14/12/2020

Dear Lucia and Guy and Dr. Lazure,

Many thanks for responding to the comments of the reviewers and for the V2 of the
paper. I read it last night. As I implied last time, I am not sure I buy into the
NAO links you suggest, but I am prepared as an editor to go along with the suggestion
for this paper, hoping it might stimulate further work. That is, as long as the paper
itself is respectable generally.

I did not send this V2 to another editor or reviewer as you can see that there are
some things I think could be edited, I don't think most will take you long. What I suggest is
that you rapidly make a V3 following my comments below and we can take things from there.

Phil

There are 2 main things I came across that worried me:

In Figure 12, your suggestion in (a) and (b) is that higher pressure in the south (in (a)), and
the implied redistribution of water masses, leads to larger M2 (in (b)). ok, fine - that is
what you are implying in Figure 6 (I think mention of (a) and (b) here should refer in the
text back to Figure 6).

(I noticed that in your V1 you had figure (a) but said it was MSL and not air pressure -
I assume you corrected that now.)

But then in (c) we have higher pressure in the south in Jan rather than July and yet M2 is
larger in July. That is the opposite way round, or maybe I am missing something.

I thought at first that (c) might have been mis-labelled from July-Jan but I think it is
correct - see for example Figure 3a of Chen et al., Tellus A, 49, 613-621. They show July-Jan
but, allowing for the sign, it agrees with your (c). And as we know, during the winter (Jan)
the Icelandic Low deepens a lot, so blue in the north ok.

Anyhow, I am confused by all this. Could you investigate?

The second thing is that at line 273 you say the MSL in the following regressions was
after VLM correction i.e. Figure 5(b). However, it seems to me that you should surely
use the uncorrected MSL (Figure 5a), which is actual water depth depth of whatever origin
(VLM or whatever, who cares?) and so has some physical logic to it, rather than corrected for
VLM which is a more scientific concept to do with climate budgets etc. You clearly disagreed
(line 273) so please say why you made this choice.

Some more detailed comments are as follows, many trivial:

7 - The secular trends ..

what does 'overall' mean. Perhaps 'most of them ..'

12 - distribution of mean sea level (corresponding to water depth) from

16 - have been observed

21 - a few

24 - a few ... in many estuaries

27 - such as changes

40 by Pickering

where does the -20 to 20 cm refer to, around the whole ocean?

46 characteristics to

49 on the M2

60 mentioned

61 of the NAO

66 observed tidal changes

96 found no differences in the

97 using either T_Tide

98 temporal sampling was 3 hours

103 years --> station-years

104 ditto

116 centres

131 were low-pass filtered

135 We employed ... (20CR) data set version ???

137 sea-level pressure at each grid point

why do you use 2015 when you use 2018 for the tide gauge time series below?

139 pressure at each grid point.

144, 146 - what do 'very similar' and 'high correlation' mean? I think you have
to refer the reader to the more quantitative statements below

149 have noticed

152 and the shape

153 such as the migration

155 such as narrowing

157 that have gradually

I don't understand the point being made here, and also below. You are implying that
most of the Brest data is now affected by engineering when it is was more open ocean
before? In which case you can't really claim it as part of the wide-scale variability
in the last century mentioned below? I think this needs rewording. Also I think you need to say here
that your focus below will be on the 20th century (after 1910).

164 - no obvious linear trend

line 1 Fig 2 caption - .. amplitude 9-year filtered .. The stars in (b) in the 1860s

179 0.8 cm respectivey

189 in Figure ... in Figure

193 in Figure

195 only 0.36

17.5 cm mean amplitude

I don't understand the sentence 'The very slow ...'. You are talking about normalised
amplitudes here.

198 Gulf

203 trends ... provides some consistency with the hypothesis ... formulated from the
analyses of Brest and otherdata in Figure 2(a) i.e. ..

But then see above, you seem to rule Brest out!

what is climate-scale? This seems to be from your old title.

204 introduces some breaks

210 I think a problem with Halifax is the big gap. Who knows what changes to instrumentation
and the port etc. there were in between

215 that the later results

253 - the normalised trends? But these are not shown in the Table. The unnormalised trends
vary a lot so I don't understand line 254

227 - I don't understand this 9 and 3, and you have two 'negatives'. You have 18 stations
in the table and 5 are from NE Atlantic. How do you get to 12? It looks to me that
10 are negative and 8 positive

231 The largest trends for 1910 onwards ?

252 .. to 5 i.e. their ..

255-258 - this sentence on modelling is just rambling and the mention of 10 km,
coarseness and rough is all the same thing. I would drop (1), leave that alone,
and just keep (2) which is true.

261 - The MSL records ... include

273 see my second main question above. I am not convinced you have to make VLM
corrections, and then you could drop figure 5b, but I realise you would have to
recompute the regressions - but that is easy enough.

276 MSL as well as climate indices

273 - see above

277 - such as

281 act on

fig 5 caption line 2 - MSL values are filtered using 9-year windows.

294 on the tide

295 .. variability, the NAO index tends

fig 6 - I was intrigued that the NAO and AO indices have different magnitudes. Maybe ok.
But why don't you use normalised indices anyway?

Fig 6 line 1 - M2 amplitudes

301 are filtered using the same

Fig 7 line 2 - filtered using

313 - why do you exclude these 4 for having no NAO correlation? They would appear to
be ideal candidates for model 1

321 - 0.7 does not sound like 'no significant correlation'

322 of the NAO

323 adding an additional regression ..

327 contribution of the NAO

328 whereas it is negligible

Fig 8 - see above - why don't you have the other 4 in this plot. Also I can't see
one with a red bar of 0.7 mentioned above. Also please remove the 1.2 on the y-axis, the
max for that is 1.0

line 2 - filtered using the

fig 9 line 2 - the size of each large dot

fig 10 line 1 - filtered using

329 sensitive to the NAO

332 at all four stations

The text here doesn't mention the Brest sub-figure

340 pressure gradient

358 - drop 'very'

360 They ran

371 - a few cm, similar to its variations seasonally?

figs 11 and 12 - say in caption what the contour intervals are

fig 12 - see above

385 - whereas variations appear between stations in the ..

386 from one station to

395 on the M2 tide

398 and currents

405 studies (e.g. Ray, 2006) in

423 - why pick these 2 places and refer to Kemp? I would drop this. There must
be many places you would like to look at

storminess on the tide

459 - space after am2

figure (a) - you should say in the caption or text why this is detrended and (b)
isn't

fig B1 - the delfzijl and hoek van holland have enormous increases in M2 around
1960 which look very odd. Is there no previous mention of this in the literature?
Adding them hasn't benefitted the paper much if they are so obviously different.

references - some titles are not capitalised (e.g. Araujo and Pugh) but some are
(e.g. Colisi and Munk). Some journal names are given in full and some are
abbreviated. Please use the OSD style.

27/10/2020

Dear Lucia and Guy and Dr. Lazure,

Many thanks for the latest version (V3) of the paper which I have read. Thanks for
attending to the 'MSL without VLM correction' issue. I think that reads better now. But
I have two remaining issues with the paper.

One issue is the usual set of small edits which I list below. These will not take you
long. I am sure you are getting as tired of these things as I am but until the
second issue is dealt with I may as well continue to send along these minor edits, which you
can accept or not as you want.

But the second is, as I mentioned last time, to do with the discussion of Figure 12
and the air presssure pattern on seasonal timescales and its implications for M2
amplitude being the OPPOSITE of the relationship on interannual timescales.

I see that you changed Fig 12 (a) and (c) to focus on the North Sea area, ok. But the
same problem remains, the inference of higher pressure in the south leads to larger
M2 (as one learns from (a) and (b)) is OPPOSITE to the inference on seasonal timescales
from (c) and (d). You have disposed of this problem with a minor edit to the text at line
361 to say 'and obtained values close to the ones in Figure 12 (a) in terms of magnitude
(a dozen of hPa)'. Well, yes, they are similar in magnitude but not in sign!

Your problem to my mind is that you are trying to make too strong a connection between
the air pressure patterns and M2. Let's leave Huess and Andersen aside for a moment.
I am reasonably convinced by the work of Mueller (2012) and Grawe et al (2015) that the main
reason for the seasonal M2 variability is the seasonal stratification. They may be wrong,
but they are the latest sets of work on this topic. If that is the case, then
you can try till the cows come home (as we say) to make a connection between air pressures
and M2 on seasonal timescales that will look anything like the pattern you show for
interannual timescales (if also true). I hope you see what I mean. The seasonal M2
variability could still of course be connected to water depth of a similar magnitude
as you have interannually but for entirely different reasons.

So I think the para 'We conducted .. available observations' needs to be completely
rewritten so show that you understand things seasonally and not dispose of Muller etc. at the
end of the para as 'other hypotheses'. They are not - they are the best
hypotheses at the moment. You should by all means mention the Huess and Andersen work of course,
and keep all of Fig 12 if you want, but make it clear in the text that the simple association of patterns does
not have the right sign and not try to hide that with disingenuous words like 'in terms of magnitude'.

I have actually spent some time on trying to understand why Huess and Andersen obtained
maximum M2 in summer (which I assume is what their phase of 180 degrees means) by
turning on winds for surges, which must be dominated by winter westerlies as in your
Figure 2(c). I haven't fully understood all that and perhaps that can be returned to by
someone else again, but for now I think you
should not base most of your seasonal discussion on that, rather than Mueller etc.

I hope you can be patient and address this issue seriously. Until you do that I do
not feel I can accept the paper. I am conscious that
I am functioning here more as a reviewer more than an editor, so I have copied this
set of comments to the executive editor (as I did for my comments on the V2 version)
and he can let me know if I am not doing my job correctly. I hope you can see that I
am on your side here (because I am intrigued if there really is an NAO connection).
However, I can also if you wish send the latest draft to the original two reviewers
although I doubt if they will be pleased to see it again.

Thanks again.

Phil

Small comments on the text and figures:

Would you be willing to change the title to read '.... tide along North ...'? It is up
to you but I think that would be better.

line 1 - M2 along North ..

7 - up to 2.5 mm/yr at Wilmington since 1910.

19 - and Boston

20 - at the North Atlantic basin scale .... and at a quasi-global ..

21 some coastal stations

23-24 I would remove this sentence. This example is from the Pacific which looks odd in
a paper about the Atlantic. You do mention this paper lower down ok so it will be included.

If you agree then: For example, Ray and Talke ..

32 - Cap --> Cape

45 - Hawaii

50 - waters

65 The first section below ['below' because this isn't the first section in the paper]

70 Tide gauge selection

73 by the Rijkswaterstaat (RWS) in the Netherlands.

77 - 249 from UHSLC

93 - .... in 1846, 1879 and 1896 respectively ...

96 developed

99 - I guess this is optional at this stage, but don't you think 3 hour sampling is a bit
coarse? Have you checked that this sampling results in the same M2 amplitudes as for hourlies?
You could test that using an hourly series and then have something like:

... was 3 hours. (We checked with an hourly time series from Brest, with a comparable M2
magnitude to the Netherlands, that 3-hourly sampling
did not result in a reduction of M2 amplitude in a tidal analysis compared to hourly sampling.)
[If true of course]

100 .. seasonal modulation affecting the computed amplitudes.

line 3 of Table 1 caption: ... since 1990 up to 2018 in each case (standard ..

125 - I would drop 'with yearly values'. What does that mean anyway, you are using winter
values

128 - the NAO index

137-140 - I would drop from 'We computed' to the end of the para and replace with:

However, we made use only of data from 1850 to be more consistent with the temporal
coverage of the tide gauge measurements. This will be discussed in Section 4.

The reason for this is that the reader will ask why you defined winter as Dec-Feb when the
NAO and AO indices are Dec-March. But I come back to this below.

142 - you use US spelling here (harbor) and UK spelling (harbour) at line 163 and elsewhere.
Please check you use standard spelling throughout (preferably UK).

155 - .. Cuxhaven could therefore be sensitive

162 - would this link not be better as a reference in the normal way?

168 - drop comma after result

176 archaeology

182 .. magnitude of unnormalized M2 ..

199-200 I still don't understand this sentence about small amplitude at Key West. You are
discussing normalised changes so it shouldn't matter what the actual amplitude is.

209 between the 1870s

215 However, note that ..

234 - (1) some spatially-coherent changes .. [you have two recents in the same sentence]

238 - I think 1990 here should be 1910 or the paragraph doesn't make sense. Right? If so,
I would move this para to line 230 so all the information for 1910- comes in two paragraphs,
followed by the one ('The trends') on 1990-.

264 - with 9-year

278 - current profiles .. shelves ... modifies

280 circulation on the tide

When you computed these correlations and the p-values did you allow for serial correlation?
If so, say so. If not, the correlations will be less significant than you think.

295 - (the NAO index ..

300 - two indices are closely related.

310 - NAO and MSL at each station: there ... below. [i.e. drop the brackets]

324 - ... dependent) naturally captures better the ..

[this is an obvious statement!]

325 observed since 1990

333 ... (Figure 11(b)). (We define winter here as December-February.) This way ..

335 .. onshore by westerly winds

338 I don't understand how you infer -15 to 24 from Figure 12(a). I see +/- 10 on average.

342 .. yield centimetric changes in M2 amplitude.

343 is roughly un agreement

347 Note that the M2 amplitude

But the para 345- needs considerable revising, see above

380 - since 1910 and since 1990

Fig 12 - see my comments above, but can the right plots be moved more right as the 'hPa' of
the left is confusingly close to the 'M2(cm)' on the right

395 tends

411 may not be

10/11/2020

Dear Lucia and Guy and Dr. Lazure,

Many thanks for the latest version (V4) of the paper which I have read. I have decided to
sign off on it now as we are in danger of going on for ever. I understand completely
why the reviewers did not like the first version (and I supplied many comments on that myself) but
the various incremental changes have helped to improve the document itself. Just like you, I
remain intrigued by the possibility of a connection between the NAO and M2 changes, so I think
my job as editor is to bring that possibility to the attention of readers. I am hoping that
this paper encourages more work that perhaps you could do yourself later on.

As for this V4 version, I found a couple of other things to mention, see below. Please could ensure
that you attend to these when uploading your final (V5) version? I will not need to see that again.

Many thanks for supporting the special issue.

Phil

line 276 - current profiles

292 - in your last response you said that you not take serial correlation into account
but that a statistical test showed that it existed.

Could you please discuss this with Guy who understand statistics well? I think there is no
impact on your presentation of actual correlation coefficients (or r or r-squared values as in
Figure 7). The problem is that the computed p-values will be too optimistic. So I suggest
to be clear that at line 292 you say something like:

As a rough test of significance, the correlations obtained are considered as significant
only if a simply-computed p-value (without consideration of serial correlation) is lower than
0.05 (95% significance level).

326-358 - I think this reads better than you had before. Thanks for revising this. However:

331 - you say that stratification is 'the main hypothesis invoked in Ray (2006) to explain secular
changes in M2 amplitude in the Gulf of Maine'.

This is not correct. If you read his paper again you will see that Ray concluded he had no idea
what could be causing the changes. The words 'stratification' and 'temperature' do not occur at
all in the paper. However, stratification is indeed considered as (I would say) 'one of main
possible hypotheses' and not 'the main' in Ray and Talke (2019). Please read this again to check
their actual statements.

line 350-1 - I still can't see how -15 to 24 cm relates at all to Figure 12, which has a maximum of 12 mbar,
so that would correspond to +/- 12 cm, wouldn't it? I read your last response but I still
cannot look at that figure and conclude -15 to 24. Apologies if I am misunderstanding something
(but then so will other readers).

453 - the title of the appendix comes after the figure in that appendix.