the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 12 Sep 2025
Non-linear Climatic Response to the Weakening of the Atlantic Meridional Overturning Circulation During Glacial Times
Abstract. The climatic response to the weakening of the Atlantic Meridional Overturning Circulation (AMOC) is investigated under glacial conditions representative of Heinrich Stadial 5 using a fully coupled Earth System Model (ACCESS-ESM1.5), with a focus on Southern Hemisphere and Australian hydroclimate. We find that the climatic response to an AMOC slowdown or shutdown, respectively representing Dansgaard-Oeschger (D-O) and Heinrich stadials, is non-linear. Global mean temperature and precipitation anomalies increase linearly with AMOC weakening; however, crossing the threshold of AMOC shutdown results in non-linear and more complex atmospheric circulation and climate responses. A shutdown of the AMOC in the simulations leads to an enhanced and expanded northern winter Hadley Cell (HC), with a southward shift of its ascending branch. The southern winter HC is weaker but increased in width due to a northward shift of the ascending branch due to AMOC shutdown. This change in the HC drives seasonal variations in the Northern and Southern Hemispheres subtropical high pressure systems and subsequently, changes in the cross-equatorial atmospheric circulation, as well as the Southern Hemisphere mid-latitude westerly winds and other climate features such the monsoon systems. The simulation results are broadly consistent with available proxy records for Heinrich and D-O stadials as well as previous model simulations. The study demonstrates the potential location of a threshold in the climate system between linear weakening and nonlinear shutdown of AMOC with differing climate impacts, further highlighting the importance of not crossing the threshold of AMOC shutdown in the future.
Competing interests: At least one of the (co-)authors is a mem-ber of the editorial board of Climate of the Past. The peer-review process was guided by an independent editor, and the authors also have no other competing interests to declare.
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 paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(10953 KB)
-
Supplement
(3916 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(10953 KB) - Metadata XML
-
Supplement
(3916 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4212', Shih-Yu Lee, 09 Oct 2025
- AC1: 'Reply on RC1', Yanxuan Du, 01 Jan 2026
-
RC2: 'Comment on egusphere-2025-4212', Marlene Klockmann, 20 Oct 2025
Review of Du et al - Non-linear Climatic Response to the weakening of the Atlantic Meridional Overturning Circulation During Glacial Times
Du and colleagues compare the response of the climate system to weakened AMOC versus a collapsed AMOC, representative of Greenland Stadials (in relation to Dansgaard-Oeschger events) and Heinrich Stadials. They find that both temperature and precipitation change approximately linearly, as long as AMOC is decreasing linearly. As soon as the AMOC weakening crosses a threshold such that the AMOC collapses, also the climate system response becomes non-linear. The study initially analyses global fields, with a second focus on the Southern hemisphere/Australasia. In response to AMOC shutdown, both the Northern and Southern hemisphere winter Hadley cells (HC) increase in width, while the northern winter HC strengthens and the southern winter HC weakens. This leads to an expansion of the Indo-Australian summer monsoon and increased precipitation over most of the Australian continent.
The study is mostly well written and the figures well designed. Studies on the impact of AMOC weakening are of great general interest and relevance. I have a few major and a list of minor issues that should be addressed before publication of the study.
Major comments
1. Framing
In the abstract and the conclusions, the study is framed in the context of a potential future AMOC shutdown/threshold. Hoe transferable are the results of this study to future climate, given that the background climate (glacial vs global warming) are fundamentally different? This is briefly mentioned in l.424-429, but deserves a some more discussion.2. Mechanistic link between AMOC weakening and Hadley cell changes
In parts, the manuscript remains very descriptive. I am missing some discussion of the mechanism how the AMOC changes lead to the described ITCZ and HC changes. Is it only the change in heat transport and the resulting change in temperature gradients?3. The results of this study in the broader context of previous literature
In the introduction (l.68-73), the authors summarise the findings of previous studies on the atmospheric response to glacial AMOC weakening/shutdown. It would help in highlighting the relevance of this study, to emphasise more strongly the open questions that this study addresses and in the conclusions also to emphasise the new knowledge gained by this study. I understood that the results or this study are broadly consistent with previous literature based on both climate models and proxies. But what is the key new insight from this study?4. Setup of the hosing simulations
The setup seems not 100% consistent. Does it affect the results that the hosing is increased successively in the weaker hosing experiments? Would the 49ka_0.3Sv have the same state if it was started from 49ka_control? Or vice versa, would the response to 0.4Sv hosing be the same if you had continued to successively increase the hosing?
And yes, 49ka_0.3Sv and 49ka_shutdown have the same total integration time, but 49ka_0.2Sv does not. And also the integration time under the same hosing strength is different for the two DO-analogues vs the shutdown experiment. It probably does not have a big effect on the results, but it would be good to have this at least discussed.5. Significance/Robustness of very small changes in latitude
In many occasions, very small changes in latitude are reported (0.1° or even smaller). Given that the latitudes are interpolated to 0.5°, are changes smaller than 0.5° even significant/robust/detectable? Is not any change <0.5° below the accuracy of the spatial resolution?6. Consistent way of referring to the simulations
Sometimes "AMOC weakening" refers to only the 0.2SV and 0.3Sv simulation, sometimes it seems to refer to all three simulations, including the shutdown. This makes it sometimes hard to clearly follow the argument. You could e.g. consistently refer to the 0.2SV and 0.3Sv simulations as "the DO simulations" and to the shutdown simulation as "the HS simulation" and then be very clear, whether your current result applies only to the DO or HS simulations or to both. Other clear nomenclatures are of course possible as well.7. Absolute changes vs changes normalised by AMOC change
This is more a question that I had while reading: would it make sense to express changes in other variables (HC strength, width, etc) also as a function of AMOC change?Minor
l.17: "global mean temperature and precipitation anomalies increase linearly" This is not reflected in the main text, where you emphasise non-linearity also for the global mean fields (factor of 1.3 for both variables).
l.55: HS5 is introduced quite abruptly. Any particular reason why you focus on HS5?
l.84-85: what do you mean by "consistent response here" what are the open questions?
Tab1: how are albedo, vegetation, topography and runoff obtained for 49ka? what are they based on?
l.119-120: this sentence is not fully clear. The 1555 years are the spinup of the 49ka_control simulation? Or are the 760 years of 49ka_control part of the 1555 years? why were boundary conditions changed stepwise and how?
l.113-122: This section relies heavily on information from Saini et al 2025b. While you of course don't need to go as much into the details as Saini et al, it shoud still be possible to understand the relevant aspects of the setup without having to read Saini et al in addition.
l.140/Fig.S1: include 49ka_control in Figure S1 for comparison
l.175: what is the additional benefit of the cubic spline? are the results very sensitive to the fitting parameters?
l.188-189: was the AMOC reduction in the simulations with the 40% heat transport weakening of a similar magnitude as in 49ka_shutdown? If the residual AMOC strength after the "shutdown" was stronger than in 49ka_shutdown, this could be an explanation for the difference in heat transport reduction. Also, these numbers are probably very model dependent to begin with.
l.191: How is this huge change in the subpolar North Atlantic possible? is that sea-ice related? I can see that air temperatures over sea ice may change dramatically, but a similar cooling seems to be happening in the SST as well. This implies that the subpolar North Atlantic must have temperatures of around 25° or higher, is it really that warm? But perhaps the SST changes are not as large and only the non-linear spacing of the colourbar in Fig. S3 makes it hard to read the actual magnitude of the SST change.
l.198: warming over Antarctica is also non-significant in 49ka_shutdown (no stippling over Antarctica).
l.209: To me, the precipitation patterns do not seem more diverse than the temperature patterns. At least not on on the global scale. Which is probably also reflected by the fact that the global mean temperature and precipitation both change by a factor of 1.3 between DO and HS simulations.
l.257-259: this sentence should refer to the change in HC width, not the absolute HS width
l.300: this seems to be true mostly over Africa/the subtropical Atlantic/eastern pacific.
l.380-382: These two sentences seem to contradict each other "[...] Australia and New Guinea receive increased monsoon precipitation [...]. [...] thus no significant changes in precipitation are simulated over New Guinea."
l.474-475: What were the previous studies based on?
Editorial
l.31-32 nomenclature: the sentence in its current form seems to imply that the warming/cooling transitions are the Greenland Interstadials/Stadials. But Interstadial/Stadial refers to the more or less stable periods between the transitions. Please reformulate
l.34 same as above. Also, stadial was already introduced in the sentence before.
l.40 "contain" instead of "contains"
l.40-43: please shortly explain what is debated and which usage you follow.
l.48: Add "The" before "climatic response" at the beginning of the sentence.
l.59: "global temperature was" instead of "is"
l.121-122: should this not be "relative stable surface air and sea surface temperatures" rather than "stable changes in SAT and SST changes"? l.134: "more likely to close to a complete shutdown", something is missing in the sentence, please reformulate
l.197-199: sentence does not work. please reformulate
l.305-309: super long sentence. please break it down to smaller, easier sentences.
l.356-357: sentence does not work. please reformulate
l.383-391: difficult to follow which part applies to all or only the DO or HS simulations. Please reformulate for clarity
Fig.9 (caption): the identifiers "left column", "middle" and "right column" do not fit the structure of the figure. It should probably be only "top" and "bottom" behind DJF and JJA, respectively.
l.415: "strengthen and migrate" instead of "strengthens and migrating"Citation: https://doi.org/10.5194/egusphere-2025-4212-RC2 - AC2: 'Reply on RC2', Yanxuan Du, 01 Jan 2026
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4212', Shih-Yu Lee, 09 Oct 2025
A) Experimental realism & robustness
1. Freshwater forcing magnitude, duration, and geometry: The shutdown uses 0.4 Sv for 500 yr over 50–70° N, 70–0° W (Table 2)—an idealized choice that likely exceeds plausible HE freshwater flux histories. Please (i) discuss physical plausibility versus “shock” idealization; (ii) provide a brief sensitivity or cite prior ACCESS ESM1.5 tests to hosing shape (pulsed vs ramped), duration, and release region (e.g., including/substituting the Nordic Seas or Labrador shelf). Even a short 100 yr/0.4 Sv pulse test or a reduced area hosing would help demonstrate that the non linear atmospheric reorganization at shutdown is not a by product of sustained extreme hosing.2. Internal variability and sampling: Results rely on 50 yr windows. Please consider a simple signal to noise check by resampling 50 yr blocks from the control and from each experiment (or show running 30 yr means across the last 150 yr) to demonstrate that key patterns (DJF ITCZ latitude, HC strengths/widths, STR latitude, SH westerly latitude) exceed internal variability.
B) Mechanistic clarity/suggestion
5. Energetics of the ITCZ shift: Since the story hinges on interhemispheric energy transport compensation, would it be possible to exam TOA energy transport decomposition (DJF/JJA) that connects the ∼1 PW reduction in NH ocean heat transport at 30° N to the southward DJF ITCZ jump in shutdown. Even a zonal mean cross equatorial energy flux figure would make the mechanism crisper.6. Southern Ocean/sea ice feedbacks: You note notable SAT cooling near the Ross/Weddell sectors and stronger SH westerlies in shutdown (Fig. 2a; Fig. 7). The argument will be more comprehensive to show how wind/ice/ocean coupling amplifies the SH response.
7. Basin contrasts in SH westerlies: The Atlantic sector behaves differently from the Pacific/Indian in slowdowns; shutdown realigns them (Fig. 7). Would be good to clarify why the Atlantic deviates.
8. Carbon cycle configuration and outputs: ACCESS ESM1.5 includes an interactive carbon cycle, and the author discuss potential CO₂ links via SH westerlies (p. 22). Please state whether carbon was prognostic or prescribed here and, if active, showing simulated air–sea CO₂ flux (Southern Ocean sectors), DIC/alkalinity, and atmospheric CO₂ response will be of great interest.
Citation: https://doi.org/10.5194/egusphere-2025-4212-RC1 - AC1: 'Reply on RC1', Yanxuan Du, 01 Jan 2026
-
RC2: 'Comment on egusphere-2025-4212', Marlene Klockmann, 20 Oct 2025
Review of Du et al - Non-linear Climatic Response to the weakening of the Atlantic Meridional Overturning Circulation During Glacial Times
Du and colleagues compare the response of the climate system to weakened AMOC versus a collapsed AMOC, representative of Greenland Stadials (in relation to Dansgaard-Oeschger events) and Heinrich Stadials. They find that both temperature and precipitation change approximately linearly, as long as AMOC is decreasing linearly. As soon as the AMOC weakening crosses a threshold such that the AMOC collapses, also the climate system response becomes non-linear. The study initially analyses global fields, with a second focus on the Southern hemisphere/Australasia. In response to AMOC shutdown, both the Northern and Southern hemisphere winter Hadley cells (HC) increase in width, while the northern winter HC strengthens and the southern winter HC weakens. This leads to an expansion of the Indo-Australian summer monsoon and increased precipitation over most of the Australian continent.
The study is mostly well written and the figures well designed. Studies on the impact of AMOC weakening are of great general interest and relevance. I have a few major and a list of minor issues that should be addressed before publication of the study.
Major comments
1. Framing
In the abstract and the conclusions, the study is framed in the context of a potential future AMOC shutdown/threshold. Hoe transferable are the results of this study to future climate, given that the background climate (glacial vs global warming) are fundamentally different? This is briefly mentioned in l.424-429, but deserves a some more discussion.2. Mechanistic link between AMOC weakening and Hadley cell changes
In parts, the manuscript remains very descriptive. I am missing some discussion of the mechanism how the AMOC changes lead to the described ITCZ and HC changes. Is it only the change in heat transport and the resulting change in temperature gradients?3. The results of this study in the broader context of previous literature
In the introduction (l.68-73), the authors summarise the findings of previous studies on the atmospheric response to glacial AMOC weakening/shutdown. It would help in highlighting the relevance of this study, to emphasise more strongly the open questions that this study addresses and in the conclusions also to emphasise the new knowledge gained by this study. I understood that the results or this study are broadly consistent with previous literature based on both climate models and proxies. But what is the key new insight from this study?4. Setup of the hosing simulations
The setup seems not 100% consistent. Does it affect the results that the hosing is increased successively in the weaker hosing experiments? Would the 49ka_0.3Sv have the same state if it was started from 49ka_control? Or vice versa, would the response to 0.4Sv hosing be the same if you had continued to successively increase the hosing?
And yes, 49ka_0.3Sv and 49ka_shutdown have the same total integration time, but 49ka_0.2Sv does not. And also the integration time under the same hosing strength is different for the two DO-analogues vs the shutdown experiment. It probably does not have a big effect on the results, but it would be good to have this at least discussed.5. Significance/Robustness of very small changes in latitude
In many occasions, very small changes in latitude are reported (0.1° or even smaller). Given that the latitudes are interpolated to 0.5°, are changes smaller than 0.5° even significant/robust/detectable? Is not any change <0.5° below the accuracy of the spatial resolution?6. Consistent way of referring to the simulations
Sometimes "AMOC weakening" refers to only the 0.2SV and 0.3Sv simulation, sometimes it seems to refer to all three simulations, including the shutdown. This makes it sometimes hard to clearly follow the argument. You could e.g. consistently refer to the 0.2SV and 0.3Sv simulations as "the DO simulations" and to the shutdown simulation as "the HS simulation" and then be very clear, whether your current result applies only to the DO or HS simulations or to both. Other clear nomenclatures are of course possible as well.7. Absolute changes vs changes normalised by AMOC change
This is more a question that I had while reading: would it make sense to express changes in other variables (HC strength, width, etc) also as a function of AMOC change?Minor
l.17: "global mean temperature and precipitation anomalies increase linearly" This is not reflected in the main text, where you emphasise non-linearity also for the global mean fields (factor of 1.3 for both variables).
l.55: HS5 is introduced quite abruptly. Any particular reason why you focus on HS5?
l.84-85: what do you mean by "consistent response here" what are the open questions?
Tab1: how are albedo, vegetation, topography and runoff obtained for 49ka? what are they based on?
l.119-120: this sentence is not fully clear. The 1555 years are the spinup of the 49ka_control simulation? Or are the 760 years of 49ka_control part of the 1555 years? why were boundary conditions changed stepwise and how?
l.113-122: This section relies heavily on information from Saini et al 2025b. While you of course don't need to go as much into the details as Saini et al, it shoud still be possible to understand the relevant aspects of the setup without having to read Saini et al in addition.
l.140/Fig.S1: include 49ka_control in Figure S1 for comparison
l.175: what is the additional benefit of the cubic spline? are the results very sensitive to the fitting parameters?
l.188-189: was the AMOC reduction in the simulations with the 40% heat transport weakening of a similar magnitude as in 49ka_shutdown? If the residual AMOC strength after the "shutdown" was stronger than in 49ka_shutdown, this could be an explanation for the difference in heat transport reduction. Also, these numbers are probably very model dependent to begin with.
l.191: How is this huge change in the subpolar North Atlantic possible? is that sea-ice related? I can see that air temperatures over sea ice may change dramatically, but a similar cooling seems to be happening in the SST as well. This implies that the subpolar North Atlantic must have temperatures of around 25° or higher, is it really that warm? But perhaps the SST changes are not as large and only the non-linear spacing of the colourbar in Fig. S3 makes it hard to read the actual magnitude of the SST change.
l.198: warming over Antarctica is also non-significant in 49ka_shutdown (no stippling over Antarctica).
l.209: To me, the precipitation patterns do not seem more diverse than the temperature patterns. At least not on on the global scale. Which is probably also reflected by the fact that the global mean temperature and precipitation both change by a factor of 1.3 between DO and HS simulations.
l.257-259: this sentence should refer to the change in HC width, not the absolute HS width
l.300: this seems to be true mostly over Africa/the subtropical Atlantic/eastern pacific.
l.380-382: These two sentences seem to contradict each other "[...] Australia and New Guinea receive increased monsoon precipitation [...]. [...] thus no significant changes in precipitation are simulated over New Guinea."
l.474-475: What were the previous studies based on?
Editorial
l.31-32 nomenclature: the sentence in its current form seems to imply that the warming/cooling transitions are the Greenland Interstadials/Stadials. But Interstadial/Stadial refers to the more or less stable periods between the transitions. Please reformulate
l.34 same as above. Also, stadial was already introduced in the sentence before.
l.40 "contain" instead of "contains"
l.40-43: please shortly explain what is debated and which usage you follow.
l.48: Add "The" before "climatic response" at the beginning of the sentence.
l.59: "global temperature was" instead of "is"
l.121-122: should this not be "relative stable surface air and sea surface temperatures" rather than "stable changes in SAT and SST changes"? l.134: "more likely to close to a complete shutdown", something is missing in the sentence, please reformulate
l.197-199: sentence does not work. please reformulate
l.305-309: super long sentence. please break it down to smaller, easier sentences.
l.356-357: sentence does not work. please reformulate
l.383-391: difficult to follow which part applies to all or only the DO or HS simulations. Please reformulate for clarity
Fig.9 (caption): the identifiers "left column", "middle" and "right column" do not fit the structure of the figure. It should probably be only "top" and "bottom" behind DJF and JJA, respectively.
l.415: "strengthen and migrate" instead of "strengthens and migrating"Citation: https://doi.org/10.5194/egusphere-2025-4212-RC2 - AC2: 'Reply on RC2', Yanxuan Du, 01 Jan 2026
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 8,947 | 2,313 | 192 | 11,452 | 380 | 176 | 225 |
- HTML: 8,947
- PDF: 2,313
- XML: 192
- Total: 11,452
- Supplement: 380
- BibTeX: 176
- EndNote: 225
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Yanxuan Du
Josephine R. Brown
Laurie Menviel
Himadri Saini
Russell N. Drysdale
David K. Hutchinson
Calla N. Gould-Whaley
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(10953 KB) - Metadata XML
-
Supplement
(3916 KB) - BibTeX
- EndNote
- Final revised paper
A) Experimental realism & robustness
1. Freshwater forcing magnitude, duration, and geometry: The shutdown uses 0.4 Sv for 500 yr over 50–70° N, 70–0° W (Table 2)—an idealized choice that likely exceeds plausible HE freshwater flux histories. Please (i) discuss physical plausibility versus “shock” idealization; (ii) provide a brief sensitivity or cite prior ACCESS ESM1.5 tests to hosing shape (pulsed vs ramped), duration, and release region (e.g., including/substituting the Nordic Seas or Labrador shelf). Even a short 100 yr/0.4 Sv pulse test or a reduced area hosing would help demonstrate that the non linear atmospheric reorganization at shutdown is not a by product of sustained extreme hosing.
2. Internal variability and sampling: Results rely on 50 yr windows. Please consider a simple signal to noise check by resampling 50 yr blocks from the control and from each experiment (or show running 30 yr means across the last 150 yr) to demonstrate that key patterns (DJF ITCZ latitude, HC strengths/widths, STR latitude, SH westerly latitude) exceed internal variability.
B) Mechanistic clarity/suggestion
5. Energetics of the ITCZ shift: Since the story hinges on interhemispheric energy transport compensation, would it be possible to exam TOA energy transport decomposition (DJF/JJA) that connects the ∼1 PW reduction in NH ocean heat transport at 30° N to the southward DJF ITCZ jump in shutdown. Even a zonal mean cross equatorial energy flux figure would make the mechanism crisper.
6. Southern Ocean/sea ice feedbacks: You note notable SAT cooling near the Ross/Weddell sectors and stronger SH westerlies in shutdown (Fig. 2a; Fig. 7). The argument will be more comprehensive to show how wind/ice/ocean coupling amplifies the SH response.
7. Basin contrasts in SH westerlies: The Atlantic sector behaves differently from the Pacific/Indian in slowdowns; shutdown realigns them (Fig. 7). Would be good to clarify why the Atlantic deviates.
8. Carbon cycle configuration and outputs: ACCESS ESM1.5 includes an interactive carbon cycle, and the author discuss potential CO₂ links via SH westerlies (p. 22). Please state whether carbon was prognostic or prescribed here and, if active, showing simulated air–sea CO₂ flux (Southern Ocean sectors), DIC/alkalinity, and atmospheric CO₂ response will be of great interest.