Review Goris et al., Gulf Stream and interior western boundary volume transport as key regions to constrain the future North Atlantic Carbon Uptake

This study aimed for regional optimization of the emergent constraints for projecting future North Atlantic carbon uptake. A previous study (Goris et al., 2018) identified two indicators, i.e., seasonal pCO₂^sea anomaly in middle-to-high latitude and fraction of anthropogenic carbon inventory below 1000m, for future carbon uptake projection in the North Atlantic. The authors apply a genetic algorithm to further find out which spatial area and depth ranges are crucial for the emergent relationships. This study is scientifically interesting to constrain the projections of the North Atlantic ocean carbon uptake, and also practically provide guidance for monitoring and observational strategies. However, this manuscript needs some further work and clarification to be published.  

Major comments:

-Inconsistency of the season for pCO₂^sea anomaly: it is winter time pCO₂^sea in this study, but the cited paper (Goris et al., 2018) used summer time pCO₂^sea. The correlations should be reversed but are the same in both manuscripts.

-The constrained relationship of winter pCO₂^sea anomaly is relative small (r=0.79),  maybe it is because the definition of winter months (November to April) in this study. The variations in different months should be quite different, especially in the transit seasons, i.e., spring and autumn. Definition of the focus season with less months, e.g. December to February, or January to March, might end up with clear relationship and higher correlation.

-This study presented several predictors including the two from Goris et al. (2018). As shown in Fig. 8, each predictor provides an different estimate of the constrained range of future North Atlantic carbon uptake. Which estimate is more plausible?

-How are the uncertainty range of the predictand C_ant*-uptake in Fig. 1, 8 and Table 1 calculated? I suppose they should be determined by the cross points of the linear regression line and the vertical lines of the observational uncertainty, but apparently it is not the case as shown in Fig. 1(c and e)  and Fig. 8.

- It is not very clear how to interpretate Fig. 5 and Fig. 7.  The results in the two figures seem contradict to each other, both upper ocean (Fig. 5) and deeper ocean (Fig. 7) have high correlations. How to combine the information?  In addition, L467-469: “…the deep ocean southward volume transport between 700m-4700m at 26N.” Is this statement based on Fig. 7? This figure shows the 700m-5300m and 21N has reached the largest correlations.

Minor comments:

-The information of figures are incomplete. I would suggest the authors to ensure all the figures are more or less self-explainable.

Fig. 3: the titles of x-axis and y-axis are missing, the y-axis’ title is relative easy to guess, but the x-axis is not so straightforward. The readers need to check back and forth of the context to figure out that it should be number of iterations.

Fig. 4: the unit of the presented variable is missing on both plots and in the figure caption. Why are the color shadings much lighter in Fig. 4c-d than in the Fig. 4b, as they are presenting the same quantity? The same question is also for Fig. 6c-f.

-Fig. 8: as specific model like CESM1-BGC is mentioned to perform well in L413-414, and more information and comparison can be made if the authors present the models with colors as in Fig. 1.

-Abstract L3: “A previous study…” needs to add the reference paper citation so that the readers get the context. From reading the main text, I guess this study refers to Goris et al. (2018).

-Abstract L5: “…winter pCO₂^sea – anomaly…”, but the previous paper (Goris et al., 2018) suggested the pCO₂^sea anomaly in summer (May to October) NOT winter (November to April). As the winter and summer are taken actually half a year in this study, respectively, I guess the counterpart season should be with the same magnitude of correlation but a reversed sign. So I am quite confused that this study based on winter months and the previous study based on summer months get exactly the same correlations as shown in Fig. 1c. 

-Some relevant details need to be described in this paper, so that the readers don’t need to refer to Goris et al. (2018) all the time. For instance:  how is the pCO₂^sea anomaly defined, is it relative to the annual mean or long-term specific season mean? Which time periods are 1990s, 1997s, and 2090s?

-L350, 354, 363, Figs. S01, Figs. S03 and S04 are inconsistent with the figure numbering in the supplementary.  

-L411: “…consistent which…” -> “…consistent with…”

-L487: “…averaged aver…” -> “…averaged over…”

I’m surprised to see this paper in review for GMD as it does not obviously meet any of the journal’s manuscript types. It seems like the direct utility of this work is thinking about how to guide observational strategies to constrain N. Atlantic carbon uptake. This is hwoever a call for the editor.

Major comments.

I find the manuscript comes across a bit as a dump of all the work the authors have done in this area, and as such, I feel it would benefit from some curating. The manuscript seems to be doing all of the following:

1. Identifying specific regions where people should be making observations to constrain future N. Atlantic CO2 uptake (and in doing so they refine existing published emergent constraints slightly).

2. Exploring how a genetic algorithm can be used to select the optimum area of observational sampling to constrain models.

3. Expanding on the mechanisms behind the emergent constrains that the authors have previously put forward.

4. Better understand which key processes are leading to uncertainty in projections of future N. Atlantic CO2 uptake (which links quite closely to 3).

As it is written it is doing 1, suggesting that it is doing 2, and doing a bit of 3 and 4 around the edges. The editor will be able to provide guidance on which of these a GMD paper should be doing, but I would argue that 2, 3 or 4 done fully would make the most useful papers, while1 is useful for a very specific audience. As it stands 2, 3 and 4 are the less developed parts of this manuscript. Perhaps it is OK to do all of these things, but if that is what is done, a much clearer structure needs to be imposed on the manuscript and introduction of what is being done and why, so that the reader knows what information they should be getting from each section, and can efficiently take what they need from it. My preference would be to be clear about what the manuscript is trying to achieve and focus the manuscript on that, bringing in the other bits perhaps only as part of the discussion.

Fundamentally I can’t see any mistakes beyond that raised by the other reviewer. I would echo the other reviewer’s comments about it being difficult to interpret some of the figures, and would add that the manuscript would benefit from some careful editing for readability.

Minor comments:

• Just a comment - I’m pleased to see the desire for mechanisms in emergent constraints!
• The title does not make sense. “Gulf Stream and interior western boundary volume transport as key regions to constrain the future North Atlantic Carbon Uptake” Should it perhaps read “Gulf Stream and interior western boundary as key regions to constrain the future North Atlantic Carbon Uptake”?
• It seems to me that the ‘competition’ described in section 2 might benefit from a more detailed diagram than Fig 2.
• Line 27 refer --> referred
• Line 37 “Despite many progresses” --> “Despite much progress”
• Line 37: ‘have not necessarily’ – be specific have they or haven’t they, or in what areas have they.
• 72: “could highly gain from” --> “could gain from”
• Line 214: “we advice against” --> "we advise against"
• Line 487 aver should be over