Generalized stability landscape of the Atlantic Meridional Overturning Circulation

Willeit, Matteo; Ganopolski, Andrey

doi:https://doi.org/10.5194/egusphere-2024-1482

Preprints

https://doi.org/10.5194/egusphere-2024-1482

Preprints

23 May 2024

| 23 May 2024

Generalized stability landscape of the Atlantic Meridional Overturning Circulation

Matteo Willeit and Andrey Ganopolski

Abstract. The Atlantic Meridional Overturning Circulation (AMOC) plays a crucial role in shaping climate conditions over the North Atlantic region and beyond and its future stability is a matter of concern. While the stability of the AMOC to surface freshwater forcing (FWF) has been investigated in numerous model simulations, its equilibrium response to changing CO₂ remains largely unexplored and precludes a comprehensive understanding of AMOC stability under ongoing global warming. Here we use a fast Earth system model to explore the stability of the AMOC to combined changes in FWF between -0.25 and +0.25 Sv in the North Atlantic and atmospheric CO₂ concentrations between 180 and 560 ppm. We find four different AMOC states associated with qualitatively different convection patterns in the North Atlantic. Apart from an Off AMOC state and a Modern-like AMOC with deep water forming in the Labrador and Nordic Seas, we find a Weak AMOC state with convection occurring south of 55° N and a Strong AMOC state characterized by deep water formation extending into the Arctic. Several of these AMOC states can be stable under the same boundary conditions for specific combinations of CO₂ and FWF. Generally the model shows an increase in equilibrium AMOC strength for higher CO₂ levels. It is also noteworthy that, while under preindustrial conditions the AMOC off state is not stable in the model, it becomes stable for CO₂ concentrations above ~400 ppm, suggesting that an AMOC shutdown in a warmer climate might be irreversible.

Received: 17 May 2024 – Discussion started: 23 May 2024

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Matteo Willeit and Andrey Ganopolski

Status: closed

CC1:
'Comment on egusphere-2024-1482', Justin Gérard, 28 Jun 2024

We would like to thank Willeit and Ganopolski for having considered or work [Gérard and Crucifix, 2021, 10.5194/esd-15-293-2024, 2024] for their analysis of generalised stability landscape of the Atlantic Meridional Overturning Circulation.

In their section 3, ll. 91-95 the authors note, referring to us, that "recently analyzed the AMOC response to a slow CO2 increase and found a gradual AMOC weakening and eventual collapse at CO2 above ∼1500 ppm. However, their rate of change of CO2 of 0.14 ppm yr−1 is much larger than the rate that we use in this study (∼0.005 ppm yr−1) and it has been shown that the AMOC response is highly sensitive to the applied rate of temperature change.".

Hereafter, we provide the maximum values of the overturning streamfunction, as measured in our paper, comparing the values reached at equilibrium with those obtained in the hysteresis experiments. This is valid in the experimental conditions described in our paper and simply serves as additional information for the authors.
[CO2] (ppm) Equilibrium (Sv) Transient (Sv)
280 14.44 14.44
560 12.86 12.38
840 11.93 11.58
1120 11.21 11.05
1400 9.41 9.58
We observed that, in this particular experimental setup, differences between values obtained at equilibrium and those of the transient experiments are modest. Crucially, in cGENIE and, again, in the modern continental configuration as described in our paper, the maximum or the overturning streamfunction decreases as CO₂ increases. A shutdown occurs at 6xCO₂.

Justin Gérard and Michel Crucifix

Citation: https://doi.org/10.5194/egusphere-2024-1482-CC1
- AC1: 'Reply on CC1', Matteo Willeit, 11 Jul 2024
  
  We are very grateful to Gérard and Crucifix for their comment, which provides us with important information. When we wrote our paper, we were not entirely sure that we could directly compare our results with those of Gérard and Crucifix (2024), as their rate of change of CO₂ when tracing the stability diagram in the CO₂ phase space was rather small, but still much larger than in our study. However, now Gérard and Crucifix have presented additional modelling results confirming that the AMOC dependence on CO₂ shown in their Fig. 7a very closely represents the equilibrium response. Thanks to the commentary by Gérard and Crucifix, we can now make a definite statement about the differences in our results.
  
  In their model, the quasi-equilibrium AMOC weakens with increasing CO₂, which is opposite to our modelling results. Such a difference between results of different models is not surprising. While the oceanic components of our models are similar (but not the same; in the GOLDSTEIN ocean model version used in CLIMBER-X we have made a number of changes as described in Willeit et al. (2022), the most noteworthy being the implementation of a less diffusive advection scheme), the atmospheric components are very different. cGENIE model used by Gérard and Crucifix is a 2D energy-moisture balance model, while CLIMBER-X uses a more dynamical quasi-3D statistical-dynamical atmospheric model.
  
  Citation: https://doi.org/10.5194/egusphere-2024-1482-AC1
RC1:
'Review of Willeit and Ganopolski’s “Generalized stability landscape of the Atlantic Meridional Overturning Circulation” by Yvan Romé, University of Leeds', Yvan Romé, 05 Jul 2024
Review of Willeit and Ganopolski’s “Generalized stability landscape of the Atlantic Meridional Overturning Circulation” by Yvan Romé, University of Leeds

General comments

The manuscript explores the combined effect of changes in CO2 concentrations and North Atlantic freshwater forcing on the existence of multiple AMOC modes in the fast climate model ClimberX. After plotting the AMOC intensity hysteresis cycles resulting from the two independent parameters and including discussions about the role of the initial AMOC state and the rate of change, Willeit and Ganopolski produced a stability landscape of the AMOC modes in the input space formed by the CO2 concentrations and the freshwater forcing.

This work offers a new multidimensional approach to understanding AMOC stability in climate models. The stability landscape map is a convincing and comprehensive way to explore the domain of occurrence of AMOC mode shifts. The volume of simulations produced for this article is remarkable, and the experimental design allows advanced conclusions on the effect of AMOC mode shifts on the climate.

I have, nonetheless, major reservations about the clarity of the manuscript, as well as the justification of some arguments. The manuscript is, at times, difficult to read, and the description of the experiment and the calculations need to be revised to be able to evaluate the validity of the conclusions.

In summary, this paper is a strong contribution to the highly relevant question of AMOC manuscript in climate models. However, some work is required on the text to support the interpretations and conclusions. I recommend major revisions of the manuscript before publication in ESD. The main concerns I identified are the following.

The abstract is not reflective of the work and inconsistent with the conclusion, which is very clear. In particular, the abstract implies that the main aim of the paper is the impact of CO2 concentrations on AMOC stability, instead of, as it is written in the conclusion, performing “a systematic analysis of the AMOC stability in the FWF–CO2 space.”

In the introduction, the impact of CO2 on the AMOC stability is said to “remain largely unexplored”, and the freshwater forcing outside of the 20-50°N band to be a rare occurrence. I believe both of these views are outdated, and the introduction is missing key references and discussion points that provide an accurate and comprehensive picture of the current state of the research. If these comments only concerned modern days, it should clearly be stated, and the results from the palaeo community should be discussed. I recommend splitting the second paragraph of the introduction into three parts. A first one on the CO2 effect, including missing references (e.g. Brown and Galbraith 2016, Zhang 2017, Klockmann 2018, Vettoretti 2022), a second one on the FWF including missing references (e.g. Smith and Gregory 2009, Roche 2010, Kageyama 2013, Ivanovic 2018, Romé 2022) and a third one on the need for combined CO2xFWF analysis and an introduction of your paper, which is currently too short.

I do not believe that the authors can claim to be the first to attempt to draw a landscape of AMOC stability in the CO2 x freshwater forcing space is true, see Brown and Galbraith 2016 for instance. However, I would say that this paper presents the most comprehensive and robust method up to date. If this claim only applies to modern-day studies of the AMOC, it needs to be clearly stated and put in context with palaeo studies.

Significant mode shifts and overshoots on the hysteresis cycles are not discussed in the text. In particular, in Figure 1, the transition in the red solid line around 0.05Sv is remarkable: Is it different from an overshoot? Why is it sustained for about 1000 years? Could this be an occurrence of millennial-scale variability? Could you link this to Willeit 2024?

The definition of the different states comes too late in the paper and lacks precision. How do you define the different modes, using the AMOC index, the mixed layer depth or manually? Additionally, How do you calculate the AMOC index? What references did you use for the modern-day deep water formation sites, and how do they compare to your modern mode? Over what time slices was Figure 3 plotted?

The interpretation of the freshwater flux needs to be clarified in this article, and it becomes a problem when comparing pre-industrial to modern conditions. Would it not be more accurate to account for changes in CO2 and freshwater forcing when comparing the two? Otherwise, what is the point of using a two-dimensional landscape? In addition, the following statement from the conclusion “Our results indicate a generally stronger and deeper AMOC at equilibrium under warmer climate conditions. This is in contrast to the projected AMOC weakening response to anthropogenic global warming […]” is only valid if one considers the sole CO2 effect, but freshwater forcing is expected to increase with Greenland melt, which could take us into a region of the landscape where all four modes exist. I think the comparison between past, present and future states should include a discussion about the role of excess freshwater induced by ice sheet melting.

The details about the construction of the stability landscape is lacking precision and its validity cannot be evaluated. This all the more important as you highlighted the dependence of the direction of variation in Figure 1,2 and B1.

Specific comments

Abstract

L8 - Can you briefly define the OFF and Modern AMOC states?

L11-12 (“In general, the model shows an increase in equilibrium AMOC strength for higher CO2 levels.”) - This does not reflect the actual nature of the work, which goes way beyond this sole aspect. This statement is valid for the standard CO2 experiment in Figure 2, but not consistent with Figure 4 (ex. in Figure 4a, an increase of CO2 can trigger a weak mode). The abstract needs not to focus only on the CO2 experiment but also on the stability landscape.

Introduction

L28 (“There is no consensus as to whether the AMOC is in a monostable or a bistable regime under present climate conditions”) - This needs a reference; I am aware of discussions on the potential weakening of the AMOC, less so about the current state of the AMOC.

Results

L59 (“In particular, there is a range of FWF over which the AMOC has two stable states has two stable states”) – Can you be precise about the range of FWF you are talking about? It also depends on your definition of stability, as I would argue that the dip around 0.05 Sv in Figure 1 is a sign of instability.

L61 - Does “preindustrial conditions” mean 0 Sv in this case? Also, according to the methods, this experiment has pre-industrial CO2 concentrations but a modern-day ice sheet. I would be more careful about using “pre-idustrial” conditions, what about "initial state" instead?

L63 (“suggesting a prominent role of convective instability”) - Could you show that this is a convective instability, showing deep water formation sites activity, for example?

L65 - “This is the result of a collapse of deepwater formation […] of observed past Dansgaard-Oeschger events.” : Here again, the manuscript is missing a plot with the deep water sites dynamics to verify this statement, and a reference about convection in D-O records.

L67 – A definition of what the authors mean by Off, Weak, Modern-day, Strong is needed at this point of the paper.

L82 (“which is possibly more relevant for the ongoing global warming”) - I disagree, both CO2 and meltwater discharge are relevant to future climate changes.

L85-93 - I find the wording of this section confusing. Is the rate of CO2 increase in this paper slower than the “slow” increase in Gérard and Crucifix? How do you explain that you see a strengthening of the AMOC when Gérard and Crucifix 2024 saw a decrease? Could this simply mean that the AMOC response to CO2 is highly uncertain and model dependent?

L105 (“For CO2 above ∼250 ppm, the convection pattern resembles the present-day state with deep water forming in the Labrador Sea and in the Nordic Seas”) - Are you talking about Willeit et al. 2024 or this manuscript?

L156 (“If the climate would be in equilibrium with present-day CO2 concentrations of ∼420 ppm, the model suggests that the Modern AMOC state would not be stable, but that the AMOC would rather be in the Strong state instead”) – Back to the point about the interpretation of future freshwater forcing, would the accelerated melt of Greenland not move the system along the Modern AMOC conditions diagonal instead? Otherwise, this is a major caveat of the analysis that needs to be discussed (although, arguably, the last paragraph of the discussion introduces this idea of model dependency).

Figure 4 - Good Figure but missing information about the criteria used for the clustering of the AMOC states. Is it purely based on the AMOC index, or on the mixed layer depth?

Figure 5 - I think this is the most important Figure of the paper and it deserves to be bigger. Also, why are the Off mode treated differently than the different modes? It makes the Figure difficult to interpret at first reading. Could you use different colours or shading and include all four states in the stability landscape?

L181 - This is a good section and a convincing way to perform this analysis. However, as you showed, the sea ice extent is driving the most significant temperature changes, so I think it would be good to conclude on the impact of less extensive Arctic sea ice in future climate on these results.

Discussion

L186 - Can you say a word about the Modern and Off AMOC modes.

SI

L256 - The paths used in Figure A1 need to be clearly explained in the Figure or the text, otherwise it is impossible to validate the protocol in Figure 4. Do they correspond to the black or the green arrows? Do they include the standard experiments?

Figure B2 does not have a caption.

Technical corrections

Introduction

L17 - Do you have an example of a “societal” change?

L19 - Add Bellomo 2021 for state-of-the-art climate model references.

Results

L51 – “Willeit et al. (2020)” to “(Willeit et al., 2020)

L69 (“most AMOC hysteresis experiments to date have been performed with FWF at lower latitudes (usually between 20N and 50N)”) – Needs references.

L71 (“In our hysteresis experiment”) – Change to “In Figure 1”, there are multiple hysteresis experiments in this paper.

L80 (“give the wrong impression that the AMOC Off state is also stable”) – Is it a wrong impression or relative to the rate of meltwater discharge? Can you give an order of magnitude of the expected meltwater discharge for future melting or during a D-O event/Heinrich event?

L83 – “In an experiment where” to “In Figure 2”?

L94 (“two discrete transition”) - what do you mean by discrete? Abrupt?

Figure 2 - It is difficult to distinguish the solid and dotted lines on this Figure. Maybe two columns?

L110 (“but this has been shown to not be a requirement for the existence of millennial-scale variability”) – Vague, can you say more?

Figure 3 - The Figure needs to be bigger. Please add labels to the colour-bars. Does seasonal mean winter in this case?

L130 - “net freshwater flux” to “net surface freshwater flux”.

L133 (“for CO2 doubling”) – I would say “for double the amount of CO2” because you are not doing a CO2 doubling experiment, which, as you highlighted, could have a different impact on the AMOC.

L139 - “in the model the net surface freshwater flux into the whole Atlantic Ocean shows the opposite trend” to “the net surface freshwater flux into the whole Atlantic Ocean shows the opposite trend in our model”?

L144 - remove “a stabilizing effect”?

L150 - “can be investigating by tracing their stability through” to “can be investigated by tracing the AMOC response in”

L169 - “to explore the pure effect of the different AMOC states on climate” to “to isolate the effect of the changes of AMOC states on the climate”.

Figure 6: Can you use a different colour scheme for the temperature?

Discussion

L193 - “explains” to “demonstrates”?

L202 (“anthropogenic global warming”) - Add Bellomo 2021 to the references for state-of-the-art climate models.

SI

Figure B1: Hard to distinguish the solid and dotted lines on this Figure. Maybe two columns?
Citation: https://doi.org/10.5194/egusphere-2024-1482-RC1
- AC2: 'Reply on RC1', Matteo Willeit, 20 Aug 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1482/egusphere-2024-1482-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2024-1482-AC2
RC2:
'Comment on egusphere-2024-1482', Anonymous Referee #2, 18 Jul 2024
Review comments on egusphere-2024-1482
I thank the author team for their interesting contribution. The study finds an interesting AMOC stability landscape in CLIMBER-X by systematically varying freshwater forcing and atmospheric CO2. I have not seen such a comprehensive study of AMOC stability across varying backgrounds before, and I think it provides an intriguing template to move beyond single-perturbation AMOC studies and broaden our understanding of the processes affecting AMOC stability on multi-millennial time scales. This merits publication in itself, but I think the authors could increase the relevance of their manuscript by expanding their discussion of the implications of their findings for our understanding of past or future AMOC states. I recommend publication with minor corrections.
Main points
I did not understand how many and which simulations were run in total to explore the CO2-FWF space. Was each combination of CO2 and FWF run for initial ‘on’ and ‘off’ states? Were the ‘pathway’ simulations run transiently, or iteratively run into equilibrium?

The existence of a ‘strong’ AMOC mode above 370 ppm is a key result of the study. The authors already compare their results to other model studies and mention that it shows the potential for stronger-than-present AMOC states in the past. Given that this CO2 threshold is close to both current and Pliocene CO2 concentrations, and that the Arctic might have been seasonally ice-free in the Eemian, it would be interesting to discuss this result also in the context of observations and proxy data. Is this solution purely theoretical or is there evidence that such a circulation pattern has existed in the past?

Besides the ‘strong’ AMOC state, the study shows that the ‘off’ state is also stable for high CO2 concentrations. If I understand correctly, this is the conclusion from simulations that were initialised with an ‘off’ state at high CO2. What forcing is required to transition from a ‘strong’ AMOC state into an ‘off’ state under high CO2?

What forcing is required to tip from a ‘modern’ into a ‘strong’ AMOC? What are the climatic impacts of this shift? Do the authors think that tipping into an ‘off’ or ‘strong’ AMOC is possible under persistent anthropogenic climate change?

Minor points
Page 12: Please add an explanation of the grey and black arrows and the black dots to the figure caption or a legend.
Page 14: Please add a caption for Fig B2.
Citation: https://doi.org/10.5194/egusphere-2024-1482-RC2
- AC3: 'Reply on RC2', Matteo Willeit, 20 Aug 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1482/egusphere-2024-1482-AC3-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2024-1482-AC3
RC3:
'Comment on egusphere-2024-1482', Anonymous Referee #3, 22 Jul 2024

Review of the ms. “Generalized stability landscape of the Atlantic Meridional Overturning Circulation” by Matteo Willeit and Andrey Ganopolski”
Using a fast, low-resolution Earth system model, the authors present a systematic and comprehensive exploration of the dynamical stability landscape of the AMOC under quasi-equilibrium forcing conditions. Notably, AMOC stable states are explored in a combined phase space of atmospheric CO2 concentration and North Atlantic freshwater forcing. One of their main results is the possibility for a presence of more than two stable AMOC states if the present-day CO2 concentration and freshwater forcing are run into equilibrium. Among these, there is a stable “off” state, but also a stable “strong” state with deep water formation in the Kara Sea.
This study is a very timely and useful contribution to the discussion about the near-term evolution of the AMOC – a discussion that has very recently gained much traction again. Knowledge of AMOC stability in quasi-equilibrium climates – as opposed to the fast-changing transient that is our reality – is essential context for the scientific understanding of the latter.
The paper is very well written and has a clear layout. I offer a few comments for the authors to consider while recommending the ms. for publication without a further round of reviews.
Comments
The study explores a range of CO₂ concentrations between 180 and 560 ppm. Of course this is a reasonable choice given the pre-industrial concentration. Sadly though, it is not entirely unlikely that the Earth System might have a CO₂ concentration larger that 560 ppm in the foreseeable future. Could the authors comment on why they didn’t explore that part of the phase space?
CLIMBER-X does not have internal interannual climate variability, as stated on l.220. The presence of such variability could lead to some states being indistinguishable (as said on l.222). Another possibility is that a stable state, while technically present, is occupied only with a very small probability (Monahan [2002], JPO 32, 2072-2085).
For the three quantities plotted in Figure 2 we need the spatial context. Where, in e.g. latitude and depth, did you diagnose the AMOC maximum and the AMOC heat transport maximum? Is the delta net FW in panel (c) diagnosed over the entire surface of the Atlantic basin?
I have one concern that I’d like the authors to comment on. Figure 3g tells us that the overturning streamfunction of the “modern” AMOC state shows a complete absence of the abyssal AABW (Antarctic bottom water) cell. We know that in reality the AABW cell is present (Talley et al. [2003], J. Clim. 16, 3213-3226; Johnson [2008], DOI 10.1029/2007JC004477). Does that imply that the CLIMBER-X phase space significantly deviates from present-day conditions? Or should we say that the state labelled “weak” here is the actual modern-day state? Fig. 3f, after all, displays the correct shape of the streamfunction, and Fig. 2a suggests that the amount of overturning is realistic in this “weak” state too. If the model did have variability, perhaps it would convect in the correct regions too?
Could you consider a different type of colour scale for Figure 6? Currently it’s really hard to work out the displayed colling patterns quantitatively using the small hue difference from one colour level to the next.
When discussing Figure 6 (around lines 170 to 180), could it be worthwhile mentioning that an AMOC transition as in Figure 2a, driven by increasing CO2 concentration, would trigger a Modern to Strong transition, and thus a warming actually, with the negative of the patterns in Figure 6 b, f, j?
I think for Figure B2 no legend was provided.

Citation: https://doi.org/10.5194/egusphere-2024-1482-RC3
- AC4: 'Reply on RC3', Matteo Willeit, 20 Aug 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1482/egusphere-2024-1482-AC4-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2024-1482-AC4

Status: closed

CC1:
'Comment on egusphere-2024-1482', Justin Gérard, 28 Jun 2024

We would like to thank Willeit and Ganopolski for having considered or work [Gérard and Crucifix, 2021, 10.5194/esd-15-293-2024, 2024] for their analysis of generalised stability landscape of the Atlantic Meridional Overturning Circulation.

In their section 3, ll. 91-95 the authors note, referring to us, that "recently analyzed the AMOC response to a slow CO2 increase and found a gradual AMOC weakening and eventual collapse at CO2 above ∼1500 ppm. However, their rate of change of CO2 of 0.14 ppm yr−1 is much larger than the rate that we use in this study (∼0.005 ppm yr−1) and it has been shown that the AMOC response is highly sensitive to the applied rate of temperature change.".

Hereafter, we provide the maximum values of the overturning streamfunction, as measured in our paper, comparing the values reached at equilibrium with those obtained in the hysteresis experiments. This is valid in the experimental conditions described in our paper and simply serves as additional information for the authors.
[CO2] (ppm) Equilibrium (Sv) Transient (Sv)
280 14.44 14.44
560 12.86 12.38
840 11.93 11.58
1120 11.21 11.05
1400 9.41 9.58
We observed that, in this particular experimental setup, differences between values obtained at equilibrium and those of the transient experiments are modest. Crucially, in cGENIE and, again, in the modern continental configuration as described in our paper, the maximum or the overturning streamfunction decreases as CO₂ increases. A shutdown occurs at 6xCO₂.

Justin Gérard and Michel Crucifix

Citation: https://doi.org/10.5194/egusphere-2024-1482-CC1
- AC1: 'Reply on CC1', Matteo Willeit, 11 Jul 2024
  
  We are very grateful to Gérard and Crucifix for their comment, which provides us with important information. When we wrote our paper, we were not entirely sure that we could directly compare our results with those of Gérard and Crucifix (2024), as their rate of change of CO₂ when tracing the stability diagram in the CO₂ phase space was rather small, but still much larger than in our study. However, now Gérard and Crucifix have presented additional modelling results confirming that the AMOC dependence on CO₂ shown in their Fig. 7a very closely represents the equilibrium response. Thanks to the commentary by Gérard and Crucifix, we can now make a definite statement about the differences in our results.
  
  In their model, the quasi-equilibrium AMOC weakens with increasing CO₂, which is opposite to our modelling results. Such a difference between results of different models is not surprising. While the oceanic components of our models are similar (but not the same; in the GOLDSTEIN ocean model version used in CLIMBER-X we have made a number of changes as described in Willeit et al. (2022), the most noteworthy being the implementation of a less diffusive advection scheme), the atmospheric components are very different. cGENIE model used by Gérard and Crucifix is a 2D energy-moisture balance model, while CLIMBER-X uses a more dynamical quasi-3D statistical-dynamical atmospheric model.
  
  Citation: https://doi.org/10.5194/egusphere-2024-1482-AC1
RC1:
'Review of Willeit and Ganopolski’s “Generalized stability landscape of the Atlantic Meridional Overturning Circulation” by Yvan Romé, University of Leeds', Yvan Romé, 05 Jul 2024
Review of Willeit and Ganopolski’s “Generalized stability landscape of the Atlantic Meridional Overturning Circulation” by Yvan Romé, University of Leeds

General comments

The manuscript explores the combined effect of changes in CO2 concentrations and North Atlantic freshwater forcing on the existence of multiple AMOC modes in the fast climate model ClimberX. After plotting the AMOC intensity hysteresis cycles resulting from the two independent parameters and including discussions about the role of the initial AMOC state and the rate of change, Willeit and Ganopolski produced a stability landscape of the AMOC modes in the input space formed by the CO2 concentrations and the freshwater forcing.

This work offers a new multidimensional approach to understanding AMOC stability in climate models. The stability landscape map is a convincing and comprehensive way to explore the domain of occurrence of AMOC mode shifts. The volume of simulations produced for this article is remarkable, and the experimental design allows advanced conclusions on the effect of AMOC mode shifts on the climate.

I have, nonetheless, major reservations about the clarity of the manuscript, as well as the justification of some arguments. The manuscript is, at times, difficult to read, and the description of the experiment and the calculations need to be revised to be able to evaluate the validity of the conclusions.

In summary, this paper is a strong contribution to the highly relevant question of AMOC manuscript in climate models. However, some work is required on the text to support the interpretations and conclusions. I recommend major revisions of the manuscript before publication in ESD. The main concerns I identified are the following.

The abstract is not reflective of the work and inconsistent with the conclusion, which is very clear. In particular, the abstract implies that the main aim of the paper is the impact of CO2 concentrations on AMOC stability, instead of, as it is written in the conclusion, performing “a systematic analysis of the AMOC stability in the FWF–CO2 space.”

In the introduction, the impact of CO2 on the AMOC stability is said to “remain largely unexplored”, and the freshwater forcing outside of the 20-50°N band to be a rare occurrence. I believe both of these views are outdated, and the introduction is missing key references and discussion points that provide an accurate and comprehensive picture of the current state of the research. If these comments only concerned modern days, it should clearly be stated, and the results from the palaeo community should be discussed. I recommend splitting the second paragraph of the introduction into three parts. A first one on the CO2 effect, including missing references (e.g. Brown and Galbraith 2016, Zhang 2017, Klockmann 2018, Vettoretti 2022), a second one on the FWF including missing references (e.g. Smith and Gregory 2009, Roche 2010, Kageyama 2013, Ivanovic 2018, Romé 2022) and a third one on the need for combined CO2xFWF analysis and an introduction of your paper, which is currently too short.

I do not believe that the authors can claim to be the first to attempt to draw a landscape of AMOC stability in the CO2 x freshwater forcing space is true, see Brown and Galbraith 2016 for instance. However, I would say that this paper presents the most comprehensive and robust method up to date. If this claim only applies to modern-day studies of the AMOC, it needs to be clearly stated and put in context with palaeo studies.

Significant mode shifts and overshoots on the hysteresis cycles are not discussed in the text. In particular, in Figure 1, the transition in the red solid line around 0.05Sv is remarkable: Is it different from an overshoot? Why is it sustained for about 1000 years? Could this be an occurrence of millennial-scale variability? Could you link this to Willeit 2024?

The definition of the different states comes too late in the paper and lacks precision. How do you define the different modes, using the AMOC index, the mixed layer depth or manually? Additionally, How do you calculate the AMOC index? What references did you use for the modern-day deep water formation sites, and how do they compare to your modern mode? Over what time slices was Figure 3 plotted?

The interpretation of the freshwater flux needs to be clarified in this article, and it becomes a problem when comparing pre-industrial to modern conditions. Would it not be more accurate to account for changes in CO2 and freshwater forcing when comparing the two? Otherwise, what is the point of using a two-dimensional landscape? In addition, the following statement from the conclusion “Our results indicate a generally stronger and deeper AMOC at equilibrium under warmer climate conditions. This is in contrast to the projected AMOC weakening response to anthropogenic global warming […]” is only valid if one considers the sole CO2 effect, but freshwater forcing is expected to increase with Greenland melt, which could take us into a region of the landscape where all four modes exist. I think the comparison between past, present and future states should include a discussion about the role of excess freshwater induced by ice sheet melting.

The details about the construction of the stability landscape is lacking precision and its validity cannot be evaluated. This all the more important as you highlighted the dependence of the direction of variation in Figure 1,2 and B1.

Specific comments

Abstract

L8 - Can you briefly define the OFF and Modern AMOC states?

L11-12 (“In general, the model shows an increase in equilibrium AMOC strength for higher CO2 levels.”) - This does not reflect the actual nature of the work, which goes way beyond this sole aspect. This statement is valid for the standard CO2 experiment in Figure 2, but not consistent with Figure 4 (ex. in Figure 4a, an increase of CO2 can trigger a weak mode). The abstract needs not to focus only on the CO2 experiment but also on the stability landscape.

Introduction

L28 (“There is no consensus as to whether the AMOC is in a monostable or a bistable regime under present climate conditions”) - This needs a reference; I am aware of discussions on the potential weakening of the AMOC, less so about the current state of the AMOC.

Results

L59 (“In particular, there is a range of FWF over which the AMOC has two stable states has two stable states”) – Can you be precise about the range of FWF you are talking about? It also depends on your definition of stability, as I would argue that the dip around 0.05 Sv in Figure 1 is a sign of instability.

L61 - Does “preindustrial conditions” mean 0 Sv in this case? Also, according to the methods, this experiment has pre-industrial CO2 concentrations but a modern-day ice sheet. I would be more careful about using “pre-idustrial” conditions, what about "initial state" instead?

L63 (“suggesting a prominent role of convective instability”) - Could you show that this is a convective instability, showing deep water formation sites activity, for example?

L65 - “This is the result of a collapse of deepwater formation […] of observed past Dansgaard-Oeschger events.” : Here again, the manuscript is missing a plot with the deep water sites dynamics to verify this statement, and a reference about convection in D-O records.

L67 – A definition of what the authors mean by Off, Weak, Modern-day, Strong is needed at this point of the paper.

L82 (“which is possibly more relevant for the ongoing global warming”) - I disagree, both CO2 and meltwater discharge are relevant to future climate changes.

L85-93 - I find the wording of this section confusing. Is the rate of CO2 increase in this paper slower than the “slow” increase in Gérard and Crucifix? How do you explain that you see a strengthening of the AMOC when Gérard and Crucifix 2024 saw a decrease? Could this simply mean that the AMOC response to CO2 is highly uncertain and model dependent?

L105 (“For CO2 above ∼250 ppm, the convection pattern resembles the present-day state with deep water forming in the Labrador Sea and in the Nordic Seas”) - Are you talking about Willeit et al. 2024 or this manuscript?

L156 (“If the climate would be in equilibrium with present-day CO2 concentrations of ∼420 ppm, the model suggests that the Modern AMOC state would not be stable, but that the AMOC would rather be in the Strong state instead”) – Back to the point about the interpretation of future freshwater forcing, would the accelerated melt of Greenland not move the system along the Modern AMOC conditions diagonal instead? Otherwise, this is a major caveat of the analysis that needs to be discussed (although, arguably, the last paragraph of the discussion introduces this idea of model dependency).

Figure 4 - Good Figure but missing information about the criteria used for the clustering of the AMOC states. Is it purely based on the AMOC index, or on the mixed layer depth?

Figure 5 - I think this is the most important Figure of the paper and it deserves to be bigger. Also, why are the Off mode treated differently than the different modes? It makes the Figure difficult to interpret at first reading. Could you use different colours or shading and include all four states in the stability landscape?

L181 - This is a good section and a convincing way to perform this analysis. However, as you showed, the sea ice extent is driving the most significant temperature changes, so I think it would be good to conclude on the impact of less extensive Arctic sea ice in future climate on these results.

Discussion

L186 - Can you say a word about the Modern and Off AMOC modes.

SI

L256 - The paths used in Figure A1 need to be clearly explained in the Figure or the text, otherwise it is impossible to validate the protocol in Figure 4. Do they correspond to the black or the green arrows? Do they include the standard experiments?

Figure B2 does not have a caption.

Technical corrections

Introduction

L17 - Do you have an example of a “societal” change?

L19 - Add Bellomo 2021 for state-of-the-art climate model references.

Results

L51 – “Willeit et al. (2020)” to “(Willeit et al., 2020)

L69 (“most AMOC hysteresis experiments to date have been performed with FWF at lower latitudes (usually between 20N and 50N)”) – Needs references.

L71 (“In our hysteresis experiment”) – Change to “In Figure 1”, there are multiple hysteresis experiments in this paper.

L80 (“give the wrong impression that the AMOC Off state is also stable”) – Is it a wrong impression or relative to the rate of meltwater discharge? Can you give an order of magnitude of the expected meltwater discharge for future melting or during a D-O event/Heinrich event?

L83 – “In an experiment where” to “In Figure 2”?

L94 (“two discrete transition”) - what do you mean by discrete? Abrupt?

Figure 2 - It is difficult to distinguish the solid and dotted lines on this Figure. Maybe two columns?

L110 (“but this has been shown to not be a requirement for the existence of millennial-scale variability”) – Vague, can you say more?

Figure 3 - The Figure needs to be bigger. Please add labels to the colour-bars. Does seasonal mean winter in this case?

L130 - “net freshwater flux” to “net surface freshwater flux”.

L133 (“for CO2 doubling”) – I would say “for double the amount of CO2” because you are not doing a CO2 doubling experiment, which, as you highlighted, could have a different impact on the AMOC.

L139 - “in the model the net surface freshwater flux into the whole Atlantic Ocean shows the opposite trend” to “the net surface freshwater flux into the whole Atlantic Ocean shows the opposite trend in our model”?

L144 - remove “a stabilizing effect”?

L150 - “can be investigating by tracing their stability through” to “can be investigated by tracing the AMOC response in”

L169 - “to explore the pure effect of the different AMOC states on climate” to “to isolate the effect of the changes of AMOC states on the climate”.

Figure 6: Can you use a different colour scheme for the temperature?

Discussion

L193 - “explains” to “demonstrates”?

L202 (“anthropogenic global warming”) - Add Bellomo 2021 to the references for state-of-the-art climate models.

SI

Figure B1: Hard to distinguish the solid and dotted lines on this Figure. Maybe two columns?
Citation: https://doi.org/10.5194/egusphere-2024-1482-RC1
- AC2: 'Reply on RC1', Matteo Willeit, 20 Aug 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1482/egusphere-2024-1482-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2024-1482-AC2
RC2:
'Comment on egusphere-2024-1482', Anonymous Referee #2, 18 Jul 2024
Review comments on egusphere-2024-1482
I thank the author team for their interesting contribution. The study finds an interesting AMOC stability landscape in CLIMBER-X by systematically varying freshwater forcing and atmospheric CO2. I have not seen such a comprehensive study of AMOC stability across varying backgrounds before, and I think it provides an intriguing template to move beyond single-perturbation AMOC studies and broaden our understanding of the processes affecting AMOC stability on multi-millennial time scales. This merits publication in itself, but I think the authors could increase the relevance of their manuscript by expanding their discussion of the implications of their findings for our understanding of past or future AMOC states. I recommend publication with minor corrections.
Main points
I did not understand how many and which simulations were run in total to explore the CO2-FWF space. Was each combination of CO2 and FWF run for initial ‘on’ and ‘off’ states? Were the ‘pathway’ simulations run transiently, or iteratively run into equilibrium?

The existence of a ‘strong’ AMOC mode above 370 ppm is a key result of the study. The authors already compare their results to other model studies and mention that it shows the potential for stronger-than-present AMOC states in the past. Given that this CO2 threshold is close to both current and Pliocene CO2 concentrations, and that the Arctic might have been seasonally ice-free in the Eemian, it would be interesting to discuss this result also in the context of observations and proxy data. Is this solution purely theoretical or is there evidence that such a circulation pattern has existed in the past?

Besides the ‘strong’ AMOC state, the study shows that the ‘off’ state is also stable for high CO2 concentrations. If I understand correctly, this is the conclusion from simulations that were initialised with an ‘off’ state at high CO2. What forcing is required to transition from a ‘strong’ AMOC state into an ‘off’ state under high CO2?

What forcing is required to tip from a ‘modern’ into a ‘strong’ AMOC? What are the climatic impacts of this shift? Do the authors think that tipping into an ‘off’ or ‘strong’ AMOC is possible under persistent anthropogenic climate change?

Minor points
Page 12: Please add an explanation of the grey and black arrows and the black dots to the figure caption or a legend.
Page 14: Please add a caption for Fig B2.
Citation: https://doi.org/10.5194/egusphere-2024-1482-RC2
- AC3: 'Reply on RC2', Matteo Willeit, 20 Aug 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1482/egusphere-2024-1482-AC3-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2024-1482-AC3
RC3:
'Comment on egusphere-2024-1482', Anonymous Referee #3, 22 Jul 2024

Review of the ms. “Generalized stability landscape of the Atlantic Meridional Overturning Circulation” by Matteo Willeit and Andrey Ganopolski”
Using a fast, low-resolution Earth system model, the authors present a systematic and comprehensive exploration of the dynamical stability landscape of the AMOC under quasi-equilibrium forcing conditions. Notably, AMOC stable states are explored in a combined phase space of atmospheric CO2 concentration and North Atlantic freshwater forcing. One of their main results is the possibility for a presence of more than two stable AMOC states if the present-day CO2 concentration and freshwater forcing are run into equilibrium. Among these, there is a stable “off” state, but also a stable “strong” state with deep water formation in the Kara Sea.
This study is a very timely and useful contribution to the discussion about the near-term evolution of the AMOC – a discussion that has very recently gained much traction again. Knowledge of AMOC stability in quasi-equilibrium climates – as opposed to the fast-changing transient that is our reality – is essential context for the scientific understanding of the latter.
The paper is very well written and has a clear layout. I offer a few comments for the authors to consider while recommending the ms. for publication without a further round of reviews.
Comments
The study explores a range of CO₂ concentrations between 180 and 560 ppm. Of course this is a reasonable choice given the pre-industrial concentration. Sadly though, it is not entirely unlikely that the Earth System might have a CO₂ concentration larger that 560 ppm in the foreseeable future. Could the authors comment on why they didn’t explore that part of the phase space?
CLIMBER-X does not have internal interannual climate variability, as stated on l.220. The presence of such variability could lead to some states being indistinguishable (as said on l.222). Another possibility is that a stable state, while technically present, is occupied only with a very small probability (Monahan [2002], JPO 32, 2072-2085).
For the three quantities plotted in Figure 2 we need the spatial context. Where, in e.g. latitude and depth, did you diagnose the AMOC maximum and the AMOC heat transport maximum? Is the delta net FW in panel (c) diagnosed over the entire surface of the Atlantic basin?
I have one concern that I’d like the authors to comment on. Figure 3g tells us that the overturning streamfunction of the “modern” AMOC state shows a complete absence of the abyssal AABW (Antarctic bottom water) cell. We know that in reality the AABW cell is present (Talley et al. [2003], J. Clim. 16, 3213-3226; Johnson [2008], DOI 10.1029/2007JC004477). Does that imply that the CLIMBER-X phase space significantly deviates from present-day conditions? Or should we say that the state labelled “weak” here is the actual modern-day state? Fig. 3f, after all, displays the correct shape of the streamfunction, and Fig. 2a suggests that the amount of overturning is realistic in this “weak” state too. If the model did have variability, perhaps it would convect in the correct regions too?
Could you consider a different type of colour scale for Figure 6? Currently it’s really hard to work out the displayed colling patterns quantitatively using the small hue difference from one colour level to the next.
When discussing Figure 6 (around lines 170 to 180), could it be worthwhile mentioning that an AMOC transition as in Figure 2a, driven by increasing CO2 concentration, would trigger a Modern to Strong transition, and thus a warming actually, with the negative of the patterns in Figure 6 b, f, j?
I think for Figure B2 no legend was provided.

Citation: https://doi.org/10.5194/egusphere-2024-1482-RC3
- AC4: 'Reply on RC3', Matteo Willeit, 20 Aug 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1482/egusphere-2024-1482-AC4-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2024-1482-AC4

Matteo Willeit and Andrey Ganopolski

Viewed

Total article views: 744 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
446	155	143	744	21	17

HTML: 446
PDF: 155
XML: 143
Total: 744
BibTeX: 21
EndNote: 17

Views and downloads (calculated since 23 May 2024)

Month	HTML	PDF	XML	Total
May 2024	90	49	4	143
Jun 2024	105	38	13	156
Jul 2024	106	33	14	153
Aug 2024	58	21	5	84
Sep 2024	39	10	105	154
Oct 2024	48	4	2	54

Cumulative views and downloads (calculated since 23 May 2024)

Month	HTML	PDF	XML	Total
May 2024	90	49	4	143
Jun 2024	105	38	13	156
Jul 2024	106	33	14	153
Aug 2024	58	21	5	84
Sep 2024	39	10	105	154
Oct 2024	48	4	2	54

Viewed (geographical distribution)

Total article views: 746 (including HTML, PDF, and XML) Thereof 746 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 01 Nov 2024

Short summary

Using a fast Earth system model we trace the stability landscape of the Atlantic Meridional Overturning Circulation (AMOC) in the combined freshwater forcing – atmospheric CO₂ space. We find four different AMOC states that are stable under different conditions and a generally increasing equilibrium AMOC strength with increasing CO₂concentrations.

[CO2] (ppm)	Equilibrium (Sv)	Transient (Sv)
280	14.44	14.44
560	12.86	12.38
840	11.93	11.58
1120	11.21	11.05
1400	9.41	9.58


Total:	0
HTML:	0
PDF:	0
XML:	0