Review of « Seasonal overturning variability in the eastern North Atlantic subpolar gyre: A Lagrangian perspective” by Oliver J. Tooth et al. submitted to Ocean Science. 

Using a NEMO-based forced global numerical hindcast simulation, the authors characterize the seasonal variability of the overturning circulation across the OSNAP-east section in the Subpolar North Atlantic. Combining Euleriand and Lagrangian approaches, they found that such seasonality is critically controlled by the transit time of water parcels north of the section, with a 8.5 month threshold beyond which diapycnal transformation is irreversible (i.e. contribution to the mean overturning, not the seasonality). They further describe the pathways and mechanisms underlying the seasonal signal, and show the key role for wind-driven changes in recirculation time of upper water masses in Irminger Sea. Distinguishing the respective pathways and timescales that characterize the seasonal and mean overturning is a key asset of this work.

It was overall a pleasure to read this manuscript. It is well and precisely written, with a rigorous and very comprehensive analysis of well-posed scientific questions. I think it comprises significant findings and will represent, alongside its already-published companion paper on the mean overturning state, a timely contribution to the field. Therefore, I have only a few general and minor comments, which Iist below.

General comments.

• The paper is long, quite dense and detailed, which could give some readers a hard time to eventually extract the key take-home messages. I would suggest shortening the text where possible and only keep the key findings in the main text, and maybe put the complementary diagnostics in supplementary materials. This is more of an advice than a request.
• It should be made clearer (in the abstract notably) that the results apply to the OSNAP-east section only. In some instance, the findings are presented as relevant for the entire eastern SPNA (e.g. line 13-16). Although the sensitivity of the results to the specific section location is mentioned at line 579-584, I think this should be better emphasized throughout the paper. In fact, I wonder if replacing “in the eastern North Atlantic subpolar Gyre” by “across the OSNAP-east section” in the title would be indeed more correct.
• Although the authors start to elaborate on the possible impact of using higher-resolution runs at the very end of the manuscript, I believe more could be said on their use of a single numerical model to infer general conclusions about the overturning. I think the authors should better acknowledge in their manuscript that their results might be model-dependent, and specifically point out the most sensitive diagnostic/results accordingly. On the same topic, references to previous validation of that particular simulation in the SPNA should be added in Section 2.1 (how well does ORCA025-GJM189 represents the basic features of the subpolar North Atlantic?)
• I was left wondering whether the results could be sensitive to the chosen parametrization of turbulent convective mixing within the mixed layer (random perturbation of vertical velocities)? Could the authors comment on this in their Methods section?

Minor comments.

1. 65-69: I am not sure to follow the point made here. Comparing the seasonal AMOC amplitude (4 Sv) with that of its potential surface forcing (20 Sv) makes sense (although the volume term hampers this comparison, as stated), but comparing it to the mean strength (16 Sv) is less clear to me.
2. 152: Integrating from the sea-surface (instead of from the bottom) implies that the MOC strength includes the net transport through the section, because one can assume that the northward transport into the Arctic takes place in the shallowest layers. Some studies indeed use bottom-up integration to strictly capture the overturning (in fact the authors here remove this net throughflow to provide the mean at line 190). Therefore, I wonder whether the Eulerian seasonal signal does include a contribution from the seasonal variability of the net throughflow? Was it also removed from the total signal?
3. 249-250: What explains the 1 Sv difference between the peak-to-peak amplitude of the seasonal MOC (4.1 Sv) and LMOC (5.1 Sv)? This should be explained.
4. 348: Vage et al (2011) is an observational analysis, so it is not obvious whether their definition of boundary-interior limit (500 km) applies in the model too. Are the simulated IG and IC characteristic in line with observed ones ?

Review of “Seasonal overturning variability in the eastern North Atlantic subpolar gyre: A Lagrangian perspective” by Tooth et al.

The manuscript describes an in-depth analysis of Lagrangian-derived overturning in the eastern subpolar gyre, with an emphasis on the seasonal cycle. The analysis is conducted wholly in the ORCA025 model run that extends from 1958-2015. The authors compare the Eulerian AMOC in the model to a Lagrangian-derived AMOC by seeding Lagrangian particles in the northward currents across OSNAP East and calculating their net water mass transformation by the time the water particles recirculate southwards across OSNAP East. The authors find that the seasonal cycle across OSNAP East is primarily driven by fast-moving particles that recirculate in the region within 8.5 months.

I found the manuscript fascinating and the figures beautiful. The text is quite long, but each of the sections provided interesting information. I commend the authors for the amount of work and diligence it must have taken to prepare a manuscript with this much material. For this reason, I was torn about how to review this paper: on one hand, the paper is extremely polished, while on the other hand, I found a few major concerns about the manuscript (described below). I have decided to recommend the paper be reconsidered pending major revisions, mostly because my concerns underlie the basis of the paper, and they give me serious pause when trying to learn what to take away from the paper. Without sufficiently addressing these concerns, the manuscript lacks a central message despite the fascinating results along its circuitous journey (which is analogous to the journey these Lagragian particles take around the Iceland Basin…).

Major comments:

1. It is unclear to me what the goal of the paper is and/or what signal the authors are trying to explain. Is it the observed seasonality at OSNAP East? Or possibly the model’s version of seasonality at OSNAP East? If the authors are trying to explain the observed seasonality at OSNAP East, it has not yet been identified in the published literature, so it seems strange to try to explain it without the signal being identified. Can the authors diagnose the OSNAP East seasonal cycle from the publicly-available data and use that as a motivation for the current study? And if it is the model’s seasonality, the authors should explain the importance of a single model’s representation of the seasonal cycle, especially whether it resembles the observations.

2. Part of my concern in #1 arises from ambiguity in the introduction – individually the sentences are factually correct and well-written, but I often didn’t understand how one sentence led to the next. This is true throughout the section, though I will highlight the sentence starting “It therefore remains an open question…” (line 80) because it is critical to motivating this paper. The previous sentences were discussing seasonal versus interannual variability, then in this sentence the authors shift to comparing the mean AMOC to its seasonal cycle. I generally agree with the sentences individually up to this point, I just don’t know how it leads to the authors’ question that they seek to address in the paper. I also didn’t understand the importance of this question: why does it matter whether the particles that determine the mean state are the same as the ones that determine the seasonality? To address this issue throughout the introduction (not just the example I provided), I recommend the authors highlight a single (or set) of questions that they aim to address in the manuscript, and provide motivation for why those questions are important in the introduction. Otherwise, the text seems to ramble through a lot interesting topics, but lacks clear, identifiable results. I also believe that much of the text could be condensed and made more readable if the goals of the study were outlined early.

3. The authors use ORCA025 exclusively and do not motivate why this would be a good, or even sufficient model to use for this analysis. There are certainly higher resolution models run for similar time periods readily available, so I would hope that there is a reason to use this model over the others (e.g. HYCOM, VIKING, other NEMO-based model runs, ECCO, etc.). Though resolution is not the only component of a model that determines its quality, I am concerned that a 1/4° resolution model cannot resolve some of the important processes in the eastern subpolar North Atlantic shown in the literature (e.g. Gary et al., 2018; Houpert et al., 2018; Zhao et al., 2018; Devana et al., 2021), specifically the transformation of water, and any vertical velocities, both of which are highly resolution-dependent, yet important to the AMOC. I am also concerned that a ¼° cannot properly resolve the three currents that enter the Iceland Basin (Holliday et al., 2020), which are critical to understanding the circulation in the region. My final concern about the resolution of the model concerns running Lagrangian trajectories through a coarse-resolution velocity field involves a lot of interpolation between points, with the resultant figures (which are beautifully presented) at much higher resolution than the underlying data, and potentially evoking higher confidence in the results than one might otherwise given 1/4° data. The authors acknowledge this issue in the last paragraph of the paper, but it needs to be addressed in the data and methods section (if not earlier).

Again, I want to underscore how impressed I was with the quality of the paper, which reflects quite strongly on the authors.

The manuscript by Tooth et al. presents a detailed and thorough model analysis of the seasonal overturning variability happening between the OSNAP array and the Greenland-Scotland Ridge. Combining insights from a Eulerian analysis at the OSNAP array and a Lagrangian framework where the sensitivity of the transformation to inflow characteristics is tested, the authors show that the majority of the seasonality in the overturning is due to water parcels that exhibit a relatively short (< 8.5 months) recirculation time within the eastern North Atlantic Subpolar Gyre.

The analysis is scientifically sound, very well embedded in the existing literature and is valuable to e.g. better interpret OSNAP measurements and improve our understanding of water mass transformation processes. However, I do agree with most of the comments of the other reviewers. The paper is very long, and, due to the smart but tricky to understand Lagrangian method, it can be challenging for the readers to fully grasp the content. Also, the motivation of the study can be more clearly defined in the abstract and introduction. Therefore, I hope to provide some recommendations and suggestions to improve the readability of the manuscript and would advise minor revisions before publication, but with sufficient revision time to restructure the paper.

General Comments

1. Abstract

• More clearly state the motivation / current lack of knowledge in your abstract and how your approach provides new insight. E.g. One of your main findings is that you are only able to explain the minimum MOC in autumn seen in the OSNAP measurements if you use a Lagrangian approach (“This convergence of southward… wind forcing”). This is at the moment not clear in your abstract.
• The statement “recirculation race against time” is nice to mention in the paper, but might confuse readers in the abstract. And I don’t think you need it in the abstract, as it is already clear from the last sentence what your main finding is (“The seasonality of Lagrangian overturning… in the eastern SPG”).

2. Introduction

• Try to get to the main research question / motivation for this study within the first two paragraphs. I find some hint for motivation in Ln.78, but I would suggest to get to this much quicker, and state clearly how this is related to the research question and approach of your study.
• In general, this introduction can be shortened quite a bit, there are many details that are not needed to understand the motivation of the study, and some can be moved to relevant parts in the manuscript.
  • g. Ln. 35-39 not necessary in the introduction (can move that to the method section where you explain how you define overturning)

3. General structure of the paper

• The structure in the abstract differs from the general structure of the paper, and I think it makes a bit more sense to indeed first fully discuss the insights from the Eulerian analysis, before moving on to the Lagrangian one. That would also help to more clearly state what the added benefit is for looking at the relevant mechanisms from a Lagrangian perspective. So e.g. move largest part of section 6 to follow 3.1. Or, have a full section 3 focused on the Eulerian perspective where you have a dedicated section for validation (what now is mainly section 3.1) to also argue why the model you’re using is the right choice to address the seasonal variability and related mechanisms. Furthermore, it would make the interpretation of the Lagrangian results a lot easier when readers have seen the general Eulerian flow structure in this region and the full overturning characteristics from a Eulerian perspective.
• The different seasonal cycle seen when comparing the Eulerian to the Lagrangian framework can be explained a little better. I’m not sure whether readers fully understand this. Maybe as a thought experiment, think what would happen when you would define the Lagrangian overturning “backwards”. So, tracing the Southward flow backwards, and define the LMOC overturning in that way. This would again change the seasonal variability observed as you would focus on the seasonality of the outflow, instead of the inflow.
• Check the length of your paragraphs, some of them are extremely long. Try to keep it to one or two main take-aways per paragraph, and keep them in general short (e.g. max length ~12 lines in the current template format).

4. Discussion and conclusions

• Currently the focus is too much on conclusions, and the relevance and importance of the results can be more strongly communicated.
• Maybe also put your findings more in the context of the OSNAP observations.

Specific Comments

1. 1 - Why MOC and not AMOC?
2. 2 – I find the SPG abbreviation confusing (maybe change to ESPG?), in particular for people that only read the abstract. Even when mentioning the eastern part of the Subpolar Gyre I have a bigger area in mind than the one North of OSNAP and south of GSR. I think the region of interest should be more clearly defined already in the abstract.
3. 7 – Also here, it is not clear for the reader where exactly you are defining this seasonal cycle (minimum AMOC), I do think you should mention the OSNAP array in the abstract.
4. 142 – How did you define the Greenland-Scotland Ridge in your model?
5. 2 – If I understand the calculation correctly, it should be V_south(sigma, t<tau<tau_max) , to make clear that any parcel that returns within this period of 7 years is added to the LMOC?
6. 220-225 – You could already make a link here why you need a Lagrangian framework to explain why this is the case (now this text might insinuate that already cold and dense waters transported Northward somehow lead to maximum overturning strength).
7. 243 – Confusing what the transports mentioned in the brackets are
8. 244 – 243 – I don’t understand what is meant here with ‘close correspondence’, is that somehow visible in one of the figures? Did you calculate a correlation?
9. 267 and elsewhere. In general I think care should be taken when talking about seasons in relation to the LMOC definition as there is a time lag involved in the actual calculation (e.g. when the transformation of the water masses occurs), so it is very difficult to interpret what a minimum in May actually means.
10. 277 – “in contrast”, how does the context of this sentence is in contrast with the previous one?
11. 302 – The recirculation time itself is probably also seasonally variable? Maybe already address that here?
12. 316 – Unclear sentence, which is better explained in the following sentences. Maybe just say “We have identified a threshold recirculation time of 8.5 months”. And then continue with explaining what happens to particles < recirculation time, and then > recirculation time.
13. 330 and elsewhere – I would change the abbreviation of this pathway to Ic-Irm and Ro-Irm, as all pathways are defined by their entry and exit locations and not by crossing RR. Makes it easier to remember for the reader.
14. 335 – “70%” make clear where the reader can find this result and in which figure.
15. Section 5.2 – This section is extremely long, and due to all the different decompositions very difficult to keep track of what is happening. I would suggest to restructure this, shorten, and maybe split in different sub-sections if needed.
16. 515-519 – This is one of your key findings, the built-up to this result can be made clearer, by already talking about Wang’s conclusions in the introduction and stating why these might be insufficient arguments?

Figures

Figure 1

• I could not find in the text where you reference panel 1b. Also, refer to 1a relatively early in your introduction, to make clear how you define the Eastern SPG (and maybe then use as abbreviation “ESPG” instead of “SPG”).
• Caption: “ volume transports across the model-defined OSNAP East array.” (also in caption Fig. 2 and Fig. 3)
• Dotted line panel b at 1990 does not seem necessary for the storyline.

Figure 4

• Panel b, it would be good to add the main flow features already here (not wait until Fig 8d)
• Panel b, unclear what the thick black line represents
• Use of white region vs. white line is confusing, also because the white line is missing in your colorbar. I do like the use of the white line to indicate the threshold-time, but then maybe use gray for the masked areas?

Figure 5

• You start introducing different decomposites based on pathway at the beginning of section 5.1, but then continue to further decompose them (e.g. IC and IG, and the eastern and western part of the IC-Irm pathway). Maybe it would be good to directly make this clear in one figure, so readers can better follow the story and understand the choices made?
• Specify Ic and Ro separately in the figure, now it is unclear how you kept these two pathways separate in your analysis. Maybe, as the decomposition is based on in- and outflow location, add those separation lines in panel b instead with all relevant abbreviations and switch the two panels (b = a). In panel (b) also add ‘Reykjanes Ridge’.
• Use of color, check the ‘colorblind’ rules, I think the difference between red and orange is very difficult to distinguish (especially in panel b).

Figure 6 & 7

• What do the colored boxes represent? Again, difference between the purples and blue not well visible.
• Panel c and d can be left out, as this is also seen in e and f.
• I would merge this figure with Fig. 7, and panel 7b is not needed. Also make clear why you only look at Ic – Irm, and not Ro – Irm.

Figure 8

• As mentioned earlier, move this figure to section 3a.