Comments on ‘Extension of high temporal resolution sea level time series at Socoa (Saint Jean-de-Luz, France) back to 1875’ by Khan et al. (ESSD)

I have no major objection to this paper. It is clear that a large amount of work has been done to provide the new data set from the Socoa tide gauge. However, all this work is rather let down by providing a draft that is full of minor English-type oddities which detract from reading it. The co-authors should really have helped the lead author with the writing. Therefore, I have done my best to correct things using track changes included in the attached pdf. I have also made some comments in capitals in the pdf. I will also send these directly to the lead author.

Otherwise, I don’t think all the detail of section 3.2.1 is needed. There is an excellent Wikipedia page for example which describes the equation of time.  And also you can quote for example:

Hughes, D. W., Yallop, B. D., and Hohenkerk, C. Y.: The equation of time, Mon. Not. R. Astron. Soc., 238, 1529–1535, https://doi.org/10.1093/mnras/238.4.1529, 1989.

But I agree there needs to be something to show the differences in timing in this section. You should make it clear that MST at this point is local mean time.

In sections 3 and 4.1, you can check for bad times or dates easily by comparison to tidal predictions that are based on a modern record you can trust. Errors of an hour (say) are easy to spot in the resulting tidal residuals. I suspect you must have done that but it is not clear – you just say a lot of tests were made. General reference could be given in this section to the Pugh and Woodworth 2014 book.

I am not sure that all the sub-figures are explicitly referred to in the text as they should be – the lead author can check.

I am afraid I have not looked at the ancillary material or the new data set itself.

I have no objection to my identity being revealed.

Comments on ‘Extension of high temporal resolution sea level time series at Socoa (Saint Jean-de-Luz, France) back to 1875’

Abstract

This paper presents the extension of sea level data at Socoa (France) back to 1875. New sea level data come from historic ledgers (i.e. record books with handwritten hourly sea levels) and charts (i.e. paper maregrams from old mechanical tide gauges). Both these supports were recovered from national and local archives, through extensive research. Important efforts were undertaken to reconstruct a continuous and consistent time series. Documents were scanned and digitalized, either manually or semi-automatically. Time data were converted to UTC time system. Successive vertical datum were investigated, to check that there was no jump between historic datasets vertical reference. An intensive quality control allowed to correct the data when needed (e.g. height or time correction). In addition, the sea level data were flagged when doubtful, e.g. during periods of slowing down of the clock, during periods of siltation of the stilling well (1956-1963 and late 1990’s) or when the floating device seemed to possible malfunction. The final hourly sea level dataset consists of ??? [not mentioned in the paper ?] years of data (instead of 54 years (not mentioned in the paper?) at present), spanning from 1875 to 2021. This new historic dataset will be useful for sea level climate studies, investigating variability, trends and long-term changes in mean sea level, tide and/or extreme surges.

General comments

This work is an important effort of data archaeology, which is essential for climate studies, and we strongly recommend to publish it. The data were not only recovered, but also corrected and flagged when wrong or doubtful (which is a very time-consuming task). The paper is well structured and generally quite easy to follow. However, some parts are not always very clear and the general writing could be improved (see below in detailed comments). Some arguments could also be added to demonstrate in a stronger way the importance of data archaeology, see for example Talke and Jay (2017). Another general comment: the results of this study being the sea level datasets, the raw and final datasets should be described more precisely in the Data availability section (or somewhere else in the paper), and in the given repository (see below in detailed comments). For example, the temporal resolution (5’, 15’ or 1h) of each dataset (raw data, digitized data, final dataset) is not always very clear (see details in specific comments). Finally, the perspectives could be further developed, detailing for example how many stations could be recovered along the French coasts and/or how many station-years are available in archives. This could potentially motivate future investigations.

Specific comments. Most of them are minors or suggestions.

- line 9 “ledgers and charts” we suggest to precise somewhere what we are talking about, describing ledgers (record books with handwritten hourly sea levels) and charts (paper marigrams from old mechanical tide gauge)

- line 12 “at hourly (for ledgers) to 5-minutes (for charts) sampling” it is not clear what is the temporal resolution of the raw dataset/final dataset, once corrected and flagged. This could be clarified in the paper, for example Table 1 (with 2 columns, sampling of raw data/sampling of the final data), or in the Data availability section, as the choice is made further to digitize only hourly values

- line 13 “Analysis of the precise levelling information reveals that the datum of the tide gauge site has been stable”, it should be mentioned that were are talking about vertical datum continuity, to avoid any confusion with the vertical stability of the ground

- line 15 the data flag could be mentioned in the abstract: despite “siltation is found to be a recurrent problem of the stilling well”, such periods were successfully identified, and corresponding data were flagged as doubtful

- line 29 “data archaeology” (without capital letter), possible additional reference to:

UNESCO/IOC 2020. Workshop on Sea Level Data Archaeology, Paris, France, 10-12 March 2020. Paris, UNESCO, IOC Workshop Reports, 287, 39 pp. English. (IOC/2020/WR/287) https://unesdoc.unesco.org/ark:/48223/pf0000373327

- line 46: “GMSL rise also raises questions regarding associated long-term changes in tide” the beginning of the sentence (GMSL rise) suggests that GMSL rise is the main driver for changes in tide, but it is not the only one, and the following examples (Pouvreau et al., 2006; Colosi and Munk, 2008) do not really conclude that MSL rise is the driver of observed changes in tide: Pouvreau et al. (2006) did not find any significant trend in M2, but rather an oscillation of amplitude 1.1 cm and period 141 years, which remains unexplained; Colosi and Munk (2006) attributed changes at Honolulu to a 28◦ rotation of the internal tide vector in response to ocean warming. Other papers discuss more directly the impact of MSL rise on tide (e.g. Pickering et al., 2012, Idier et al., 2017; Schindelegger et al., 2018). We rather suggest a general introduction to this paragraph (as for ESLs, see next paragraph), e.g. “Long-term sea level datasets are needed to investigate changes in tide...”.

- line 49: “Pouvreau et al. (2008)” should be Pouvreau et al. (2006) or Pouvreau (2008)?

- more generally for this paragraph on tide changes, some research works show that changes are not linear but rather with break points, which is a strong argument to go back to the XIXth century when possible. This could be discussed further in the introduction. See for example the values of M2 at Eastport, Portland, New-York in the 1860s, which are not consistent with the large increase observed in the XXth century (Talke et al., 2014; Ray and Talke, 2019; and Fig 2b in Pineau-Guillou et al., 2021).

- more generally, for the paragraph on extreme sea levels, it is well known that extremes vary at first order with MSL. Letetrel et al. (2010) is mentioned here for Marseille, but many others demonstrated it at larger scale, e.g. Menendez and Woodworth (2010), Wahl and Chambers (2015), Marcos and Woodworth (2017). The same way, IPCC (2021) reported that sea level rise is the first driver of changes in extremes sea levels. More interesting, once this contribution is removed, storm surges display large strong interannual and multidecadal variability, and it is challenging to separate the long-term trend from the natural variability (i.e. climate variability in link for example with the North Atlantic Oscillation). In other words, trends detected on a short period could be the signature of the multidecadal variability rather than a long-term trend. These are strong arguments to go back to the XIXth century in sea level data, and could also be discussed further in the introduction. Another argument is that for extreme value analysis, longer times series means uncertainties reduction, which means a better risk assessment.

- line 58 “They illustrate how long-term sea level can help to separate the relative contribution of climate, and local changes.” I would rather say that they separated the contribution of the natural variability (climate variability linked with the North Atlantic Oscillation) with a long-term trend (which remains unexplained).

- lines 73-74: the temporal resolution (hourly ?) of present Socoa record could be added

- line 91: move “rescued” before “in this study”, and this unclear sentence should be rephrased

- lines 102-104 this paragraph could be moved at the end of previous paragraph (line 99), which also describes the tide gauge. Next paragraph rather refers to water level records.

- line 106: what is the difference between Chazallon/Brillie tide gauge, is it the same mechanism?

- line 107: “large type model”, are there other models?

- line 120-126: this refers to a tide gauge from 1942-1944, whereas the paragraph is “1950 to 2004” which is confusing. This could be moved to another subsection. It could also be rewritten more clearly, and it could be mentioned explicitly that these data are finally considered (though the tide gauge location is different).

- line 125: “the other side of the Socoa bay”, location on Fig. 1?

- lines 127-151: what is the temporal resolution of modern instrumentation?

- Table 1: we suggest a separate column for the “Sampling” and “Time System”

- Table 1: “Sampling” column, are we talking about the sampling of raw data? Digitized data? Final dataset? This is not clear in the text.

- Table 1: What means “Highres and hourly” mentioned in column 1 since 2011, whereas the corresponding “Sampling” is 1 hour?

- Fig. 2 (b): Chazallon, Brillie and modern instrumentation periods could be mentioned on the figure (suggestion)

- lines 166-216: the structure of these two paragraphs is not clear. “3.1.1 Ledgers” rather introduces the Chazallon period (ledgers and charts description, even if charts are not used) and “3.1.2 Charts” introduces the Brillie charts. Please restructure moving the charts description from 3.1.1 to 3.1.2, or change the titles accordingly.

- line 176 “hourly intervals” for ledgers (to speed up the digitization process), whereas as 15 mn is the sampling in Table 1, again please clarify the temporal resolution of each dataset (see previous comment)

- line 184: the whole recovered archive “is scanned” could be “was of sufficient quality to be scanned”, which was not the case for Chazallon-era charts

- line 187: “for applying corrections, where appropriate” it is not clear here which corrections are we talking about

- line 194: “several categories” could be detailed as an introduction of the paragraph, for example “3 categories depending on their conditions (good, mildly or strongly damaged from mould, faded)”

- line 210: “the overall process was time-consuming”, yes, and this could even be a separate paragraph to underline this aspect (suggestion)

- lines 227-246: this paragraph on time conversion is quite long and not very concise, it could be rewritten in a simpler way. The equation of time E(t) gives the time difference between Apparent and Mean Solar Time, it can be expressed as a function of (M,t), with M=6.240060 + 6.283019552t. We consider the formulation of the Bureau des Longitudes (2011) for E(t), despite it is applied for 1900-2100.

- line 246 “minor errors”, which order of magnitude?

- line 254: no capital letter at “typically”

- for this section on time conversion, it should be clearly mentioned at the beginning of the section that the objective is to convert the time in UTC (if correct)

- line 267: 3.3 should rather be “Vertical datum continuity” ?

- line 274: “it was possible to reconstruct the relationship of the tide gauge and tide pole zeros to the current benchmarks” not clear what is the tide gauge zero here, are we talking about the Chart Datum?

- lines 288-290 the origin of the difference of 18 cm for Chart Datum is not clear, does that come from heavy siltation?

- line 291-292: “It appears that this tide pole is a different from the tide pole during (1873-1920), and the hydrographic zero at 24cm below the zero of the tide pole.” The sentence is not clear, and should be rephrased. Possibly, the presence of the two tide poles could be introduced earlier, rather that at the end of the paragraph?

- lines 293-300: this paragraph is not very concise, and could be rephrased

- line 301: this is the main result of the section, it could be mentioned at the beginning of the paragraph, for more clarity

- lines 308-319: this paragraph on flags could be a dedicated subsection “Data quality flag”

- line 304 “Following these two steps…” the sentence is unclear, it should to be rephrased

- line 316: “The idea is like…” it should be reformulated

- line 328: “tide gauge journal”, ledger?

- lines 339-342: this paragraph should be moved to the end of the section, as it introduces the following section

- line 344: “hourly values” again, please mention more clearly the time resolution of the datasets (e.g. Table 1 refers to 15 mn, see previous comments)

- line 349: “The applicable corrections are applied as described above.” Why mention this here, whereas previous section (4.1) already focused on corrections?

- line 351: it is difficult “to” apply

- line 352: “These values are flagged as values with low confidence (third bit in the flag set to 1)”, which order of magnitude of the slowing down of the clock? Is it possible to mention how many of the data are concerned? (for example in %)

- lines 354-357: The title “Delayed rising/falling curve” should rather be “Possible malfunctioning of the float device” (to be consistent with other titles, referring to the potential problems rather than their impact on data). Does this malfunctioning of the floating device leading to delayed rising/falling curve has been already reported and referred? If yes, the references could be added.

- lines 358-392: “4.2.3 Siltation” This paragraph on siltation could be more concise, giving clearly the periods with problems of siltation, and mentioning which data were flagged accordingly.

- line 361: “The first major siltation problem with the data recording was noticed within the first few years of operation” which years? This problem of siltation does not seem to appear on Fig. 5?

- line 365: “After starting the reoperation of the tide gauge in 1950, the stilling well exhibited siltation and blockage related problems (Robertou 1963).” Same question, on which period exactly? Are we talking about 1956-1963?

- lines 380-382: siltation problems are mentioned during 1956-1963 and 1996-2000. It is not clear how these problems were solved, was there any intervention on the tide gauge? This could lead for example to recommendations to avoid this type of problems.

- line 383: “4.3 Buddy checking” this section is mainly “Buddy checking for vertical datum continuity” and could be part of Section 3.3 Vertical datum continuity (or at least, the title could be more precise)

- line 385: “The difference with nearby...” precise monthly mean sea levels differences

- Fig 6: the legend is unclear and should be rewritten. Difference between monthly MSL at Socoa and Brest (black) and Santander (red). The fact that MSL over the period 1965-2000 has been removed should also be mentioned.

- line 408: “However, in the Santander minus Socoa timeseries (red), we see a jump during (1976-1980), indicating a potential shift of 5 cm during that time at Santander.” It is not useful to repeat again “in the Santander minus Socoa timeseries (red)” (already in previous sentence)

- line 408: why focus on this “jump”, whereas there are others similar, as for example in the 1990’s?

- line 409-411 “Recently, Marcos et al. (2021)...for Socoa.” this comment should rather be at the beginning of the paragraph, when introducing the data (line 395 for example).

- lines 412-425 the section “5. Trend analysis” is very short (around 10 lines), it is difficult to consider it as a full section, and it does not seem to be completely finished. Supplementary “5 Inflexion point analysis” could be added in this section (rather than in the Supplementary). Why is this analysis conducted only over the 1875-1920 period rather than on the full period? What are the possible causes of these inflexion points?

- line 412: “The trends estimates” Trends of which data? Hourly final datasets?

- Table 2 Period “Available”, which is the period available for each harbour?

- line 414: “The benefit of a long time series is clear here – longer the time series, tighter the error bar.”, yes, another benefit is that changes are not always linear, and inflexion points can be investigated, as shown in the analysis below.

- line 426 Data availability This section is of great importance as the final sea level dataset is the main result of the study. Details should be given to be more precise. Note that there are many files in the repository, the readme is quite short, and there is no detailed header in the final file.

This section could at least: give the name of the file corresponding to the final sea level dataset, as well as its vertical reference, time system, level units, start date, end date, number of years with data.

In addition, concerning the file itself, it would be very useful for users to have these essential information in the header (which is not the case in the repository), and add the 4-bit quality flag in a third column (Time/Sea Level/Flag)

- lines 431-450 the conclusion section is also very short. It could be improved, with a better description of what has been done, and a development of possible perspectives. A synthesis of the corrections/flags would be appreciated. What are the main corrections? Which percentage of data is finally corrected? What are the periods flagged for siltation?

- line 434: “international sea level databanks”, such as?

- line 437: “siltation” periods of siltation that are flagged?

- line 438: “associated corrections”, synthesis of these corrections?

- line 438: “final sea level dataset” very brief description of the file (see previous comments)

- line 438 “raw data” very brief description of the file, as the differences between raw data and final data are not always very clear in the paper (see previous comments)

- lines 448-450: “there are many more stations in France where the existing sea level record could be extended” this paragraph could be developed, which stations have already been identified? How old are these data? How many additional years could be extracted?

The perspectives could be further developed, at national and/or international scales. For example, Talke and Jay (2017) reported that “more than 6,500 station-years of previously lost or forgotten tide data have been identified” (http://archives.pdx.edu/ds/psu/21294)