the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 30 Jan 2026
Effect of a Bering Strait Closure on AMOC Resilience in a Climate Box Model
Abstract. Recent simulations in an EMIC (CLIMBER-X) have shown that a constructed closure of the Bering Strait can shift the safe carbon budget of the Atlantic Meridional Overturning Circulation (AMOC). In this study we extend a conceptual ocean box model by introducing an atmospheric temperature forcing and a freshwater transport induced by the Bering Strait Throughflow (BST). With this model we can replicate the results produced by CLIMBER-X, and test their sensitivity with respect to forcing and BST parameters. Bifurcation analyses show that a closure of the Bering Strait has a destabilizing effect on an AMOC perturbed by freshwater hosing, but can have a stabilizing effect on an AMOC forced by a polar amplification in atmospheric temperatures – provided the freshwater hosing is limited. A temperature-induced weakening of the AMOC sees a reversal of the BST, which then exports relatively saline waters out of the North Atlantic, and so a closure can have a stabilizing effect. The effectiveness of a closure to prevent a temperature-induced AMOC collapse is sensitively dependent on both the BST parameter values and the rate of the applied forcing. Moreover, the timing of the last preventive closure relies heavily on the forcing rate as well. These conceptual results are important for understanding the feasibility of a Bering Strait closure in order to prevent an AMOC collapse under climate change.
- Preprint
(5427 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 22 Oct 2026)
- RC1: 'Comment on egusphere-2025-6565', Anonymous Referee #1, 09 Apr 2026 reply
-
RC2: 'Comment on egusphere-2025-6565', Anonymous Referee #2, 12 Oct 2026
reply
Soons and co-authors present a conceptual study on the role of the Bering Strait for AMOC stability. They extend the five-box model by Woods2019 with a simple parameterization of the Bering Strait throughflow and explore its impact on AMOC stability on idealized hosing and so-called Arctic amplification experiments. The authors find that the box model is able to reproduce some features also present in the CLIMBER-X intermediate complexity model and may thus serve as a toy model to better understand certain dynamics involving the Bering Strait.
The manuscript is generally well written, though rather technical, which makes it partly hard to follow as it also uses a lot of terminology specific to this box model and abbreviations. While I do see the value of such simple conceptual models for fast hypothesis testing and understanding dynamics, I am not fully convinced by the motivation. The authors make repeated comparisons to results with the CLIMBER-X EMIC (mostly the author’s earlier study on the same topic), and how the box model can reproduce a limited number of findings from the more complex model. However, I wonder what the benefit of this box model truly is. CLIMBER-X is a very fast ESM with simplified but complete physics, that is fast enough to do all the experiments done here with the box model with very moderate computational resources. One could therefore argue to directly use the EMIC and not the box model. I want to reiterate that I’m not saying that such box models are useless, I’m only questioning the framing in this study. Generally, I would advise the authors to drop the comparison to CLIMBER-X, and to treat this really just as a useful but conceptual model. I’m also skeptical of the entire framing of a Bering Strait closure as a geoengineering option to mitigate an AMOC collapse. Instead, I would change the framing to one of an interesting academic problem, but not more. There also seems to be a more fundamental issue with the numerics of the timescale, which needs to be resolved (see my comment below).
Nevertheless, I think this paper may become a valuable contribution to ESD after major revisions.
Major points
1. You only seem to describe simulations from Soons and Dijkstra 2025 in section 2.1. I think a reference or a brief summary would have been enough. Though, generally I think that the extended comparisons to CLIMBER-X go too far here. Tuning the model to the EMIC is fine, but a comparisons to emission driven experiments in CLIMBER-X are not really commensurable and should be removed.
While the box model extension is described in detail, I’m missing a description of the hosing and arctic amplification experiments in the main text. Especially, the latter is somewhat obscure at least for earth system modellers, and needs to be better explained and motivated.
2. I tried to reproduce the results of the extended Wood model myself, and found a mismatch in the timescale of Fig. 8. I cannot reproduce the timescale of that figure by a factor of 100, such that the transient response is wrong. With Table C1 as printed, the response is slow. Box flushing times are V_N/q_n about 60 yr, V_T/q_n about 160 yr and V_B/q_n about 1800 yr. The AMOC in Fig. 8a (F_H = -0.2 Sv, r = 1/yr, ΔT = 8.6 °C) needs more than 100 yr to reach 0 Sv and about 250 yr to approach -7 Sv (Fig. E). The paper shows q_n = 17.5 -> 0 Sv in about 5 yr and the final state of -10 Sv by about 35 yr. Multiplying all time derivatives by 100 (equivalent to dividing all volumes by 100) reproduces the paper’s transient results. It seems t_dim is off by a factor of 100. This would mean that all time axes are 100 times too short and all forcing 100 times too fast. I urge the authors to double check and correct the results, as this may compromise many results of the study especially the comparison to CLIMBER-X, because the forcings rates are 100 times faster in the box model.
Also, there seem to be some inconsistencies between the the equations in the appendix and the Zenodo code, such as terms missing in eqs. A1b and A3b, or a wrong sign in the hosing equation.
Minor/technical points
L1: EMIC has not been defined in the abstract.
L65: What is the Bering Strait Throughflow in CLIMBER-X under PI conditions? Please mention this here and how this compares to the observations.
L219 and following: The parameterization of the Arctic amplification needs to be better explained and why this is equivalent to a global warming scenario.
L 225: Give a brief explanation why arctic amplification on its own causes the AMOC tipping in general (I presume due to its effect on the buoyancy forcing) and how realistic the mechanism really is.
L225-226: This sentence is confusing (at least to me). Do you mean to say that a closed Bering Strait allows for higher Arctic amplification temperatures in the negative freshwater hosing regime? Please clarify.
L226: To me it seems that the effect of a closed versus open Bering Strait is vanishingly small for the high hosing regime (Fig. 5e, 6a) and not the reverse of the negative hosing regime. Am I missing something.
L273: How can you distinguish between hosing induced versus arctic amplification induced when both may contribute to the tipping point for F_H > 0, especially in the regime just below 0.3 Sv, close to the hosing induced tipping point?
L275: I don’t really understand the link to the CLIMBER-X remaining carbon budget. I think the connection to CLIMBER-X has not been established here, at least with regard to this specific box model setup of combined hosing and arctic amplification. Because the latter ha no clear equivalent in the EMIC. I think showing the CLIMBER-X result here is rather suggestive that the simple box modelling can be connected to a remaining carbon budget, which it cannot. I therefore strongly suggest to remove this panel from the figure.
L310: I guess you mean the forcing profile of equation 4? Clarify.
L357: While I see this effect in the salinity, I don’t see the delay in the AMOC panels of Fig, 8a,d.
L361: If you want to mention CLIMBER-X, give some examples of such destabilizing effects in the model (including some supplementary figure).
Figures: Color of the experiments with different hydraulic constants are hard to distinguish in all figures and should be changed.
Figure 6: This is the Hopf point not the saddle-node. The ΔT_N^{a,1} in matches the first loss of stability along the ON branch, which is a Hopf bifurcation. See also L.240-249. Also, Fig. 6c does include ν20 at q_0 = 12 Sv, although the text says this case is omitted.
Figure 7: The y-axis should have units of °C not Sv.
Citation: https://doi.org/10.5194/egusphere-2025-6565-RC2
Model code and software
Extended Wood Model: model formulation & bifurcation analysis Jelle Soons https://doi.org/10.5281/zenodo.17977682
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 1,940 | 873 | 123 | 2,936 | 96 | 152 |
- HTML: 1,940
- PDF: 873
- XML: 123
- Total: 2,936
- BibTeX: 96
- EndNote: 152
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript extends the five-box AMOC model framework of Wood et al. (2019) to study how an open versus closed Bering Strait affects AMOC stability. The authors (i) include an idealised Bering Strait Throughflow parameterisation linking BST transport (q_b) to AMOC strength (q_n) (ii) add a prognostic temperature framework (enabling temperature-forced weakening/tipping in addition to freshwater hosing). The extended box model is calibrated to a stable pre-industrial state of FAMOUS, and results are compared qualitatively with earlier CLIMBER‑X experiments.
I see value in this type of conceptual modelling: while a box model should not be interpreted quantitatively as a predictor of real-world thresholds, it can be useful for mechanism isolation and hypothesis generation, especially when it is tied back to behaviour seen in a more complex model.
My main concern is not the scientific motivation but the presentation: I found the narrative overly complex given the conceptual nature of the study. Several key conclusions are present, but the manuscript would benefit strongly from clearer “take-home messages” per section and from avoiding ambiguous phrasing that mixes open- and closed-strait configurations.
Major comments
1. Abstract clarity and configuration separation (OBS vs CBS): The abstract states that “closure … has a destabilizing effect” and shortly afterwards that a temperature-induced weakening “sees a reversal of the BST”. This can be read as internally inconsistent unless it is explicitly stated that BST reversal occurs only in the open Bering Strait configuration (OBS), whereas in the closed configuration (CBS) there is by definition no throughflow to reverse. I recommend adding one short clause in the abstract to make this explicit (“In the open-strait configuration the throughflow can reverse…”, or similar).
2. Section-level summaries and signposting: From Section 3 onwards, I suggest adding 1–2 sentences at the end (or start) of each subsection summarising: (i) what was tested, (ii) what the result is (direction of change), and (iii) why it happens physically (one mechanism). This will substantially improve readability without changing results.
3. Clarify the statement around (q_0=5.8) Sv and the tipping point (around line ~201): The text says: “This is because at (q_0 = 5.8) Sv the BST switches at the tipping point, since (q_n|{F_H^1}=5.8) Sv then as well.” Please clarify how the value (q_n|{F_H^1}=5.8) Sv is obtained..
4. Clarify the core “message” of the paper as a small set of regimes: The results depend on (a) the forcing type (freshwater hosing vs temperature forcing) and (c) BST parameters ((q_0), (\nu)). Consider adding a short schematic or a concise paragraph in the summary/discussion that states, in plain language, the regimes under which closure stabilises vs destabilises the AMOC in this framework.
I recommend the paper for publication after minor revision.