Review of egusphere-2025-6172: “Nordic Overturning Increases as AMOC Weakens in Response to Global Warming”

General comments

This study investigates the joint evolution of the Atlantic Meridional Overturning Circulation (AMOC) and the Nordic Seas Overturning Circulation (NOC) under future climate change scenarios.

In a first part, a transient CO2 forcing simulation with a high-resolution climate model is analyzed to determine the changes in meridional heat transport (MHT) into the subpolar North Atlantic and Nordic Seas in response to increasing atmospheric CO2, including a decomposition into dynamic vs. thermodynamic contributions as well as overturning vs. gyre components. The authors find an increasing MHT into the Nordic Seas attributed to dynamic changes in the overturning circulation.

In a second part, a simplified 7-box model of the Atlantic ocean circulation is specified and used to study the impact of surface warming and freshwater hosing on the overturning circulation. A two-part response is found: first, the AMOC decreases while the NOC slightly increases; second, both the AMOC and NOC collapse as convection shuts down. This behavior is interpreted through an intuitive mechanism relating the decreasing subpolar Atlantic surface density to both the Atlantic density further south and the Nordic Seas density further north.

Overall, this study is a novel and valuable contribution to ongoing debates about a potential AMOC decline/shutdown and the fate of the North Atlantic circulation more generally. The methodology is mostly sound and the results are presented in an understandable fashion, except where noted below. While the Discussion mentions various caveats for the box model results, the climate model results should also be handled with care.

I can recommend this manuscript for publication in Ocean Science after the comments below have been addressed.

Specific comments

Abstract: given the limitations of the box model (appropriately discussed in the Discussion section, Ln. 321-351) and of the climate model simulation (not yet discussed), the Abstract should at least briefly acknowledge these caveats or perhaps tone down some of the confident language.

Ln. 30: for completeness, this paragraph should mention that the historical evolution and potential weakening of the AMOC is currently still a subject of scientific debate (whereas a future AMOC decline is fully expected). In particular:

• Worthington, E. L. et al. A 30-year reconstruction of the Atlantic Meridional Overturning Circulation shows no decline. Ocean Sci. 17, 285–299 (2021). https://doi.org/10.5194/os-17-285-2021
• Latif, M., Sun, J., Visbeck, M. et al. Natural variability has dominated Atlantic Meridional Overturning Circulation since 1900. Nat. Clim. Chang. 12, 455–460 (2022). https://doi.org/10.1038/s41558-022-01342-4
• Terhaar, J., Vogt, L. & Foukal, N.P. Atlantic overturning inferred from air-sea heat fluxes indicates no decline since the 1960s. Nat Commun 16, 222 (2025). https://doi.org/10.1038/s41467-024-55297-5

Similarly, the past evolution of the Nordic MOC should be briefly mentioned for context, e.g.

• Rossby, T., Chafik, L., & Houpert, L. (2020). What can hydrography tell us about the strength of the Nordic Seas MOC over the last 70 to 100 years?. Geophysical Research Letters, 47, e2020GL087456. https://doi.org/10.1029/2020GL087456

Ln. 36: “the AMOC has demonstrated the capacity to...”: since this is referring to model results, care should be taken to use wording that does not imply any directly observed “capacity” of the real-world AMOC

Ln. 48-49: “This increased overturning...”: this sentence is too vague, it is unclear where exactly those warm and salty waters appear, and what kind of changes in water mass transformation and circulation the authors have in mind

Ln. 88: “initial simulation”: does this refer to a spin-up simulation to quasi-equilibrium before starting the preindustrial and 2xCO2 experiments? If so, that should be stated, since on Ln. 86 it is only said that “the model is initialized” with certain conditions, which does not sound like a proper spin-up simulation.

Equation (1): introduce all symbols and the values of the constants

Sections 2.2 - 2.3: it could be good to cite Msadek et al. (2013) which seems share a lot of the methodology used here:

• Msadek, R., W. E. Johns, S. G. Yeager, G. Danabasoglu, T. L. Delworth, and A. Rosati, 2013: The Atlantic Meridional Heat Transport at 26.5°N and Its Relationship with the MOC in the RAPID Array and the GFDL and NCAR Coupled Models. J. Climate, 26, 4335–4356, https://doi.org/10.1175/JCLI-D-12-00081.1.

Fig. 1 caption: “winter” in what hemisphere (what months specifically)? And why winter?

Equation (5): what are the values of the constants in this equation?

Ln. 141: “the light blue arrow flips”: specify which arrow exactly, since several of them are light blue

Box model equations (Eqs. 6-21): this collection of equations is a bit hard to parse. I would suggest putting it in an Appendix to save some space in the main text. Also, since each term in each equation belongs to one of several categories (i.e., restoring terms with lambda, overturning terms with m, gyre terms with K, convection terms with c, diffusion terms with eta), it could be beneficial to align these terms vertically so that e.g. all terms involving lambda are vertically aligned, then all overturning terms with m, etc. Lines with e.g. no restoring term (for boxes below the surface) would then just have empty space in the restoring column, if that makes sense.

Maybe this would take up too much space but it would certainly increase the readability of the model equations.

Ln. 195: “The weights for the salinity and temperature differences are the values of α and β in the linear EOS.” This sounds intuitive, but is there really a physical motivation to choose the loss function weights in this way? I would assume that the weights are arbitrary hyperparameters that could in principle be tuned, even if the chosen values (1.7 and 7.9) may happen to give good results.

Uncertainties (section 3.1): all values reported in tables 1-3 come without uncertainty estimates/confidence intervals. This makes statements about particular decompositions and “dominant” processes etc. less convincing. If more than one ensemble member has been run for this configuration, the ensemble spread should be stated. If similar runs from other climate models are available (e.g., from the CMIP archive), these could be used to get some sense of the variance in these results. Otherwise, the Discussion section should fully discuss these caveats, not only the limitations of the box model.

Ln. 228-229: “Hence, the findings imply an anticorrelation between AMOC and NOC-driven MHT changes under increased CO2 forcing in CM2.6.” Since there are really only single data points presented in these tables (e.g. a decreasing AMOC and an increasing NOC), “anticorrelation” might not be the most accurate term. It would be interesting to test whether the evolution of the AMOC and the NOC are correlated in time on decadal (or even inter-annual) time scales in CM2.6, as well as across hosing/warming experiments in the box model in later sections.

Table 2: the heat budget terms for the subpolar North Atlantic contain a large compensating effect of the dynamic vs. nonlinear term (-0.36PW vs. +0.35PW). Is it thus really possible to interpret that circulation changes “dominate” (Ln. 221 and 228)? It could also be said that nonlinear effects dominate, or that dynamic and nonlinear effects cancel and the thermodynamic effect dominates.

Ln. 238: “...the weakening of the overturning circulation is the dominant factor...”: I would argue that a 57.5% contribution is not very “dominant”, especially since no uncertainty bounds are stated (see comment about uncertainties further above).

Ln. 243: It would be helpful to have a paragraph briefly summarizing the climate model results before starting the box model section.

Ln. 252: “...the imposed warming accelerates convection shutdown due to decreasing surface buoyancy”: this is not apparent from Fig. 3a/c/e, it seems that the high- and low-end warming scenarios (red and light blue lines) bracket the response of the intermediate warming scenarios. Additionally, the red high-end warming line seems to behave qualitatively differently from the other scenarios. In any case, the differences due to different warming rates seem to be relatively small.

Ln. 314: “warming in the Nordic Seas”: it appears that Li and Liu (2025) are concerned more with the subpolar gyre warming hole than with Nordic Seas warming? See their Fig. 3a

Ln. 371: “warming and freshening of the northern Atlantic caused by an AMOC slow-down” Since both warming and freshening are actually imposed in the box model, to what extent can one really conclude that the mechanism involves the AMOC itself driving further warming and freshening? Can this feedback contribution be quantified?

Technical corrections

Typesetting: the LaTeX could be improved in a number of places:

• Uses of MHT’ and similar “math terms” in the text should be placed inside an inline equation ($...$)
• 2xCO2 should use \times and non-italic letters
• Some citations should not be in parentheses: Ln. 133, 189, 198, 205

Ln. 1: “while Nordic...” -> while the Nordic...

Ln. 2: “increase” -> strengthen

Ln. 23: “winter temperatures”: specify that this refers to (surface) air temperature

Ln. 33: “above 90% significant”: the wording should be improved

Ln. 35: “hosing” -> freshwater hosing

Ln. 41: “southern shift” -> southward shift

Ln. 45: “to name a few”: this could be removed

Ln. 79-80: “much greater fidelity, ... more realistic”: compared to what? Other models of the same generation/resolution?

Ln. 106-107: “In the following, ...”: this sentence is redundant with the previous sentence and could be removed

Ln. 127: refer to Fig. 2 after “... eight boxes”

Ln. 185: “The parameter c”: there are actually several parameters c_i

Ln. 185: “The parameter c...”: although I can understand the meaning of this sentence, it would be easier to understand when put in equation form (with a large brace containing the value of c_i in the two cases)

Ln. 191: “eta is the heat diffusion constant”: it is also responsible for salt diffusion (Eq. 21)

Ln. 214: “surface heat flux”: specify the sign convention

Fig. 3 caption: “Grey lines”: since they are all grey, it is not possible to tell which line belongs to which warming rate.

Ln. 261: “because of the nonlinearity in the equation of state”: expand on this a bit, I would assume this refers to the temperature dependence of alpha and the different temperatures in the two regions.

Ln. 264: “salt advection feedback”: the accumulation of freshwater in the northern Atlantic after an AMOC decline is not yet a feedback, it is only one of the two “arrows” in the feedback loop. I would thus remove this parenthesis.

Ln. 290: “warming prevents the decline of the NOC”: it only “prevents” it in the warming-only experiments. The NOC does end up declining in the hosing+warming experiments, so a more accurate term could be “delays” or similar.

Ln. 304: “Labrador Sea”: it looks like the box covers the Irminger Sea more than it does the Labrador Sea?

Ln. 340: “dsicussed”: typo

Ln. 355: “in northern Atlantic” -> in the northern Atlantic

Ln. 358: “5 mSv/year”: should this be 4 mSv/year instead?

This paper describes and analyses the response of the Northern overturning circulation (NOC) in response to global warming showing an initial strengthening while the AMOC further south weakens. This study makes use of the high-resolution coupled model CM2.6, including an eddy-permitting, if not eddy-resolving ocean component. They compare an 80-year PIC-run with a 1%-increase of CO2 run, which reaches doubled concentrations after 70 years after which the simulation is extended with 10 years at doubled CO2 concentrations. In addition, they use a box model, containing of 8 reservoirs (boxes), performing additional (hosing and hysteresis) experiments to provide support to the coupled climate model results in terms of the physical mechanisms at play.

This is a nice study showing a both physically interesting and potentially impactful result that has not attracted a lot of attention hitherto. The analyses performed are thorough and the presentation is clear and logical. Nevertheless, I have a few comments/recommendations.

1. The CM2.6 runs are, of course, not fully equilibrated. I don’t think that this is an issue, as the NOC-strengthening during AMOC decline is mostly a transient phenomenon, unless the AMOC equilibrates at medium-strength, purely driven by deep convection in the Nordic Seas, while convection in the Labrador/Irminger Seas remains collapsed. What I miss are timeseries of AMOC at say 26 and 45N and a relevant NOC-metric, together with maximum (March) mixed-layer depths from the Nordic Seas convection site. At present there is hardly any information on how CM2.6 reacts on the forcing, apart from one figure (5) showing density changes averaged over the last 20-years of the run. Also, I miss info how long (if at all) the model was spun-up before the two 0-year runs were initialised. To what extent are the NOC-strengthening and AMOC decline still further developing or s there already a plateau in AMOC decline and onset of NOC weakening at the end of the runs?
2. The authors claim that for the opposing trends between AMOC and NOC so far, no causal links have been proposed. This is not strictly true. A strengthening of the NOC concurrent with decreasing convection in the Labrador/Irminger Seas, resulting in a very weak AMOC response to freshwater hosing was discussed in “Circulation adjustment in the Arctic and Atlantic in response to Greenland and Antarctic mass loss by van den Berk et al (2021);

https://doi.org/10.1007/s00382-021-05755-3

In this article it was argued that the differential SST-response between Nordic Seas and SPG in response to freshwater hosing, induce an atmospheric response, enhancing the westerlies which, through windstress curl, drive an enhancement of the Nordic Seas gyre with more upward doming of isopycnals in the center, leading to deeper mixed layers through decreased stratification there.  This coupled ocean-atmosphere mechanism cannot be described by the box model they used, but could be checked to play a role, or not, in CM2.6, and should, even if it does not play a role in that model, be discussed. A seesaw response Between Labrador Sea and Nordic Seas in response to the NAO has been described before in many papers and is both consistent with the Van den Berk mechanism and possibly also relevant to the present study.