Reviewer Comments for Rowell et al 2023

Pete D. Akers

Overview:

The authors here present their chronology development for a borehole from Sherman Island in West Antarctica. They use multiple methods to determine their chronology, largely based on the constraints of their sampling and changes in the ice chemistry resolution with depth. They find their borehole covers back ~1200 years and propose that ice near the bedrock could be early Holocene.

Overall, I find that their methodology is sound (although I will admit that I do not have personal experience with flow or thinning modelling). It generally accomplishes its goal of presenting an age-depth scale for the location, but broader discussion is rather limited outside of extrapolating what the bedrock age of the ice might be. The writing is generally good and easy to follow, with only a few minor technical/grammar issues highlighted in the technical comments.

As a stand-alone manuscript, this comes across lacking in parts at times. This manuscript is clearly complementary to Rowell et al, 2022, but readers unfamiliar with that paper will feel rather lost reading the present manuscript. I personally didn’t understand many of the mechanics and specifics alluded to in this manuscript until I had to go read Rowell 2022. While it is of course fine to point out that the details on certain methods and analyses are in Rowell 2022, this manuscript in review needs to be able to stand alone enough that reader can understand all the basic results and discussion in the manuscript.

Following on that, the overall discussion of the final chronology and work feels a bit underwhelming and undeveloped. I don’t think that a massive expansion of discussion is required (or warranted), but it would be nice to see some broader impacts and discussion about the chronology results. Is the final accumulation rate different from what is observed elsewhere or expected from models? How does the quality/resolution compare to the SI Core or other core-based results from similar sites? Are there any broad lessons learned about in how this method could be applied or where it should be applied elsewhere? Note that these aren’t questions I’m requiring the authors to answer; they are simply pointing out some ways that a deeper discussion could make the manuscript more impactful than just largely a technical report specific only to this Sherman Island site.

Altogether, the manuscript is generally a solid report on the results of a chronology production. Should the editor(s) desire more than this, I think the paper could be made more impactful by developing the discussion more (both what exists on the bedrock age modelling and adding some broader comparative context and/or applicable lessons).

Major Comments:

Introduction: The first paragraph sets the scene well, but I struggled to understand exactly the point and details of the project in the next two paragraphs. For example, is the RAID system different from other drilling systems? Why was Sherman island chosen? It is serving as a constraint of what? Moreover, what is the actual point and objective of THIS paper? There doesn’t need to be a ton of detail on this stage, but more clarity and structure to prepare the reader for what they are about to read.

35: The use and reasoning of the RAID aren’t clear here. Since it isn’t stated what the RAID is, or how it differs from normal drilling, I don’t know why it would be chosen in light of the risk of Sherman Island. (Reviewer note: Later I see that Rowell 2022 has this information, but the basics need to be summarized here in this manuscript also).

35: Why was Sherman Island chosen, then, if it had such high risk of not contributing to the goals of the WACSWAIN project? Did it have other virtues that warranted the risk, or was is simply logistically easy?

Drilling and measurements: Some more information is needed here. Although papers with detailed descriptions are cited, this manuscript should include the minimum details required to understand the rest of the paper without having to look up a second paper. Namely, what sort of samples is the RAID bringing up? If they are chips, are they in stratigraphic order? What is modern environment at the drill site (accumulation, etc)? What is the vertical resolution of samples if they are mixed? These are all critical to understanding the rest of the paper, and a reader shouldn’t have to go find a second paper to get this information.

Figure 1. A close up map of the Sherman Island region and drill site would be beneficial to understanding the local geography and ice structure/flow.

105: Why was EPICA Dome C chosen as the assumed proportional accumulation as opposed to, say, WAIS? Is there any evidence to think that Dome C and SI would be proportional? They are very different climate systems.

110: This sentences reads awkwardly. Better fitting than the markers? Better fitting judged in what way? “It became clear” gives us no insight into what decisions were made, statistics performed, or observations made.

113: Again, if it is clear, you don’t need to state it to the reader. You should present what made it clear.

114: Annual resolution isn’t necessarily required to detect volcanic events. They are found at Dome C, for example, and the signal there is blurred over annual levels. In fact this raises an interesting question: since the sulfate fallout can extend over several months, did any of the volcanic peaks interfere with your annual dating by SO4 peaks?

140: Note that Kuwae is still disputed as a source for the 1458 eruption (https://doi.org/10.1038%2Fs41598-019-50939-x). Should become more clear with some research over the next decade, but you could add a “commonly attributed to Kuwae” or “eruption in 1458 (possibly Kuwae)” style to be safe.

Figure 4: The legend is hard to see with its small size and placement. I recommend enlarging it and placing it at the top and/or direct labelling the y-axes and lines.

233: What is known about the bed of the glacier here? Is it thought to have significant melt? Is the assumption about no basal melting simply for calculations, or is there geologic evidence supporting this assumption?

253: Again, there are references to the uniqueness or special nature of the RAID samples, but they were never described in this manuscript.

Discussion: Sections 5.1-5.3 are very short and do not add much discussion of the previous parts. I’d argue that they could simply be appended to their appropriate section in the results. Section 5.4 has some intriguing points, but is underdeveloped and feels a bit like an afterthought.

Data availability: Are the data (d²H, chemical species, etc) that led to the creation of the age model in an online archive or otherwise available? I didn’t see a link at Rowell 2022 (unless they are contained in that paper’s supplementary material).

Technical comments:

17: The acronym MISI is never used again in the manuscript, and therefore it is not necessary to define it here.

21: Both of these acronyms are only used once elsewhere. Remove acronyms and just spell out in second instance.

26: LIG is used three other times. It may be better for readability to simply spell it out each time.

45: Suggestion to consider flipping the sentences in this section so that you give a brief summary of the important necessary information here, and then end with the sentence that “a detailed description…”.

Section 3: The numbering scheme is odd with 3.0.1, 3.0.2, etc. Since there is no 3.1, shouldn’t these simply be sections 3.1, 3.2, etc.?

67: Is SI core the name of the ice core, or an abbreviation for Sherman Island? If it is the core name, perhaps adding a date and/or length to the end will make it more unique of an identifier. If it is an abbreviation, Sherman Island isn’t abbreviated elsewhere (which I think is good)..

73: This phrasing seems odd to me. Perhaps “This variability could reflect the local geography of Sherman Island…”

74: “and IS then closely by…”

Figure 2 (and others): The line thicknesses are rather thick, which makes it hard to see small details where exactly indicated lines are falling. Consider making the thicknesses thinner.

102: Sentence starting with “That is” is a fragment.

111: a priori not prior, I think here.

Figure 4 (and others): This particular shade of red and green is difficult for many colorblind people. The colors aren’t overlapping, so it isn’t as big of a concern, but something to consider if you revise your figures. However, someone colorblind would not be able to make the connection from your legend to the lines.

178: This paragraph could be merged with previous.

Fig 7: Is this the best x-axis display for this plot? It seems to add more confusion and oddness since the yr BP axis goes negative and positive on the log axis.

General Comments:

There are a couple of areas within the manuscript where I felt lost and needed to read Rowell et al, 2022, for context. These areas are discussed in the Specific Comments below. If these areas are addressed, it will make the manuscript much more accessible to readers unfamiliar with Rowell et al, 2022, and help the manuscript achieve a “stand-alone” status.

With minor revisions, this paper will be an important contribution to the Ice core science at the three poles special issue.

Specific Comments:

35: Without the context from Rowell et al, 2022, it is not clear why WACSWAIN would choose to drill at Sherman Island to investigate the WAIS during the LIG if there is high risk of no LIG ice at the site. A sentence (or two) further explaining why Sherman Island was chosen as the site to investigate the LIG will help the reader understand the scientific context for drilling at Sherman Island.

35: Without context from Rowell et al., 2022, it is not clear why the RAID was chosen for the drilling. All that is needed here is some of the context that is contained in Rowell et al., 2022, such as …high risk of not finding LIG ice at Sherman Island… the RAID was used, as opposed to a conventional ice core drill, to obtain a lower resolution water isotope record from the site which would indicate the ice age at the bedrock and give a first indication of the climate signal at Sherman Island… the RAID will save on logistics… etc..

40: The first sentence in the last paragraph talks about the “last few centuries” whereas the previous paragraph focuses on the LIG. I believe I understand your overall point – a reliable age scale is needed for the core – but the last paragraph could use some rewording to make it fit better within the context of all of the previous text up to that point.

60: It would be good to include a reference regarding “sulfur (S) isotope analysis to differentiate between background and volcanic samples”.

Technical Corrections:

30: The word “in” is not needed I the sentence that reads, “the Abbott Ice Shelf in between continental Antarctica and Thurston Island”.

45: The acronym for RAID is defined earlier in the text (at 30). Instead of “Sherman Island Rapid Access Isotope Drill (RAID) field campaign”, it can simply say “Sherman Island RAID field campaign”.

55: The title of this section is “SI:RAID age scale development”. It will help the reader if “SI:RAID” is defined somewhere. The first sentence of this section could be changed from “The Sherman Island RAID age scale” to “The Sherman Island RAID (SI:RAID) age scale” to give a reader a clear understanding of what is meant by “SI:RAID”.

65: Up to this point in the text, Sherman Island has not been assigned the acronym “SI”. So, in the second sentence, it is clearer if “20 m long SI ice core (SI:Core)” is changed to “20 m long Sherman Island ice core (SI:Core)”.

70: In some places throughout the text “sea salt” is used and in other places “sea-salt” is used.

105: In the first sentence, include the word “Island” after “Sherman”.

110: It seems like the first paragraph (both sentences) “It became clear during….dating of the deeper samples” should be in the previous section, and section 3.03 should begin with the sentence/paragraph “The records of chemical species were closely…”.

110: Suggest rewording the second sentence to: “We use the flank-flow depth/age model to guide us in dating the deeper samples.”

130: The authors are asked to double-check their stated values in the text for Rm and Rt. I think the values stated in the text are mismatched. I believe Rm should be 0.038 and not 1.78, and Rt should be 1.78 and not 0.038.

170: Incorrect use of “an”. Should read “we use a depth/age model”.

175: Should be either “Under these assumptions…” or “Under this assumption”.

200: Delete the first occurrence of the word “were” in the following sentence “Peaks present in some species in RAID data were which were not identified in the Core”. Sentence should read, “Peaks present in some species in RAID data which were not identified in the Core”

210: It looks like the word “Table” is missing in “…added to the average uncertainty at the two adjacent ties (2)”. Should it read “…added to the average uncertainty at the two adjacent ties (Table 2)”.

225: Suggest rewording sentence to “The model was used to estimate the age of the ice towards the bed at the RAID drilling site; however, in the lowest 13 meters, model outputs are meaningless.”

265: Previously in the text it is written as “S isotope” rather than “S-isotope”.

300: “IC” is defined anywhere in the text. Could use “chemistry data” to be consistent with Section 2 of the manuscript.

300: Most previous places in the text it is written as “S isotope” rather than “S-isotope”.

305: “SI” isn’t defined anywhere else in text. Suggest rewriting sentence to “The records contained in the existing Sherman Island data”

310: “SI” isn’t defined anywhere else in text. Suggest rewriting sentence to “The Sherman Island ice core was drilled by DT.”

325: The Basen et al., 2012, reference is to a discussion paper. Is there a corresponding final paper to reference? If not, a reference to a published, non-discussion paper should be used in lieu of this one.

340: The stated reference is to a pre-print. Printed manuscript reference is: Crick, L., Burke, A., Hutchison, W., Kohno, M., Moore, K. A., Savarino, J., Doyle, E. A., Mahony, S., Kipfstuhl, S., Rae, J. W. B., Steele, R. C. J., Sparks, R. S. J., and Wolff, E. W.: New insights into the  ∼ 74 ka Toba eruption from sulfur isotopes of polar ice cores, Climate of the Past, 17, 2119–2137, https://doi.org/10.5194/cp-17-2119-2021, 2021.

This is a great paper, and the authors are to be commended!   They have written the first paper to convincingly demonstrate the utility of the RAID system.  They have also provided a really impressive and careful analysis of the timescale for the Sherman Island site, which will prove useful as a climate record, and strongly supports the idea of drilling a complete "normal" ice core record at this site. 

I have a few criticisms that I suggest should be considered in a revised paper.

First, a few minor grammar/style things:

Please define abbreviations before using them.  In the asbstract, "LIG" is used, but not defined.

The terms SI:RAID and SI:CORE are used but not defined. At first, I thought SI might mean Supplementary Information!

In the abstract, it is stated that the new record is the "oldest, continuous, ice-derived palaeoclimate records for the coastal Amundsen-Bellingshausen Sea sectors."  This is a stretch.  WAIS Divide gets its snow accumulation from the Amundsen Bellinghausen Seas, and probably reflects the Amundsen-Bellingshausen Sea sectors almost as well as the Sherman Island site does.  Instead, I would say "oldest, continuous, ice-derived palaeoclimate records *IN* the coastal Amundsen-Bellingshausen Sea sector.

On Page 1, where Joughin and Alley 2011 are cited, I suggest a more primary source, rather than a review paper.

Similarly, on page 2, I don't think Deconto and Pollard is the best citation for the statement that the WAIS may have collapsed in the past.  It would be better to cite evidence for this from sea level records or other papers that provide data suggestive of this.  Deconto and Pollard is a modeling paper that *assumes* WAIS has collapsed.  It is not a source of evidence that it did.

On page 4, the model of Martín et al. (2015) is briefly described. But I would like to understand the motivation for using this model, rather than simply models such as that of Dansgaard-Johnsen, which is commonly used.  Furthermore, elsewhere in the paper other models (e.g. Liboutry) are used, so this is confusing.  Also, it is stated that horizontal advection is neglected, but nothing is said about whether this is a reasonable assumption for this site.

On page 5, I find the Caption for the graph to be confusing.  It is stated that divide flow is shown in red, flank flow in blue, and that green shows the optimized model.   This implies that the best model is inconsistent with both divide flow and flank flow.  But in the text, it's clear that this is NOT the difference between the curves.  The difference is that assumptions made about the snow accumulation rate were probably too simple.  That's a nice result, but it is not at all obvious.  As far as I can tell, the green curve actually uses divide flow, but allows for variable accumulation.  (If I am wrong, then I am confused by the text, and some work is needed to make it clear what has been done).  I strongly suggest renaming the curves, with something like "divide flow with Dome C-proportional accumulation", "flank flow with Dome C-proportional" and "divide flow (??) and optimized accumulation".   

I have one significant criticism.  Why is Dome C used for accumulation rate in the first place?  It is about as far from Sherman Island as possible.  Why not use WAIS Divide?    Even South Pole would be better!   The authors show clearly that the assumption that the accumulation history is proportional to that at Dome C does not work -- if it did, then the "optimized" timescale would be very similar to the "divide" timescales.  There is a missed opportunity here.  The authors do not show what accumulation history goes with the optimized timescale (they should show that!).   It would be very interesting to know whether that history agrees better with Dome C or with WAIS Divide or South Pole.

I would very much like to see the "divide" and "flank" timescales re-calculated, using WAIS Divide (and perhaps also South Pole), and some commentary on which one provides the closest fit to the optimized timescale.