Review of "A fjord dataset for multidisciplinary applications: Thirteen years of ocean observations in Sermilik Fjord, Southeast Greenland”

By Aurora Roth et al.

The paper is well written and should become a useful reference for future use of this data set. There is many years of dedicated effort to collect this important dataset, and the presented summer climatology is well organized. The figures are also clear and show many examples of how the observations have been synthesized to show variability. While many of the presented results were ‘as expected’ for a Southeast Greenland fjord, it is good that this is published and that the 2009 – 2023 observations make sense compared to earlier work. The suggested improvements are small and few enough to tick off the ‘minor’ revisions here.

General comment: The minor revisions needed are some unclear formulations and there were also a few errors around referencing figures. The one larger issue I have is that although an impressive number of summer campaigns is presented, it is still ‘summer only’. I strongly suggest that the data is described as summer data. In line 270 for example, you state; “all profiles collected in the fjord across all years”. This makes it sound like you have representative data for “all years” - - you don’t. You have good representation for summers. This should be corrected throughout.

Specific comments:

Line 78,79,80: This text is repeated and should be deleted. It is also unclear what 'annual gridded dataset' is compared to the summertime means.

Lilne 100: Here you need to add “SA”; Absolute Salinity (SA) 33.3 g kg-1.

Line 157: Missing figure reference; “except for 2014 and 2020 (??).”

Line 159: I don’t know what you try to indicate with the (1)… Please fix.

 Line 161: Wrong format for the deg C.

 Line 179: Probably the Greenlandic name for Midgaard Fjord is missing?

Line 180: I don’t know what you try to indicate with the (2)… Please fix.

Line 181: Wrong format for the deg C.

Line 220: I don’t know what you try to indicate with the (3)… Please fix.

Line 270 I suggest to just write; “all profiles collected in the fjord across all summers”

Line 346: And here then “each summer” instead of “each year”.

Line 350: Again here you write “all the yearly data”. I think that you did not include the March XCTD profiles? In any regard – please specify if this is summer data (July – September). 

Caption Figure 4: “The 2023 CTD and XCTD data are not combined and instead treated as two separate instances of a summer fjord state leading to a total of 14 possible grids.” This is clearly written also in the text, you only need to explain this once. I suggest to delete this here.

Line 392: Here you use the correct term. You have produced “a summer state climatology”.

Line 393: Missing figure refence.

Line 418: I think you are missing something here; “The subsurface temperature minimum of this profile indicates the core of the PW layer present on the shelf.” Surely – this must be the remnant of the lower part of the PW from the previous winter situation? Please clarify.

Figure 9 caption has missing Figure (??) reference.

Line 449: The stratification (density) is not visible in Fig. 10. Do you mean thermocline??

Line 463: Missing figure refence (??).

Line 475: Figure 8 c) shows the mean salinity – not the variability. And the variability in the mouth region is only large for Temperature. Please rephrase.

Line 528: Here there is a very welcome comment on the large seasonal variability in freshwater forcing. I think you should also mention the large (trivial) solar seasonal variability that drives it. This must explain the maximum surface variability for temperature shown in Figure 8 b).

Otherwise it was a nice paper. The data collection is impressive and the new climatology will be very useful.

Summary

The authors have compiled several sources of temperature and salinity data from 2009-2023, from both CTDs and SCTDs, into what they consider to be a gridded, optimally interpolated dataset. They accounted for the differences inherent in measurements taken with different sampling tools, techniques, and missions. All of which are well accounted for in the included tables (Table 1 + 2). These collected and QCed data were then used to create along-fjord gradients, as well as summer climatologies and their root mean squares of deviation (RMSD). These data have been made available in NetCDF format, with one file for the full transect data, and individual files for the yearly collected data.

The authors claim that this dataset should provide researchers with easier to use, and more reliable data for investigations around the mean state of the fjord, it’s gradient towards the mouth, and regions of variability therein. The authors provide a very thorough description of the optimal interpolation process, and investigation of the results. I enjoyed reading the history of the sampling effort given by the authors in Section 3.4. The text is very well written with a consistent and engaging voice used throughout.

However, it does not appear that the final compiled manuscript was checked as there are abundant spacing and broken reference issues in all sections. Generally, there are also quite a lot of spacing issues between symbols, numbers, and letters that the authors should triple check. Perhaps these were caused by LaTeX formatting issues? I also noted several formatting issues that appear to be notes between the authors to remember to insert certain pieces of information. These will need to be smoothed out before publication. A spelling question throughout, should ‘Gletsjer’ be ‘Glacier’? Also, double-check that ‘ice mélange’ always has the accent. I’m not certain what the ESSD editorial standard is, but the authors should ensure whether a space is used for °C or not. E.g. 10°C or 10 °C. Yes, the ISO standard is 10 °C, but often in written manuscripts it is given as 10°C.

Ultimately, I would argue that thes dataset presented here id not gridded data in the traditional sense of the word. Meaning, the different grid cells of data do not have evenly spaced lon/lat coordinates in any coordinate system (e.g. any satellite or model output spatial data product). I do not think that using even integers (e.g. distance from fjord mouth in km) in place of spatial coordinates can be used to call a dataset gridded. Nor are these data available at any spatial extent, they are a linear string of data following a proscribed line, with the lon/lat coordinates attributed afterwards. I think that most readers and users, upon seeing the word ‘gridded’ used repeatedly throughout this manuscript would assume that the data product created here would have even pixels across the body of the fjord, with the corresponding depth values per pixel along an evenly gridded z-axis. I agree that this z-axis has been created as expected, but not the surface values. I recommend that the authors re-evaluate their use of the word ‘gridded’ throughout the manuscript, and rather more strongly emphasize that these are transect data, and not gridded data.

That being said, I do not have any substantive comments or suggestions before recommending this manuscript for publication. Though do see my specific comments below, particularly those on the code and data availability section. As well as my thoughts on the discussion and conclusion sections.

Title

- I don’t know that the word ‘fjord’ needs to be used twice. Could just say: ‘A dataset for…’.

Abstract

- 14: ‘fjord intercomparison studies’ the issue with this statement is that, if this methodology has only been applied to this fjord for this dataset format, there are no other comparably formatted datasets with which one could perform fjord intercomparison studies. One may want to add a clause that datasets like the one produced by the authors would allow for effective fjord intercomparison studies, if more datasets like this one were created for other fjords. i.e. This methodology could be taken as a blue print for creating gridded temperature + salinity datasets for other fjords.

Introduction

- 81: This is a big statement. One should potentially add a sentence linking to the code or repository were this data node is beginning to develop. And/or a sentence explaining how the long-term continued support for such an endeavour has been assured.

Study Site

-106: Section 2.3: It may be interesting to add some of the physical features described here (e.g. the sill) to Figure 1. As well as referencing Figure 1 somewhere in this section.

- 115: ‘(GMW).In’ → ‘(GMW). In’

- 120: Section 2.4: Why not reference Figure 1 when describing the ice mélange reason? The authors have already gone through the trouble of labelling it on the map.

Data

- 157: Why the ‘(??)’? Referencing issue, or a note to explain the two missing years?

- 159: What does ‘(1)’ signify?

- 161: Throughout this and the following sub-sections, please double check the spacing between symbols, numbers, and letters. The degree symbol is incorrect.

- 179: ‘GREENLANDIC NAME’ ?

-180: ‘(2)’?

- 184: Does it not cause issues to average the CTDs to 1 m depth bins, but the XCTDs to 2 m depth bins?

- 220: ‘(3)’? I take it that these are meant to be references to the Figures? How are references to tables made? ‘Figure 1’ is referenced as such in Section 4.

Methods

- 245-251: It seems sub-optimal to me that the grids would not be consistent for both temperature and salinity, nor the same for every year. If indeed one of the objectives of the creation of this dataset is to make it useful to a wide range of disciplines, the grids should be static (i.e. the same for each variable and year). Or, it seems that the use of the word ‘grid’ in this text is not what I would expect it to be, and I’ve misunderstood the description of the methodology. Rather, the authors use the word ‘grid’ to refer to the mathematical equation used to calculate the average values per grid cell? With a grid cell being the used in the traditional sense, a la gridded data found within NetCDF files. Again though after reading the full manuscript I see that this is not the case either. I have put my thoughts on this topic in the summary above.

- 289: ‘(Eq. 4 and Eq. 4).’ -> ‘(Eq. 4 and Eq. 5).’

- 343: ‘The 2023…’ This seems like it would cause issues for users that simply want to extract annual climatology values. Ah, after accessing the .nc data files I see that the authors have chosen to give each year as an individual NetCDF file. Thereby avoiding the issue of the two different gridded datasets for 2023. This is a sub-optimal way to provide data for non-technical users as they are not necessarily going to have the competence to access and extract data well from multiple disparate files. I advise that the authors combine all annual data into one NetCDF file. Keeping the 2023XCTD data as a supplementary file.

Results

- 367: Spacing issues throughout the paragraph.

- 379: ‘mapping relative’ -> ‘mapping of relative’

- 379: ‘The user...’ This is a nice idea, but outside of physical oceanography, I don’t think many research teams are honestly going to get into this level of detail. I make this point considering that the authors hope for this dataset to serve as the locus of development for an established community resource. The more simple a product is to use, the more likely people are to use it. The more people use it, the more likely it will be funded in the future. Too much complexity may be counter-productive.

- 389: ‘Supplementary Material’ -> ‘Supplementary Material (Figures S1-S8)’

- 393: Broken figure reference.

- 463: Broken reference.

Discussion

- I find the discussion a bit odd in that, so much of the data were analysed and reported on in the results, but these are not discussed in this section. Rather the authors decided to give three considerations on the use of these data. While good to include, would it not make sense to include a sub-section discussing any of the interesting results that were made possible due to the data interpolation method used here?

Conclusions

- 571: ‘interdisciplinary disciplinary’ → ‘interdisciplinary’

- 572: ‘shared’ → ‘created’

- I find that this section also does not match with the body of the results. The authors have put quite a lot of efforts into fastidiously documenting the resulting data of their applied interpolation method, but then hardly discuss or conclude anything from this. Rather they choose to focus on the importance of FAIR data to the Arctic research community. I completely agree that this is important, therefore, it seems that throughout the results section (or discussion) the results that describe the data (e.g. mean inner-fjord deep temperature) should be provided with some context for how this is relevant to FAIR data in the Arctic.

Code and data availability

- Link given for datasets work as expected. Data can be downloaded and are in a standard NetCDF format.

- The code used for these analyses is also publicly available. In the same location as the data themselves, which is ideal.

- Upon loading the transect (gridded) data, I do not have the same smooth visual summer climatology shown in the results of the manuscript. Regardless of which x and y axis variables are used.

- If the authors are serious about this methodology being able to be easily applied by other teams to create similar datasets, it would be ideal to translate it into Python and R code. I am not recommending this be done for this manuscript. Rather something to add to a long-term TO DO list.

Table 1

Very nice. However, this table is not referred to in the text.

Table 2

Also nice, but also not referred to in the text.

Table 3

No comments.

Figure 1

If the authors want to show depth contours, it would be better to use an illustrated figure (e.g. Figure 2), rather than a satellite image. By showing the water as dark blue (true colour), this figure gives the impression that the majority of the fjord is deeper than 800 metres. While the ice mélange regions appear to be above sea level. However, I agree that it is useful to illustrate where the ice mélange region is, so I would rather just remove the depth contours as they are shown in Figure 2, and showing the depth is not the primary purpose of this figure.

Figure 2

Why is this figure not referred to in the text? One should show the x- and y-axis labels only once for the entire figure. Not once for every facet. This would allow for the labels to be a larger, more legible, font size. I like the authors choice to but the legend in the top left, and to start the first facet in the top middle of the figure.

Figure 3

A very useful figure. A good example of one of the few cases when it is correct to use a stacked barplot.

Figure 4

Well presented and clear. This figure helped me to realise I was misunderstanding what the authors meant when using the word ‘grid’ vs ‘grid cell’.

Figure 5

The residuals (b, e), shown as dot plots, are not easy to read. Though I don’t have a better suggestion. Upon zooming into the PDF I was able to make out the values relatively well. The choice of a green-purple colour scale is a good one. Otherwise a very well assembled figure. The authors may want to consider increasing the size of all numbers, labels, and text. Though I leave this to the editorial process. As with Figure 1, it could save space and text on the figure to give the x- and y-axis labels only once as they are always the same.

Figure 6

Another nice figure. I don’t think it’s necessary to state what the bin size is along the x-axis. An interested reader could determine that for themselves.

Figure 7

There is a lot going on in this figure. The caption does a good job of describing how to interpret the information, and it is interesting to look at the difference between the original profiles and the gridded data.

Figure 8

Another consistently good figure.

Figure 9

I like the choice the authors took to inset the horizontal profile panel in the upper left corner. There is a broken figure reference in the caption.

Figure 10

Considering that a light colour palette is being used, it may be better to make the lines slightly thicker. Or highlight the edges of the lines in black. Though that can lead to other visual artefacts.

Figure 11

Same comment as Figure 10. ‘larger the’ ‘larger than the’.

Figure S1-S8

The same comments for their corresponding versions in the main manuscript. I think it was a good choice by the authors to have included all of the profiles in the supplement like this.