RC1: 'Comment on essd-2022-66', Anonymous Referee #1, 09 May 2022 reply

Review of Hoppmann et al., “Mesoscale observations of temperature and salinity in the Arctic Transpolar Drift: a high-resolution dataset from the MOSAiC Distributed Network” submitted for publication in Earth System Science Data.

This is a solid manuscript that I recommend for publication after mildly major revision. The writing and figures are generally very clear. There are some confusing explanations that could use some attention.

#We thank the reviewer for the positive feedback. We tried our best to improve the explanations based on the detailed and very much appreciated comments. We hope that the new version is much clearer.

My major comment concerns a disconnect that I see between the Introduction, which has a significant discussion of the submesoscale, and the rest of the paper, which lacks analysis on these scales. See my comments near the end. The manuscript’s title has the word “mesoscale” and the eddy analysis supports this. But why is there so much discussion of the submesoscale in the Introduction?

#This issue was also caught by another reviewer, and actually we very much agree to this. Presumably this is because for quite a while we weren't sure how much science should go in the paper, and in the end there was a mismatch between the intro and the preliminary analysis. To resolve this, we now moved a significant part of the intro in the discussion part, put a little bit less focus on the submesoscale, and included a new figure to illustrate better the mesoscale aspect of the dataset.

 

Line 61: Perhaps change “mesoscale scales” -> “mesoscales”

#Done

 

Lines 93 and 101: “upstream in the Transpolar” ie insert “in”

#Done

 

Line 101: “2019/20” -> “2019”, yes?

#Done

 

Line 114: cut the word “already.”

#Done

 

Line 116-117: “the measurements ensued away from the inflow of warm Atlantic Water through the Fram Strait and as close as possible to the Siberian continental slope” I am not sure what you mean by this. Also, I suspect that “ensued” might not be the best choice of verb.

#The expedition was meant to capture local processes at different locations within the Tranpolar Drift system. The drift originates on the Siberian shelves and reaches to the Fram Strait and the Canadian Arctic Archipelago. The conditions contrast those parts of the Nansen Basin dominated by the warm and salty inflow of waters of Atlantic origin, where conditions near the surface are significantly less salines than in the Transpolar Drift. We have rephrased the sentence to better explain that idea.

 

Line 119: “to achieve that aim” What aim?

#The focus was to measure conditions and resolve processes in the upper mixed layer, seasonally deepening and subsequently restratifying, as well as the underlying stratification (upper halocline). We clarified by changing the text.

 

Line 129: An egg has a pointy end; this buoy hull looks more oval, or perhaps "flattened oval"?

#changed "egg-shaped" to "oval"

 

Line 133: Your buoy naming convention is a bit odd, ie the use of the capital letter “O” instead of the number zero “0.” But it is your choice. As noted below, this gets sloppy in the section about the eddy, when both “O” and “0” are used.

#The sloppiness has hopefully been corrected. The naming convention of these buoys is rooted in the larger context of the AWI buoy programme and the MOSAiC expedition, in which all the different buoy types are identified by the deployment year, a designated buoy-type dependent letter, and a running number. Actually, we should have included this description in the Methods part in the first place, which we have done now: "Following the naming convention of the general AWI buoy programme, which was also implemented for all MOSAiC buoys, we identify the individual buoys by an ID consisting of the deployment year (in this case "2019"), followed by a buoy type-specific letter (in this case "O") and a running number (here, "1" to "8"). This resulted in buoy IDs 2019O1 to 2019O8."

 

Paragraph starting at Line 135: I got a bit confused here. I thought perhaps the CTDs recorded 2-minute data, and then every 10 minutes, all of these data were sent to the satellite. I think instead, the satellite gets only a subsample of the 2-minute data, i.e., 10-minute sampling, is this correct? I wonder if you can make this clearer.

#This is partially correct. The buoy doesn't transmit a subset, it actually takes an additional measurement. We rewrote the paragraph substantially to make it clearer.

Lines 153-158: This material confused me. The CTDs have pressure accuracy of 0.02 - 0.002 dbar, but the transmission limitation degraded this to 0.1 dbar. So this is a significant reduction in data quality, right? Your text seems to indicate that it’s no big deal.

# An accuracy of 0.1 \unit{dbar} implies a potential error of max. 0.09 \unit{dbar}, given the deprecated pressure values. That kind of change in the top 100 \unit{m} effects negligible changes in practical salinity and potential temperature (likewise, conservative temperature and absolute salinity). The error in salinity and potential temperature is always 10^-4 or less (using the corresponding standard units). The maximum error in depth is 0.09 \unit{m}, which, again, is negligible, given that standard CTD profiles by highly accurate systems are usually averaged at 1 \unit{m} or 1 \unit{dbar} intervals. The basis of these calculation is the Gibbs Seawater Toolbox, based on TEOS10.

While the deprecation of pressure to one decimal place is unfortunate and was not known to us when ordering the units, the problem will be alleviated by the manufacturer in any future production of those units.

Further, Table 2 indicates that the pressure accuracy is always 0.1 dbar, even for the CTD data. Why? I thought a direct download could give you 0.02-0.002 dbar. I guess I’m missing something.

#Well, you are correct. The 0.1 is the most conservative accuracy, but yes, not representative for most of the dataset. We changed this to 0.1 / 0.02*, with an additional comment in the table caption. We would prefer not to give any higher accuracy, because the availability of pressure sensors was limited at the time when the CTDs were built, so for example 1000m rated pressure sensors ended up in some instruments.

Line 191: “an upper threshold of 0.8 m s-1 was applied” Why?

#Actually, no single drift speed data point was removed based on that threshold. So this statement was removed to not confuse the reader.  

Line 220: what is “sim”? I suspect it is some kind of word processing code.

#Great point, it should have been \sim, which is the LaTeX code for ~. (Fun fact: the LaTeX code in our MS was "~sim.) This is corrected now.

 

Line 222: How do you define “suspicious?”

Line 227: How do you define “questionable?”

#Beyond the defined range checks and moving average filters, a manual spike flagging was performed that was not based on a strict and stiff definition, but rather on a subjective (expert) judgement to the best of our knowledge. Examples are obvious isolated spikes (that still didn't fail the other tests, and therefore can't be considered "obvious" outliers), or suspicious "episodes" that for example weren't accompanied by a change in any other parameter.  

In that context, there are other flagging schemes that even distinguish between suspicious and questionable (which we don't), or that categorize into "probably good" and "probably bad" etc. We don't feel comfortable in using that level of differentiation, and therefore decided to flag everything that "looks odd" or inconsistent with the -questionable- flag. 

 

Line 233: “When CTD data was available…” This caused me some confusion, because the buoy has CTDs. Perhaps you can add a sentence like this to make it clearer: “Buoy data" means all data uploaded to the satellite from the buoys, including from the CTDs, at 10 min sampling, while "CTD data" refers to the directly downloaded CTD data from recovered buoys at 2 min sampling. Is that right?

#This is correct. This part has been altered to be much clearer, also based on the feedback by reviewer 3 who had the same issue.

 

Table 2: Is the surface temperature thermistor really accurate to 0.01 degC? I think these are usually not so good, e.g., 0.1 or 0.05 at best.

#This is correct, this should have read "0.05", and has been changed accordingly. This accuracy is also only valid when in water apparently.

Line 395: The MLD is not evident to me from this figure. How was it computed?

#see comment below on MLD in Figure 4.

 

Line 397: Perhaps change “a corresponding decrease in temperature” -> “a corresponding decrease in temperature along the freezing line”

#Good suggestion, this has been changed accordingly.

Lines 399-400: Perhaps change this text to: "...partly explained by a combination of upward mixing of deep salty water from below and salt rejection during ice formation from above, both forced by two February storms (with wind speeds up to 16 ms-1 and air temperatures of XX degC)."

#Good suggestion, this was changed accordingly. 

Lines 400-401: why is “vertical mixing properties” in quotation marks?

#We're not sure why. This has been changed now.

 

Line 403: I can’t see MLD in Figure 4.

#This is correct, a continuous MLD evolution naturally can't be derived from our fixed-depths dataset. However, when CTD records from different depths "join each other", one can infer that the MLD has deepened at that time (i.e. the upper 100m have the same physical properties). This has been modified to become clearer.

 

Line 404: What is “air-line distance?”

#This was changed to "straight line distance", which is hopefully more clear.

 

Line 413: “The position of the eddy remains approximately the same” The same as what?

#This sentence has been removed.

 

Lines 417 & 419: check your velocity units. m-1 is wrong, yes?

#This has been corrected.

 

Line 417: “estimated diameter of the eddy” estimated how?

#The estimate was based on the buoy drift distance.

 

Line 421: Here you are using zero “0” instead of capital “O.” There are several other examples in this eddy section.

#This has been fixed now.

 

Line 426: “by a factor of” i.e., add “a”

#Done.

 

Line 428: “radii” -> “radius”

#Done.

 

Lines 431-432: Great point!

# :)

 

Line 439: “measured the the upper ocean quasi-synoptically over submesoscales” Two comments: (1) cut one “the” (2) Your introduction has a lot of material about submesoscales, but the paper itself has no corresponding analysis. I suggest you either cut / trim down the introduction material on this subject, or do more work to follow up on it. For example, you could create a histogram for one or all the buoys with a horizontal axis of del-x = the distance travelled by a buoy over 2 or 10 minutes. If you have a lot of sampling between 100 m – 10 km (or maybe shorter? Your introduction implies that Arctic submesoscales are shorter) then yes, you can justify this statement on Line 439 and the extensive discussion of this subject in the Introduction.

#Thank you for this comment, it makes a lot of sense. As mentioned further up, we have removed the mismatch between the intro and the analysis by moving a lot of the submesoscales material to the discussion part, while also alongside this including a new Figure 8 that is in principle similar to the histograms the reviewer suggested. We hope that the match is much better now.

 

But I would also ask this question: Did you really need to sample at 2 or even 10 minutes? What if you sampled hourly, or even daily? What would this histogram look like? I suppose I am suggesting that you do some analysis to find the right time interval for buoy data recording to adequately sample the submesoscale, as you claim without proof that you are doing now.

#The choice of 2 and 10 minute sampling was based on a power budget calculation of the CTDs and buoys, and optimized towards the ideal duration of the MOSAiC experiment, which was one year.

Furthermore, resolving various processes, such as internal wave variability, requires sampling intervals of a few minutes. We have added an exemplary figure and a discussion on what time and space scales are resolved in our dataset in Section 5.3.

 

Line 473: code is available “upon request.” Is this sufficient for this journal? Should it instead be available on github or equivalent?

#Since the processing code doesn't include any particularly innovative techniques and is only tuned towards this particular data set, we decided that there is not enough added value to putting the project on Github or similar repositories. If this is a requirement for the journal though, we might reconsider this.

This is a solid contribution to the field, it does have some issue with grammar and word choice throughout the text, however, it did not hinder my ability to read or understand the content, maybe the authors could run it through a grammar checking program (I use Grammarly). 

#We thank the reviewer for the comment and advice. We now used Grammarly to correct the grammar. We also tried to improve the wording in some instances, although it is difficult to judge what words exactly the reviewer is referring to. We hope that the overall reading experience is better now.

From the title of the manuscript I expected the focus to be primarily on analysis of the data, however the focus of the paper is really on describing the buoy system and the handling of the data. I think that this does fit within the scope of the journal, but the authors may want to change the title to better fit with the focus of the paper. I have no issues with the substance of the paper, just a few comments to improve clarity.

#We appreciate the comment and concern about the title "Mesoscale observations of temperature and salinity in the Arctic Transpolar Drift: a high-resolution dataset from the MOSAiC Distributed Network". It is usually quite difficult to represent an entire manuscript by only a few words. However, we are not entirely sure why the title would suggest a focus on a scientific analysis, as words like "observations" and "dataset" should indicate that the focus is on the description of the data. We could probably have written a more technical paper and submit to a more tech-focused journal, but we wanted to put the focus on the dataset itself due to its uniqueness, but with the technical part still properly represented. In short, we do believe that the title is appropriate in its current form, but we are absolutely open and willing to change it if the editor thinks this is necessary.

The description of the data handling and QC is well thought out and helpful for others who are starting out with automated buoy systems.

#We appreciate the positive feedback.

Line 135. I was a bit confused by the statement “polled for an additional measurement by the buoy itself” I think that you mean that the CTDs recorded data every 2 mins and then they were also collected a measurement every 10 mins that was sent via iridium. This paragraph is not very clear.

#The other reviewers also criticised that this part was not very clear. The paragraph has been completely rewritten as follows: 

"In order to ensure an operational time of ~1 year, the individual CTDs were set to record data internally at 2 minute intervals, independent of the buoys' own sample and transmit intervals. The surface buoy itself recorded GPS position, surface temperature, and carried a submergence sensor. Furthermore, the buoy controller polled all CTDs for an additional measurement independent of the CTDs' internal sampling according to a pre-configured buoy sampling interval, which could in principle be adjusted by sending a reconfiguration command via iridium if necessary. However, throughout our experiment, all buoys were set to take a measurement and immediately transmit the corresponding data at a fixed 10 minute interval, chosen to ensure an operational time of at least one year. Thus, the internal sampling interval of the CTDs was 2 minutes, while an additional CTD measurement was obtained and transmitted by the buoy every 10 minutes together with the corresponding auxiliary (meta)data."

Line 404: ‘air-line distance’ do you mean the the straight line distance rather than the distance along the drift track.

#This is correct. We changed it to "straight line distance".

Line 404: Is there a better way to visualize the MLD in Figure 7?

#Actually, the evolution of the MLD can't be continuously determined from the present fixed-depth observations, and thus can't be visualized in Figure 7. See more detailed comments to reviewer 1.

The authors present a dataset of upper ocean temperature and salinity from drifting buoys that were deployed as part of the MOSAiC Distributed Network. The dataset represents valuable observations from the central Arctic Ocean over a 10-month period, including the scarcely observed Arctic winter. Observational methods, data processing steps and the resulting dataset(s) are described in detail, and some preliminary analyses of the data to investigate the signature of an eddy in the distributed network are shown.

The manuscript is in general well written with mostly clear and useful figures. Some revisions could help to improve readability and avoid confusion in an otherwise highly relevant publication.

#We thank the reviewer for the positive feedback. We incorporated the comments into our revised manuscript, hoping to remove the confusion and improve the overall quality.  

Main concerns:

There is a bit of repetition between the Methods and the Results section, esp. with 4.1. This is maybe not surprising given the nature of a data paper, but could still be avoided, and the manuscript could be more streamlined.

#We agree with this assessment, but this is somewhat intentional. This dataset as a whole is a little complicated, and there is great potential that a reader might get confused easily if only parts of the manuscript are read. We intentionally included some repetitions where appropriate to make individual sections (such as 4.1) better understandable on their own, when the other sections are not read in detail. There is absolutely a lot of room to cut information out, but we would  prefer not to do so, even if it makes the overall manuscript longer. We hope for your understanding.

    

The introduction is quite lengthy about mesoscale and submesoscale features, but then there is nothing more these throughout the methods and the results section. First in the discussion, you pick up the topic again by describing the passage of an eddy. To show that the dataset is useful to investigate (sub)mesoscale features, which you explicitly present as an aim of the design of the DN and this manuscript, a bit more analysis in this direction would be useful. For example, more detail on the drift trajectories of the buoys relative to each other beyond just the map in Figure 3 (and the short sentence on Lines 267-269) could easily provided, such as a timeseries of the relative distance of the buoys to each other or of the maximum diameter of the area covered by the buoys. That would link nicely to the discussion around the size of the eddy and could indeed demonstrate that this buoy network can capture (sub)mesoscale features.

#As also explained in our response to reviewer 1, we tried to improve the previous mismatch between the intro and the analysis by moving some of the submesoscale material to the discussion, alongside a new Figure 8 which shows TS diagrams for different measurement intervals (which is a similar concept to the distance analysis suggested by the reviewer). As there are currently 3 more manuscripts in preparation that use the present dataset to look at (sub)mesoscale features, we would prefer not to give away to many details. We therefore hope that this additional analysis is sufficient to make the point.

    

Be more precise with the variable names – I have to assume that you talk about in situ temperature and practical salinity but cannot know for sure, so please make this clear at the start of the manuscript.

#This is correct. There are indications about this throughout the manuscript, but now we state this explicitly at the beginning of the data processing section.

  

Many of the figures are not colour-blind (or greyscale friendly) – consider modifying them.

#Thanks for the suggestion, this will be fixed in the final version.

 

Other comments:

Line 21: “much greater than” – can you be more precise? E.g. give order of magnitude or similar?

#Changed text to indicate that timescales are up to the order of months.

Line 21&22: “the vertical velocity” and “the horizontal velocity” – sounds odd (like there is only one fixed vertical velocity and one horizontal velocity); suggest to change to “vertical velocities are … weaker than horizontal velocities”.

#Done.

Line 41: “submesoscale processes are responsible for … restratification” – suggest to change to “contribute to”. Du Plessis et al. (2019) still state that the main driver for springtime restratification is surface heating.

#We have changed the text to "contribute to". There are many processes driving mixing and restratifying the water column, dependent on season, region etc. Du Plessis et al. (2019) name submesoscale processes as strongly influencing seasonal restratification processes, e.g. their abstract states: "An increase of stratification from winter to summer occurs due to a seasonal warming of the mixed layer. However, we observe transient decreases in stratification lasting from days to weeks, which can arrest the seasonal restratification by up to two months after surface heat flux becomes positive. This leads to interannual differences in the timing of seasonal restratification by up to 36 days." The point of the citation in our manuscript is to put forward the importance of submesoscale processes for the upper ocean, not to give an accurate account of seasonality in the southern oceans. In the Arctic, horizontal density gradients can lead to submesoscale adjustment in the mixed layer, effecting restratification in the formerly well-mixed upper layer (e.g. Timmermans et al., 2012).

Line 66: Is “synopticity” really a word? (Or rather, a word in this context?) Suggest to rephrase.

#This has been changed and now refers to the design of hydrographic surveys.

Line 84: Which feedback? Sounds like there is only one…

#This has been changed to reflect that there are manifold feedbacks, without going into further details though.

Line 93: Do you mean upstream IN or OF the Transpolar Drift?

#We added an "in".

Line 101: Same issue: upstream in or of the TPD?

#We added an "in" as well.

Line 109-110: “the ice relative drift across the liquid ocean” – weird formulation. Rephrase?

#This has been rephrased to just "ice drift"

Line 117: “ensued” – is this the right word? Or should it rather be something like “took place” or “were done”? I assume there was a degree of planning involved in the location of the DN.

#The intro has been changed extensively, and the wording mentioned here was changed in the process.

 

Line 135-138: This part is confusing. So the CTDs are measuring every 2 minutes regardless of what the buoy is doing. Then there’s an extra measurement when the buoy is polling – so “its measurement interval” means the buoy’s measurement interval? And the data the buoy sends back via iridium – that’s only those extra measurements or all of them?

#This pararaph has been strongly modified, as reviewer 1 had the same problems understanding it. We hope it is much clearer now. You are generally right, the buoy wakes up every 10 minutes and lets the CTDs take a separate measurement, regardless of the CTD internal recording interval. Only this one is transmitted.

Following on from that, on line 142 you write “All data” – is that now the data from the measurements polled by the buoy or indeed “all” data??

#"All data" in that case refers to the CTD readings as a result of the "extra" measurement, along with GPS, surface temp and submergence. The buoy does not have any means to access the internal storage of the CTDs.

Please rework this part to clarify.

#The part was reworked.

 

Line 145 and throughout: when you write “temperature”, is that in situ temperature? Please clarify in the text.

#we added a general sentence at the beginning of the data processing section: "In this paper, we are exclusively referring to in situ temperature (T in $^\circ$C) and practical salinity (S in PSU), if not stated otherwise." We hope this makes it clearer.

 

Line 147: “the other variables” – I assume that’s conductivity, (in situ?) temperature and pressure?

#Yes, this has been made clearer now.

Line 147 (and throughout the manuscript) provide version numbers of software and toolboxes used.

#Version numbers have been added where relevant.

Line 169: Protection from what? Towels don’t strike me as the best protector from freezing in -15deg C.

#Haha, you might be right, I guess towels are not the best protection against cold air temperatures. There were significant constraints on logistics and everything was quite hectic, which led to us not having any good means to keep the instruments heated pe-deployment. It might be embarrassing, but in the end we had to use towels to at least provide some sort of protection while handling the CTDs on the ice and in the snow. We deployed the tether manually with 2 persons without the help of a tripod, so the instruments had to be placed on the ice/in the snow for up to an hour while we attached them to the cable. This also means that there were no 5 persons to handle every instrument while lowering the tether in the hole, so we actually had to drag them over the ice. To make a long story short, we did only have towels, which is better than nothing, and next time it should probably be done differently.

Line 171: What do you mean by “hydrohole”?

#This is the term for an access hole through the ice into the ocean, and is regularly used in the literature. We changed this now to "hole in the ice" though.

Line 179: The CTD data from recovered instruments then also provide higher accuracy pressure data?

#This is correct. As described further down in the data processing section, we even used the more accurate recovered CTD data to increase the accuracy of the corresponding transmitted buoy data.

Line 186: Replace “using” with “indicated by” or something similar (you didn’t “use” the flag to modify the data).

#Changed as suggested.

Line 190: How did you fill the gaps in the GPS record after removing outliers?

#The gaps were filled by linear interpolation. This has been added. (It was just a handful of outliers.)

Line 191: What was the consequence of applying the threshold? A different flag? Or removing data points?

#The data points were removed. This has been added.

Line 192: How did you determine “plausibility”?

#We just did a general range check and looked at the seasonal evolution. We did not compare or validate against other meteorological data, since this parameter is not in the focus.

Line 220: What do you mean by “sim”?

#This should have been \sim, which is the LaTeX code for ~. It is fixed now.

Line 223: How big was the drop in conductivity?

#The drop in C was from ~34.1 to ~33,5 PSU, and not accompanied by a temperature change. This has been added.

Line 229: Capital O in the buoy name.

#Done

Line 240: Add “practical” to salinity.

#Done

Line 243: Is the closing bracket in the right place? Bit confusing at the moment.

#The comma was in the wrong place, and it should be clearer now.

Line 253-254: It would be useful to include this information in one of the tables, e.g. Table 1. If you explicitly state here all buoy numbers of the ones you can provide a merged product from, then also provide the numbers of the ones with buoy data only.

#Good point. This information has been added to Table 3, where it is most relevant. Also, we included a detailed overview to the part this comment refers to.

Line 255: I guess the measured temperature is in situ temperature? And the derived salinity practical salinity (and not absolute salinity)?

#This is correct, and has been added here. We also added a general sentence as suggested in another comment.

Figure 2: Nice schematic! You could mark which of the steps were not included in the processing for buoy-only datasets.

#Actually this is the purpose of the steps that have a similar grey background as the SBE37 field at the top. Those have not been applied to the buoy-only datasets. 

Table 2: In situ temperature?

#Yes, this has been added.

Figure 5d: axis labels in the salinity panel are missing. If they are not included because of the icing issues, why show the data at all?

#Thanks for catching that, the labels should be there. The graph shows the reasonable data. This will be corrected in the next version.

Presentation of the timeseries starting Line 274: Refer to the respective figures/panels.

#References have been added to the time series description.

Line 276: delete “a”

#Done

Line 276-277: Isn’t 2.5 weeks a pretty long time for ice in the conductivity cell to disappear? Was the water at freezing point at 50 m depth throughout this period?

#We tend to agree, but we don't have any other explanation for that effect. It's not that there was an ambiguous adjustment, the salinity was completely off track. We included a sentence that this is rather unexpected, as you suggested.

Line 286: So all but the 50 m sensor?

#Well, the 50m sensor didn't give any data, so we can't refer to any data explicitly. But since all other 4 sensors showed the same consistent behaviour, we conclude that the events described must have been affected the tether as a whole. We considered to add a sentence along those lines to the manuscript, but instead changed "all CTDs" to "all remaining CTDs", which we consider sufficient.

Line 318-319: Is a drop in pressure not equivalent to instruments rising? “… causing the surface buoy to drop” => shouldn’t that then lead to an increase in pressure? I can’t follow your argument here.

#Thank you for catching this. Of course we meant that the pressure increased. his has been corrected.

Line 323-324: How long did this issue last? The entire record?

#Yes, this issue lasted the until the end of the operational period, which for this buy was anyways only two weeks. This information has been added.

Line 329: Replace “in” by “the”.

#Done

Line 338: Where below? Provide section/section number.

#Reference to Section with description of cross-validation has been added.

Line 374: What is the stated accuracy?

#The stated accuracies have been added.

Figure 7: I’m confused by the shading in panels c and d – please provide a clearer description in the caption.

#This has been clarified.

Line 472: Publish the code together with the dataset or place it on e.g. github or similar sites.

#Reviewer 1 had the same comment. We did consider it, but since the processing code doesn't include any particularly innovative techniques and is only tuned towards this particular data set, we decided that there is not enough added value to putting the project on Github or similar repositories. If this is a requirement for the journal though, we will of course reconsider this.

Figure A1 is not referred to in the text.

#This figure has now been referred to in the Results section.