Review of “The first firn core from Peter 1st Island - capturing climate variability across the Bellingshausen Sea” by Thomas, et al. for Climate of the Past

Thomas et al. provide the initial results from a firn core retrieved from Peter 1st Island. The stable water isotopes and snow accumulation are compared with reanalysis data which they also ground-truth with meteorological data from an automated weather station. The authors provide evidence of some compelling connections with regional/large-scale climate variability to support further drilling campaigns in the area. My primary concerns rest with the low sample number used in the correlation analyses and the influence of substantial melt in this shallow firn core. Although the authors address these limitations to a degree, I would offer a few suggestions or clarifications that will hopefully help bolster their results and interpretation. Nevertheless, the authors demonstrate the potential that a longer core from this site could provide to understand past climate variability in the Bellingshausen Sea region.

Major revisions

Many of the conclusions from this manuscript are reliant on correlation analysis and standard p-values. However, only 15 years of data are available, and only 13 when two melt outliers are removed. Although I am inclined to trust the relationships discussed in the paper given the strong correlation values, I do wonder about “significance”. The authors attempt to reconcile this issue by creating a pseudo-core data. I worry about the pseudo-core based solely on the ERA5 data which is then correlated with additional variables from ERA5. Could the interpolation schemes used in ERA5 generate spuriously high correlation coefficients? For instance, are there inherent correlations among adjacent or proximal grid cells that may be coming to light here. Circumventing the inability to collect more data at this stage (although I very much hope you are able to obtain a longer core from this site in the future), perhaps applying a bootstrapping method to the correlation analysis would provide somewhat more robust p-values.

The melt in the core is also a concern which relates back to the robustness of the correlation analysis. Given such a short record, the disruption to the isotope, accumulation, and chemistry records by the melt events could greatly alter the results. The authors do address this by removing the two years of greatest melt in some analyses. How much does the melt influence preceding years? For instance, the year 2012 is almost non-existent. Could its inclusion in the correlation analysis skew results? However, once you start eliminating every year potentially altered by melt, you would not have many years left in your analysis. The authors make a good point that melt events are likely more pronounced in recent years and would likely be fewer in frequency in a deeper core. This prediction supports the need for a deeper core but does not address the issues in the interpretation of the results for this manuscript. The data are what the data are; there’s no changing that. Is there a way to ensure the results are more robust? In the case of melt affecting the time scale, could making a stronger case for the volcanic marker help here (see minor revisions)? Serendipitously, the second melt event in 2006 is captured by the AWS data. Could that be utilized to provide support for the time scale and subsequent analyses? I apologize that I don’t have great suggestions for how to overcome these limitations.

Minor revisions

Lines 56-60 - citation needed

Line 68, should be 4 instead of 3 in list

Line 85-86, you mention GPR profiles indicated stratified layers in the upper 14m, and the GPR was set to 43 m. How did the layers appear below 14 m? Including a mention from the Thomas et al. 2021 paper here might help with demonstrating feasibility of a longer core record. I would suggest moving the bullet point from the conclusions (line 526) to this location, especially since it is not a conclusion of this manuscript’s analysis.

Line 101, it may be helpful to make this point in the introduction. At present, the ERA5 comparison with the AWS data appears like a separate side-project in the introduction. Making the point that reanalysis products have struggled in Antarctica and that’s the motivation for the AWS comparison would justify it a bit better in my opinion.

Line 131, were other thinning functions tested in light of not knowing the full depth? 

Is there a reason for not including MSA in figure 2?

Line 165 - greater than 50% of hourly data is a low threshold. How might changing this threshold influence the reported warm/cold bias in ERA5? With such a high correlation between ERA5 and the AWS, I imagine the variability would not be affected too much by altering the threshold, but it could make a difference in the magnitude of temperatures. Might be worthwhile since there is considerable discussion regarding the lapse rate in reconciling these differences.

Line 187 - just to clarify, the lapse rate estimate applied to the AWS greatly reduces the ERA5 warm bias, but the lapse rate estimate applied to the drill site underestimates temperature at the drill site? Is this the reported cold bias mentioned in line 380?

Line 321 - I am not sure the analysis around SAM and SOI are necessary for this manuscript given how heavily caveated it is due to the short time span. I see the need to demonstrate a longer core’s potential in exploring these dynamics, so I can understand its inclusion at this stage. I’m just not sure it adds much at this stage. I would suggest either paring it down by noting much of this analysis takes place in a positive SAM phase and merely mentioning the non-stationarity suggested by Figures 5g-i. Perhaps focus on the Amundsen Sea Low variability at this stage, which could be scaled up to explore SAM, SOI dynamics with a deeper core. 

Figure 5 c and f do not match very well in the details, although in the greater scope the match is sufficient. Does the lack of strong correlation between accumulation and zonal winds extend from post-depositional effects at the drill site or small-scale wind patterns affected topography, coastal proximity, etc.?

Line 347: Regarding the sulfate peak at 4.6m, described as the volcanic eruption. Why are there accompanying peaks in Br and Na in Figure 2? I had assumed this was an issue of percolation due to melt. Is the sulfate in Figure 2 just biogenic. If so, how did you tease it out? Would air trajectories help confirm whether volcanic material was transported to your drill site?

Line 450-454 - the potential to reconstruct blocking variability and the connection with atmospheric rivers are exciting! I think this could be fleshed out a bit more, but that’s merely my personal preference. 

Thomas et al. present a new shallow firn record from Peter the 1^st island in the Bellingshausen Sea, Antarctica. After developing the timescale of this 15-year-long record, they evaluate relationships among accumulation, melt layers, and d18O with reanalysis products and with a short automatic weather station dataset. The paper lays the groundwork for interpreting a potential deep ice core drilled at the site, which is expected to span about 200 years. With some organizational restructuring and some clarifications, the paper is worthy of publication in a journal such as Climate of the Past.

Major comments:

The development of the age model is foundational to the paper, as the accumulation rate derives from this and is used in the subsequent correlations. I am surprised by the tiny size of the seasonal variations in the major elements, and am curious what nssCa would show. Ice core records from West Antarctica typically show well-resolved annual dust peaks, so it seems like Ca could be really helpful here. If the data were collected, why are they not included? At any rate, most of the annual picks seem reasonable visually; however, the pick between 2005 and 2006 (Fig. 2) aligns with troughs in most of the major ions, rather than peaks as for most years – and there are peaks to either side of this pick that are not used. Why is this?

Second point related to the timescale is that the potential tie-points (Puyehue Cordon Caulle and the two melt years, 2006 and 2013) need to be tied in earlier in the text and not saved for the discussion. This text should be part of section 3.1 so that the reader can fully evaluate timescale development with all information in the same place. The apparent PCC peak should be highlighted in the figure. The 3-point smoothing draws the eye away from this prominent peak, but I think it would be helpful to make it more obvious. I would suggest instead of a dashed line, using a thin solid line for the full-resolution data. In the Fig. 2 caption, clarify that summer peaks are defined as Jan. 1 (or alternate date, as chosen by authors). Labeling the years between the lines (rather than directly above each line) is a bit un-intuitive for me, so it would be helpful to define exactly what the black lines represent.

I also agree with Reviewer 1’s main points, but as these concerns have already been raised, I will address other issues.

Minor comments:

Line 103: Please include actual time spans of data collection. As written, it implies one year – but the actual record is much shorter.

Line 107: Needs a transition here.

Line 120: This seems like a small subset of the major ion suite that is typically measured. Why were Mg2+, Ca2+, NO3-, NH4+, Cl-, etc. excluded from the analysis? Or were they analyzed but not included here? It seems like nssCa could help support the timescale development, so it would be nice to see it included. If the other data are to be presented elsewhere, then include a statement to that effect. Okay, re-reading this I see in section 3.7 that other major ions were also measured. Please update the methods to include all measured ions, and present the data in Fig. 2 to reinforce timescale picks, as suggested above.

Line 129: correct to “site’s”

Line 130: “ensuring” seems like an odd word here. Wouldn’t “assuming” be more appropriate? And why is the assumption that the firn core only samples the upper 10% of the ice cap’s thickness? I think a little more explanation of the reasoning here would be helpful.

Line 131: Include some uncertainty estimates based on other thinning models. How much could that 130 m estimate reasonably change based on the model you use?

Line 161: Should be “provide”

Line 172: the circle appears purple, not blue. For this figure, the black/gray/dark purple coloring makes it a little tricky to visually resolve the lines. I would suggest using more of a royal blue (i.e, more contrast with the black and gray) for the ERA5 data.

Line 191: There is a fragment here that needs revision.

Line 179 and the paragraphs that follow: Here it is stated that “a lapse rate is applied.” However, it is unclear what rate is actually used and why. This whole section (3 paragraphs about lapse rates) could use reorganization and clarification.

Line 200: This section 3.3 seems like a subheader of the section above, as the two topics are closely related.

Line 207: Replace “meltiest” with “heaviest melt” or similar

Line 211: There does not appear to be a melt layer in Dec 2006 (Fig2). Can the authors clarify or explain why this may be?

Line 231: Another fragment here that needs to live its full life as a sentence.

Line 239: This paragraph discusses accumulation, which is listed as the header of the next section. This whole part of the discussion needs reorganization in its structure and labeling. Part of this could be to remove the heading of 3.5: Accumulation, and allow the discussion of accumulation to follow smoothly from heading 3.2 as these paragraphs are so closely related.

Line 302: Change “warm-moist” to “warm, moist”

Fig. 5: Please label the y-axes next to panels d and g “ERA 5 P-E (year range)” so it’s clear exactly what the plots show.

Line 329: Change “temporality” to “temporally”

Line 350: Please cite Koffman et al 2017 for the PCC reference horizon

Section 4.1 can be moved into the first discussion of the age scale.

Line 365: Would “ice cap” be more appropriate than “ice sheet” here and throughout the paper?

Line 384: looks like a typo in the citation; colon should be a semicolon

Line 403: Correct to “or has been lost”

Line 422: “island’s”

Line 440: awkward syntax in this sentence, please revise

Line 448: Please clarify which continent (South America or Antarctica” and change “to” to “the”

Line 466: Change “is” to “can be”

Line 478: correct to “is a subject for further study”

Paragraph lines 477-484: This paragraph needs to be reorganized. Give the information and observations about melt years and then end with saying that the topic will be studied further – but leading with this does not fit well with the additional details given immediately afterward.

References:

Koffman, B. G., Dowd, E. G., Osterberg, E. C., Ferris, D. G., Hartman, L. H., Wheatley, S. D., Kurbatov, A. V., Wong, G. J., Markle, B. R., Dunbar, N. W., Kreutz, K. J., & Yates, M. (2017). Rapid transport of ash and sulfate from the 2011 Puyehue-Cordón Caulle (Chile) eruption to West Antarctica. J. Geophys. Res.  Atmos., 122. https://doi.org/10.1002/2017JD026893