Dear Dr. Tinker and co-authors,

The two reviews of your manuscript differed a lot. Referee #1 indicated some major problems, while referee #2 has a few suggestions for improvement as well.

I think you have only partly answered satisfactorily to these critiques. I think the manuscript has potential as you explained now what the real aim is, answering the strong critique of referee #1. However, this could be done better and somewhat more extensive.

Some of the answer to the comments by the referees were to my opinion not satisfactorily addressed. I have indicated that below. Also below are my corrections and suggestions for improvement. The manuscript needs major revision.

Please note that for my comments I used the author response file (with tracked changes).

You are using different data sets from other sources. Please check again the fair data use statements of those data providers whether the acknowledgements are sufficient and correct.

The references are not according to the Ocean Science format. Some are incomplete, for example:
Bourdalle-Badie, R. and Treguier, A. M. (2006). Please go through the references for possible errors or incompleteness.

Referee #1 point 1: What did you change in the manuscript following this comment?

Referee #1 point 2: I agree with the authors that a more complex analysis is not needed here. However, one could also briefly discuss what kind of physical mechanisms are at play. I encourage the authors to expand the discussion in this respect to some extent.

Referee #1 point 3: I think this is a legitimate argument by the referee. It would be good if you could go deeper into this.

Referee #1, point 4: Please answer this question by the referee: “However, just as for the second approach, the found correlations are not new and sometimes the correlated variables even miss an underlying physical connection. For example, how should T and S at the northern boundary (65°N) influence the NWS?” This is an important point.

Referee #2 Page 5, line 11:
“I would assume that the SST assimilation in the CO5 reanalysis also disturbs the heat fluxes. Please give some information about this problem.
Author’s response: The SST assimilation will improve the SST field, and will affect (improve) the upward radiation. The downward radiation that we consider in this paper is not affected. I have said this in the text.”
Editor: I am not convinced by your reply. Please explain better, especially in the manuscript.

P9, L27 The sentence does not seem to be correct. Should it start with “Giving”? Please rewrite the sentence.

P12 ,L30-31 The link between the output and the abbreviations is not clear at all
P12, last paragraph (and continuing on next page) under 2.5.1 Every sentence starts with “We”. Please try to change this to enhance the readability.

Section 2.5.1 “We compare time-series of the shelf response to time-series of the drivers to identify important relationships. We note that a statistically significant correlation does not imply a causal relationship, and so we interpret the spatial patterns of the correlation coefficients to help interpret the underlying mechanisms behind the correlations. We consider it beyond the scope of this study to undertake sensitivity studies to explore any mechanisms in detail.” I only partly agree with this. I think sensitivity studies are too much for this exercise indeed. But some ideas about the mechanisms would certainly be most welcome, the more so as relationship only based on correlations might just be showing things that make no sense at all, without any physical background.

P14, L12 “ …of surface salinity in the southern North Sea Ferry Box …” This is strange. I think you mean the salinity of ferry box data.

P15, L1, L5 Do not use psu because this is not a unit.
P19, L1 UV10 may not be common to all readers. Please define.

P20, L11-12 “Further improvements to the reliability of the NAO forecast that are in principle achievable (e.g. Eade et al. 2014) would be expected to lead to further improvements in the forecasts for NWS regions.” This is a little bit vague. The only reason that such a contention could be interesting to the reader is if these improvements are mentioned and explained. Please expand on this to some extent.

P20, L13 Is this indeed preliminary? That is not how I understand this paper. You are assessing the possibilities and chose three methods. It would be disappointing if this is only preliminary.

P20, L24 Whole point C. I find the structure to be deviating from the two earlier points. What I expect is a clear assessment whether this method works or not, or only to some extent. Maybe it is contained in the text somehow, but certainly not very clearly. I suggest restructuring this point C.

Figure 1
The map is not clear at all. It should be bigger, the lines demarcating the coast would better be in a different color, and the names of some countries should be added for orientation. By the way, what is called “southern North Sea” here is also known as the Southern Bight of the North Sea.

Figure 3 caption “as 2-year running mean” and not “with 2 year running mean”, I think.
Figure 3 It is not shown if the data are spatially comparable indeed. Which area in CO5 NWS reanalysis and in GloSea5 was used for averaging? And is this exactly the region of the ferry box data?

Figure 4 caption Twice “observed”. Time-series (typo).
Figure 5 caption: “Lower right panel shows the location” must be “Upper right panel shows the location”

Figure 6 These are really many panels. I am not sure whether this will be readable and understandable at all. If possible, could you split this figure in two or make it more readable in any other way?
Figure 7 Same as for Figure 6
Figure 8 Same as for Figure 6
Figure 9 Same as for Figure 6

Figure 10 The panels are too small. It is hardly possible to extract information from these plots. This must be changed.
Figure 11 Same as for Figure 6, even more since there are 18 panels in it.

With best wishes
Mario Hoppema

Dear Jonathan and co-authors,

Your revised manuscript has been much improved and I think it is shortly before publication. I have gone through it and below is my list of final comments, which I encourage you to account for when submitting the final version of your manuscript.

For someone who is not really familiar with the region, it is really hard to learn where everything is occurring. In the text some names of coastal towns are mentioned, but also the Scottish and Danish coast, English Channel, Irish Sea, Armorican Shelf, Celtic Sea, for example. It would be most helpful if all these geographical names used would appear in one of the figures. This also holds for the different currents that are mentioned throughout the manuscript.

P1, L11 We show that the GloSea5 system is inadequate for simulating the NWS directly.
(“can be” is strange here) (delete the last part of the sentence: and so move on from A).

P1, L15 I suggest: As to point C) we find large-scale …
P5, L12 This is the first time SSS is used. They should be defined here.

P6, L12 Wegener (typo)

P6, §2.4.2 and 2.4.3 You are using data from particular locations on the NWS. Some of the time series stations are shown in Fig. 5, but other data (for example, Hoek van Holland-Harwich) are not. Please show them in one of the figures.

P11, L40 Please be more specific with “its skill, resolution etc.” What does the etc mean?

P12, L1 Suggestion: We start to explore … (delete: can; you are going to do that, right?)

P14, L10-12 I am not sure what this sentence wants to convey.

P14, L19-26 Many times the word “may” is used here. I think in many cases this is unnecessary and tends to give the reader the impression that nothing is certain or even doubtful. Similar for some cases in the Conclusion section.

P14, L28 Here it is stated that this is a preliminary study. In your response you wrote that you had changed this after my comment to the previous version of the manuscript. I don’t think this study is preliminary. Maybe a better word is “exploratory”.

P14, L29 … its output limits direct application …
P16, L21 Wegener (typo)

Thanks and best wishes
Mario