawa and Hasumi, 2009;Hieronymus and Nycander, 2012;Mashayek et al., 2013;de Lavergne et al., 2015).By applying a spatially constant or variable heat flux in Ocean General Circulation Models (OGCMs) forced with the present day climate, it is shown that the OGH is a significant forcing that can weaken the stability of the water column, warm the bottom water and strengthen the abyssal thermohaline circulation (∼5 Sv, Sv = 10 6 m 3 .s−1 ).
Sparse observations suggest that the high oceanic heat fluxes associated with spreading centres favour bottom water thermo-dynamical changes on regional scale and centennial time scale (Detrick et al., 1974;Joyce et al., 1986;Hautala et al., 2005;Björk and Winsor, 2006).A recent study based on laboratory experiment further explored the role of the OGH.The study supports the strong effect of the OGH on the regional scale but minimises its impact on the thermohaline circulation and the turbulent mixing (Zhou et al., 2014).
From a paleo-climate perspective, the abrupt release of potential energy due to the accumulation of OGH in the deep ocean is responsible for the rapid temperature variations observed in reconstructions of the last glacial cycles based on ice and sediment cores (Adkins et al., 2005).It is also postulated that the OGH could have a large impact on the glacial overturning circulation and the deep water properties, such as the deep CO 2 storage (Adcroft et al., 2001).A climate simulation of the Neoproterozoic Era (∼700Ma ago), when the Earth was entirely covered by ice (the so-called Snowball Earth hypothesis), reveals that the ocean is not stagnant and that the OGH may be a driver of its dynamic in decreasing the density of the abyssal waters, enhancing the convective vertical mixing and homogenising temperature and salinity in the water column (Ashkenazy et al., 2013(Ashkenazy et al., , 2014)).
To our knowledge, the impact of the OGH has not yet been investigated for more recent glacial climate period, such as the Last Glacial Maximum (LGM, ∼21ka ago), when the conditions were colder, the atmospheric CO 2 concentration was lower (Monnin et al., 2001) and the ocean stratification and deep circulation stronger than those found today (see e.g., Duplessy et al., 1988;Sarnthein et al., 1994;Adkins et al., 2002;Adkins and Schrag, 2003;Curry, 2005;Lynch-Stieglitz et al., 2007;Otto-Bliesner et al., 2007).In the present study, we tested previously posed theories (e.g.Adkins et al., 2005) of the impact of the OGH on a glacial ocean state by using the forced LGM global configuration of Ballarotta et al. (2013) setup with realistic bathymetry and atmospheric forcing.
Here, we aim to 1) evaluate the impact of the OGH on the ocean circulation, in particular the North Atlantic and abyssal thermohaline circulation, the advective heat transport and the stratification; and 2) test whether the OGH could be a trigger of transition between glacial and interglacial climate.
The paper is organised as follows: the ocean simulations are described in section 2; the impact of the OGH on the LGM simulated state is described in section 3; Results are discussed in section 4 and a conclusion is given in section 5.

Model description
The NEMO-LIM2 model (Madec, 2008) was used to design the numerical experiments.The model configuration was similar to the experiment made by Emile-Geay and Madec (2009) in their study on the impact of the OGH in the present-day climate.NEMO solves the primitive equations discretised on a curvilinear horizontal mesh and was based in our study on a 2 • ×2 • Mercator grid (namely the ORCA2 global configuration).Within the tropics, the meridional resolution is increased up to 0.5 • .The vertical dimension is discretised into 31 unevenly spaced depth levels (10 meters at the surface and 500 meters in the deep ocean).The LGM bathymetry is derived from the present-day bathymetry minus 120 meters, representative of the alteration of the sea level due to the freshwater storage in the continental ice-sheets during the LGM.The vertical eddy viscosity and diffusivity coefficients required to model the vertical mixing were computed from the Turbulent Kinetic Energy (TKE) turbulent closure model (Gaspar et al., 1990;Blanke and Delecluse, 1993;Madec, 2008).The NEMO model uses the TEOS-10 equation of state (Roquet et al. , 2015).The parameterisation of the mesoscale eddy-induced turbulence was established by the Gent and McWilliams (1990) formulation, which associates an eddy-induced velocity to the isoneutral diffusion.The ocean is coupled to the Louvain-La-Neuve Ice Model LIM2 (Fichefet and Maqueda, 1997;Timmermann et al., 2005) which solves the thermodynamic growth and decay of the sea ice, the sea ice dynamics and its transport.
NEMO-LIM2 was initialised at rest and with the temperature, salinity and sea-ice fields averaged over the last 100 years of a 4,000 year long LGM experiment carried out with the MPI-OM coupled model (MPIOM LGM-W in Zhang et al. (2013)).The surface boundary conditions are computed using the CORE bulk formulae (Large and Yeager, 2004) and the atmospheric fields from a LGM quasi-equilibrated climate model experiment (Brandefelt and Otto-Bliesner, 2009).We took the atmospheric forcing from Brandefelt and Otto-Bliesner (2009) because it corresponds to a quasi-equilbrated state representative of an LGM state where the abyssal ocean is equilibrated to the LGM forcing.Note that no restoring sea surface salinity term was applied but the freshwater budget was constrained to have an instantaneous zero global mean value.A more exhaustive presentation of the experimental setup and boundary conditions can be found in Ballarotta et al. (2013).Although the initial states used in Ballarotta et al. (2013) and in the present study originate from different models, they have similar structure: the temperature, salinity and sea-ice cover distribution are similar, in particular the deep saline waters in the abyss and the simulated large sea-ice cover.
For the present study, we designed a reference experiment (REF) without OGH at the sea floor.In a second experiment (GH), spatially varying OGH fluxes were applied as the bottom boundary condition.Following Emile-Geay and Madec (2009), the OGH is computed from the age of the bedrock.
We assumed that the OGH flux during the LGM was the same as today, since it is estimated from the age of the bedrocks expressed in million of years and that the LGM continental plate arrangement was similar to the modern day condition.The total energy input from the OGH forcing is 29.9 TW (TW = 10 12 W) and the mean value over the ocean is ∼88 mW.m −2 .These values are slightly below the recent estimate of the OGH: the mean OGH value is 95.9 mW.m −2 and the total global heat flux is 30 to 31 TW (Davies and Davies, 2010;Davies, 2013).This OGH forcing modifies the heat content by changing the temperature trend in the model grid boxes just above the ocean floor.
For each configuration, NEMO was integrated for ∼14,000 years and the analysis covered the last 100 model-years.At this stage of the integration, the ocean model was close to equilibrium.
The annual mean model drift both in temperature and salinity was weak, with < 0.006 • C/century (< 0.003 • C/century) below 500 meters and < 0.09 • C/century (< 0.08 • C/century) in the upper 500 meters for REF (GH).The model drift in salinity was also weak for both GH and REF, with < 0.004PSU/century below 500 meters and < 0.012PSU/century in the upper layer.

Impact of the geothermal heat flux on the stratification
The annual mean temperature trend induced by the OGH as a function of depth is shown in Fig- ures 1a,b averaged over the Atlantic and the Indo-Pacific basins, respectively.In the GH experiment, the Atlantic basin is ∼0.4 • C colder above 1,500 meters and gains heat below 1,500 meters with a maximum warming of ∼0.9 • C formed between 2,000 and 3,000 meters.In the Indo-Pacific basin, the upper 1,500 meters layer is ∼0.25 • C warmer, whereas the deep layer is up to 0.4 • C warmer.In the Atlantic basin, the heat accumulation due to the OGH below 1,500 meters has a characteristic time scale of ∼1600 years and reaches an asymptotic limit of 0.38 • C (Figure 1c).In the Indo-Pacific basin, the characteristic time scale is ∼1200 years and the accumulated heat reaches an asymptotic limit of 0.31 • C.An equilibrium is reached after ∼10,000 years.
The annual zonal mean temperature and salinity patterns in REF for the Atlantic and the Indo-Pacific basins are shown in Figures 2a,b,c,d.The deep ocean is filled with cold (near the freezing point of sea-water) and saline waters, which agrees with paleo-proxy reconstructions (Adkins et al., 2002;Adkins and Schrag, 2003) and the simulation by Brandefelt and Otto-Bliesner (2009).Relatively fresh and cold waters are found between 40 These differences in the temperature and salinity patterns modify the sea-water density (Figure 4).The AABW becomes less dense (the density decreases by ∼0.2 kg.m −3 in the Indo-Pacific, 0.3 kg.m −3 in the Atlantic) due to warming and freshening, whereas the density increases up to 0.3 kg.m −3 in the thermocline due to colder and more saline waters.The stratification (density gradient) is hence increased by ∼3% near 2,250 meters and is reduced by ∼3% near 3,250 meters (not shown).

Impact of the geothermal heat flux on the thermohaline circulation
Most paleo-climate studies investigate the thermohaline circulation in latitude-depth coordinates.A better description of the circulation is however found in latitude-density coordinates (Döös, 1994;Döös and Webb, 1994;Ballarotta et al., 2013) because fluid parcels are essentially constrained to move along density surface without doing work against gravitational force (McDougall , 1987).Ballarotta et al. (2014) have also shown that the strength of the glacial overturning strongly depends on the choice of coordinate system.Therefore, we present hereafter the meridional overturning circulation (MOC) in latitude-density coordinates, more precisely σ 4 (i.e., referenced to 4,000 meters), in order to better capture and compare the abyssal circulation.Because the OGH affects the ocean's density structure, subtracting the streamfunctions in latitude-density coordinates is meaningless.It is however possible to identify the maximum of the AABW transport at each latitude, as well as the density where the waters are formed.The MOC in latitude-depth coordinates is presented and discussed in the Appendix A for the reader who is not familiar with the MOC in latitude-density coordinates.The MOC in latitude-density coordinates also shows that the NADW is ≈ 0.1 kg.m −3 denser in GH than in REF.Associated with it, the volume of the AABW in the Atlantic basin is eroded by ∼15% in GH than in REF after 6,000 model years (Figure 6) when the layer below 1,500 meters is warmed by ∼0.3 • C. In the Indo-Pacific basin, the volume of the AABW is only slightly larger (0.3%).

Impact on the northward heat transport
The annual mean residual (computed from Eulerian mean + eddy induced velocities) northward heat transport (in PW=10 15 Watts) for the Global Ocean, the Atlantic basin and the Indo-Pacific basin in REF is shown in Figure 7.In the Indo-Pacific basin, the heat transport reaches a maxima of 1 PW and 2 PW at 14 • N and 14 • S, respectively, and is directed northward in the northern hemisphere, and southward in the southern hemisphere.In the Atlantic basin, the heat transport is directed towards the North pole at all latitude and is maximum near 22 • N (∼0.9PW).In the Southern ocean, the heat transport is less than 0.5 PW and directed towards Antarctica.The impact of the OGH on the northward heat transport is statistically significant (based on a t-test, p-value less than 5%) in the Atlantic and Southern oceans.The AABW in the Indo-Pacific basin gains geothermal heat when it spreads northward.Most of this heat (∼0.03PW) is exported to the Southern Ocean surface where it diverges towards Antarctica (∼0.02 PW) and towards the South Atlantic Ocean (∼0.05 PW) near 50 • S. The geothermal heat transported towards Antarctica then participates in the relative freshening of the Southern Ocean surface water.The geothermal heat in the South Atlantic is transported northward and reaches a maximum of ∼0.12 PW near 40 • N, where the maximum mixed layer depths are found.

Intermediate summary and Impact on the North Atlantic deep convection
Our results suggest that the impact of the OGH on the glacial ocean stratification and thermohaline circulation is significant.The OGH warms the AABW by ∼0.Consequently, the volume of AABW is reduced by 15% in the Atlantic basin.We found that this mechanism is relatively fast and has a characteristic time scale of ∼1,500 years.It corresponds mainly to an advective time scale (associated with the dynamics of the AABW) rather than a diffusive time scale.

Discussion
The ∼0.4 • C warming of the abyssal ocean due to OGH is similar to the results found in simulations of the present-day climate (Adcroft et al., 2001;Emile-Geay and Madec, 2009;Hofmann and Maqueda, 2009).However, the largest temperature difference is found between 1,500 and 3,000 meters in the Atlantic basin due to the deepening of the thermocline.This latest result coincides with Hofmann and Maqueda (2009) in their present-day climate simulations where the North Atlantic deep western boundary current warms between 0.9 • C and 1.5 • C, but it contrasts from the solution found in the simulations from Adcroft et al. (2001) and Emile-Geay and Madec (2009) where the largest warming takes place in North Pacific below 3,000 meters depth.The mechanism, explained above, is compatible with the results found in Hofmann and Maqueda (2009).Both in our study and in Hofmann and Maqueda (2009), and in opposition to Emile-Geay and Madec (2009) and other studies, the ocean surface salinity is not relaxed towards a climatology.Therefore, the warming of the abyssal waters contributes to freshening of the Southern Ocean surface waters via the advection of heat.The large formation of AABW contributes to fresher abyssal waters.Due to the closed freshwater budget and no restoring term in the sea surface salinity in the model, the Southern Ocean freshwater supply is counter-balanced by the densification of the surface waters becoming more saline in the North Atlantic and Pacific Ocean.As a result, the AMOC is reinvigorated by the in-creased surface salinity.
Turbulent mixing can also have an impact on the results.In the present study, we followed standard practice for modelling mixing bounded with a minimum background mixing level.We are aware that the mixing in our simulations might not be completely realistic for the LGM.In particular, the mixing during the LGM was probably larger than today due to the emerged continental plateaus which allow energy dissipation (Schmittner et al., 2015).As a result, larger energy in the mixing was probably supplied in the ocean interior, which contributes to erode the stratification more easily.Based on the study of Emile-Geay and Madec (2009), our abyssal overturning will be larger in a context of larger mixing.Our equilibrated state would thus be reached more quickly due to the increased amount of energy dissipation, although we do not expect a qualitative change in our results.The uncertainty due to the mixing and its variations through time is probably not larger than the overall uncertainty on the surface forcing fields or on geothermal heating.
We found that the maximum of the AMOC is ∼11-15% larger in GH than in REF, which is similar to the anomaly found in simulation of the present-day climate (Hofmann and Maqueda, 2009).This value may be considered as relatively important in light of the estimation made for the future climate scenarios (an average reduction of 25% in Meehl et al. (2007) or on short time scale.However, it is relatively weak compared to the variation of the AMOC on climate time scales, such as the 75% reduction with respect to LGM period during Heinrich stadial 1 (∼15-18.5 ka ago), the 45% reduction during the Younger Dryas stadial (∼12 ka ago) (Ritz et al., 2013), or the values found in fresh water hosing experiments under LGM conditions (> 20% reduction in Kageyama et al. (2013)).In these experiments, the AMOC changes are linked with surface processes, such as the freshwater discharge (Heinrich, 1988;Hemming, 2004) which have a stronger and faster impact on the thermohaline circulation than the processes induced by the OGH.Similar to Adcroft et al. (2001) and Emile-Geay and Madec (2009), we found that the impact of the OGH on the northward heat transport is weak (∼10%) but non-negligible, particularly in the Atlantic Ocean and in the polar regions as a result of the large scale advection of the abyssal heat content.We found that the alteration of the ocean heat transport induced by the OGH in the North Atlantic (∼0.1PW) is ∼3 times larger than the total energy input provided by OGH (0.03 PW).However, the Southern Ocean Ekman transport prevents the accumulation of OGH in the abyssal ocean.
For a salinity gradient of ∼1PSU, a temperature gradient of ∼3 • C would be required to destabilise the water column (see Appendix B).In the present study, the OGH warms by ∼0.4 • C. Therefore the OGH alone is probably not sufficient to destabilise the water column.Yet, OGH may facilitate the transition from a glacial to an inter-glacial state by reducing the volume of saline abyssal waters by ∼15% and reinvigorating the North Atlantic overturning by ∼10%, but OGH alone could not cause abrupt climate changes.

Conclusions
In the present study, we investigated the response of the ocean to the geothermal heat flux during a glacial period, such as the LGM, when the ocean circulation and stratification were different from today, and tested previously posed theories (Adkins et al., 2005)  The deep ocean circulation and the OGH hence may speed up the transition from glacial to interglacial ocean state by reducing the volume of saline abyssal waters and reinvigorating the North Atlantic overturning.However, a new steady-state is achieved only a few thousands year after OGH is applied wherein the deep stratification, albeit weakened, remains extremely stable due to the strong salinity gradient.We thus find it unlikely that abrupt climate changes could be triggered by the action of OGH alone during the LGM period.However, the OGH should contribute significantly in the transition between glacial and inter-glacial ocean states.The OGH has a strong effect on the ventilation of the abyssal ocean and might modulate the time scale of the overturning, and in turn, the rate of CO 2 release from the deep ocean to the atmosphere.
Our results are based on a forced (i.e.prescribed atmospheric conditions) ocean simulation of the LGM period.It does not account for possible ocean feedbacks on the atmosphere.Sensitivity studies with fully coupled ocean-atmosphere simulations would be useful to assess the impact of the OGH on the global climate system.
Appendix A: Impact of the geothermal heat flux on the thermohaline circulation in latitudedepth coordinates In this section, we present the annual mean residual (Eulerian mean + eddy induced velocities) meridional overturning circulation (MOC) in latitude-depth coordinates (Figure 8a,b).The structure of the LGM thermohaline circulation agrees with the recents findings derived from multiple paleoproxies (Curry, 2005;Marchitto and Broecker, 2006;Lynch-Stieglitz et al., 2007;Evans and Hall, 2008;Tagliabue et al., 2009;Gherardi et al., 2009;Lippold et al., 2012;Adkins, 2013).The circulation representative of the North Atlantic Deep Water (NADW) in the upper 2,000 meters has a maximum transport of ∼17 Sv at 900 meters depth near 35 • N. It is slightly stronger and shallower • N and 90 • N in the North Atlantic and North Pacific basins, due to the presence of sea ice.The impact of the OGH on the zonal mean temperature and salinity patterns is shown in Figures 2e,f,g,h.The temperature differences are significant at all depths except in the upper 200 meters where the temperature variability is strong and mainly controlled by the atmospheric state.The North Atlantic cooling found in Figure 1 is mainly associated with colder surface water in the Nordic Seas (up to 0.7 • C colder) and with the intrusion of colder Antarctic Intermediate Water (AAIW) in the South Atlantic basin (up to 1.3 • C colder).The deep temperatures in the Atlantic Ocean are up to 1.3 • C warmer, particularly between 1,500 and 3,000 meters in the deep western boundary current (Figure 3), and between 30 • N and 45 • N. In the Indo-Pacific basin, the layer below 1,500 meters is up to 0.4 • C warmer and the surface layer is slightly colder (0.1 • C colder) in the North Pacific basin and ∼0.3 • C warmer in the South Pacific.The salinity differences are significant at all depths and the patterns are relatively similar between the Atlantic and the Indo-Pacific basins: the Antarctic Bottom Water (AABW) is ∼0.1 PSU fresher in GH than in REF whereas the upper layer is between 0.1 and 0.3 PSU saltier.
The annual mean residual (Eulerian mean + eddy induced velocities) MOC in latitude-density coordinates is shown in Figures 5, for REF and GH.In the Southern Ocean, relatively dense waters (σ 4 > 46.2 kg.m −3 ) are formed between 60 • S and 80 • S.These waters are then transported almost adiabatically up to 40 • N. The OGH intensifies the AABW cell by 15%-30% (from 20.5 Sv in REF to 23.4 Sv in GH in the Southern Ocean, 3.0 Sv to 3.7 Sv in the Atlantic basin, 6.3 Sv to 8.4 Sv in the Indo-Pacific basin) and shifts the maximum overturning towards lighter density classes (from σ 4 ∼47.8 kg.m −3 to σ 4 ∼47.6 kg.m −3 ).The transport associated with the North Atlantic Deep Water (NADW) is ∼4 Sv stronger in GH than in REF along the 46.5-46.7 kg.m −3 isopycnals and the maximum of the North Atlantic overturning is 11% larger in GH (17.2 Sv) than in REF (15.4 Sv).
4 • C and increases the abyssal circulation between 15% and 30% north of 30 • S in the deep Pacific and Atlantic basins.The geothermally heated deep waters are advected by the deep overturning cell and upwell at the Southern Ocean surface.When reaching the Southern Ocean surface, these waters diverge near 50 • S towards Antarctica and towards the North Atlantic basin, due to the Ekman transport.The transport towards Antarctica contributes to the freshening of the surface waters.As a result, the shelf waters that contribute to the formation of AABW become fresher and the newly formed AABW becomes less saline.Due to the global salt conservation, the freshening of the AABW is compensated by more saline surface waters in the North Atlantic and North Pacific (∼0.2PSU saltier), favouring the densification and the deepening of the NADW.The largest warming is hence found in the North Atlantic deepwestern boundary current between 2,000 and 3,000 meters due to the deepening of the thermocline.
in a global ocean/ice model with realistic bathymetry.We found that the heat flux at the sea floor is a significant forcing of the deep ocean and the global thermohaline circulation.The Antarctic Bottom Water transports geothermally heated waters from the Indo-Pacific to the North Atlantic basin, indirectly favouring the deep convection in the North Atlantic and contributing to the deepening of North Atlantic Deep Water.

Figure 1 .
Figure 1.Annual mean potential temperature trend (in • C) induced by the geothermal heat forcing as a function of depth averaged in a) the Atlantic basin and b) the Indo-Pacific basin.Contour intervals are every 0.1 • C. Note the vertical scale is increased in the upper 1000 meters.c) Time-series of the mean temperature accumulation (in • C) due to the geothermal heat below 1,500 meters in the Atlantic and Indo-Pacific basins.τ1 ∼1200 years and τ2 ∼1600 years denote the characteristic time scale, i.e. the amount of time required for the response to reach (1-1/e)≈63% of the maximum heat accumulation, in the Indo-Pacific and the Atlantic basins

Figure 2 .
Figure 2. Annual zonal mean potential temperature patterns (in • C) in the reference experiment (REF) for a) the Atlantic basin, b) the Indo-Pacific basin (Contour interval every 1 • C, thick white contour is 0 • C); salinity patterns (in PSU) in the reference experiment (REF) for c) the Atlantic basin, c) the Indo-Pacific basin (Contour interval every 0.1PSU, thick white contour is 36.3PSU); the temperature difference between REF and GH for e) the Atlantic basin, and f) the Indo-Pacific basin (Contour interval every 0.1 • C, thick grey contour is 0 • C); and salinity difference between REF and GH for g) the Atlantic basin, and h) the Indo-Pacific basin (Contour interval every 0.1PSU, thick white contour is 0PSU).The thick vertical black line shows the location of the Southern Ocean entrance at 34 • S. Note the vertical scale is increased in the upper 1,000 meters.The patterns in each Southern Ocean sector are shown in each panel between 80 • S and 34 • S. The hatched regions represent the regions where the difference is insignificant at a 95% confidence level (based on a t-test).AABW: Antarctic Bottom Water, AAIW: Antarctic Intermediate Water

Figure 3 .
Figure 3. Map of the annual mean temperature difference (in • C) between GH and REF at 2,500 meters.Maximum and minimum values are denoted in the lower left corner.The largest warming is in the Atlantic deep western boundary current

Figure 5 .
Figure 5. Annual mean density-binned residual (Eulerian + eddy-induced velocities) meridional overturning circulation (in Sv) in the experiment without geothermal heating (REF) for a) the Atlantic basin, and b) the Indo-Pacific basin; and in the experiment with geothermal heating (GH) for c) the Atlantic basin, and d) the Indo-Pacific basin.The thick black line shows the location of the Southern Ocean entrance at 34 • S. The annual mean meridional overturning circulation in the Southern Ocean is shown in each panel between 80 • S and 34 • S. Positive (negative) contours represent clockwise (anti-clockwise) circulations.Contour interval is every 4Sv, and -1 and +1 Sv contours are added.Density bins intervals every 0.01kg.m−3 .Note the vertical scale is increased for σ4 > 46 kg.m −3

Figure 6 .
Figure 6.Time-series of the AABW volume variation (in %) in the Atlantic and Indo-Pacific basins.The AABW volume is computed as the volume of water below 2,000 m where the annual mean residual overturning circulation in latitude-depth coordinates is negative (Figure 8a,b).Note that the volume of the lower meridional circulation cell does not necessarily coincide with the volume occupied by Antarctic Bottom Water, because circulation boundaries do not necessarily match water-mass boundaries

Figure 7 .Figure 8 .
Figure 7. Annual mean residual (Eulerian mean + eddy-induced velocities) northward heat transport (in PW=10 15 Watts) in the Global Ocean, the Atlantic and the Indo-Pacific basins in the reference experiment (dashed line); and the difference between REF and GH in the annual mean northward heat transport (thick line).Note that the difference is magnified by a factor 10. The green dots show where the difference is significant at a 95% confidence level (based on a t-test)