Review Servettaz et al. 2022

The climate of the Past

Summary

Air temperature is reconstructed using an East Antarctic ice core from a relatively high accumulation site, Aurora Basin North. Two methods are employed and compared, traditional d18O and 15Nexcess/borehole temperature.

I suggest that the article is accepted after the following comments has been addressed.

Major comments

Section 4.3 about teleconnections is too speculative. It needs to be removed or the analysis and argumentation strengthened. It takes away from an otherwise sound story. Who knows what type of teleconnection was active in the past? Keep in mind that the study that you have been looking for a suitable pattern in (Marshall and Thompson 2016) is only based on about 30 years of reanalysis data.

Especially the PSA2 claim is speculative. Even if the connection was the same in the past it only explains about 9% of the geopotential variability. You can not look at the pattern in the modes in the Marshall and Thompson, 2016 study and think that this would fit. A model simulation would be required to test this hypothesis. You do not have to solve everything here stick to what is certain.

The SAM and the temperature reconstruction in Figure 11 do not resemble one another. Even the long-term trends oppose.

• The 1991 end year seems arbitrary, why not go back to 1979, or calibrate with 2-year averages if the age scale isn’t reliable at deeper depths? Fig 6b the r^2=0.316 value seems low. How does this compare to other studies?
• Is it correct to use the 2mT from MAR? Consider using a simple Rayleigh-type model (e.g. SWIM) instead to reconstruct surface air temperature at the core site (Markle and Steig 2022; Jones et al. 2023). Linear reconstruction can overestimate the temperature change, which is less of an issue for surface compared to evaporation temperatures, and less for East Antarctic sites (Markle and Steig 2022). But can nevertheless introduce an error (Markle and Steig 2022). If not accounted for it feels like the latest knowledge has been accounted for in the borehole/d15N reconstruction, but not for the d18O reconstruction of surface air temperature at the core site.

Using a Rayleigh-type model would also have the advantage that you are not dependent on the age scale not having an error in the same way. Therefore, you would not be limited to data from just 1991-2013.

• Is there a winter bias in precipitation (Servettaz et al. 2020)? So, the d18O temperature reconstruction would be a winter temperature record. Perhaps it can explain the record being less variable. The winter WDC isotopes record was less variable compared to the summer and annual means (Jones et al. 2023).
• In Servettaz et al 2020, you argue that SAM- is associated with d18O peaks. However, no significant correlation with SAM is provided in either paper. The logic seems flawed since the trend in SAM is towards a more positive phase and the isotopes appear to display no trend or positive trend over the recent past, which doesn’t fit with the negative SAM argument in your previous paper.
• The trend analysis period seems arbitrary (Fig. 11). If you start the trend analysis around 1500 to present instead then you would get a positive SAM trend but there is no clear change in the isotopes over this period. From what you have presented there is no clear evidence that the isotopes are driven by SAM. This is too speculative, so it needs to be removed if no additional supporting analysis is provided. Could you use spectral analysis to check if there is a SAM and ENSO signal in d18O?
• Why is the summer SAM index displayed (Fig. 11)?
• Discussion section 4.1. Some of the discussion here is too conversational without backing up with supporting test results. Aim to be concise in the revised version.  

Minor Comments

Title. Remove the punctuation.

Abstract. Remove the text about SAM and PSA.

L158. Remove the first ‘and’. Check for this type of typo in the whole document.  

L 194. Change the word ‘thinly’.

L 203. Instead of calling it ‘resampled’ call it a 5 m moving average. As with resampled taking every 5^th m value comes to mind. Change throughout.

L230. Previously, you wrote that the water isotopes from the ABN1314 core were measured discretely on a Picarro. Here you state that they were measured on a CFA system. I guess they were measured on two setups at two labs but be clear in the manuscript.

L245. Perhaps use the word peak instead of ‘extremum’.

• Number the appendices in the order of appearance in the text?
• How were the short-core isotopes measured? Provide more information about the short core, dating, and which range of years it covers. As the start year for the calibration 1991 isn't the same as the start of the overlap with the satellite era, you cannot call the range in Fig. 6a “their overlapping period” as the full period is not used.
• Define the isotopes and which international standards were used.

L603. I wouldn’t call it ‘many’, as there aren’t that many ice core sites on the plateau. ‘The more abundant’?

L605. D18O is perhaps a proxy for winter temperature while the other represents annual temperature. Therefore, you cannot make a judgment on which proxy is best.

L608. Define SAM and the meaning of the SAM acronym at first mention in the text (L599).

L608. Marshall and Thompson, 2016 were not the first with discovering SAM’s significance on the Antarctic climate. Provide more references.

Add a paragraph that describes the model and reanalysis data that was used. State which organization provides the MAR data and reference it. And that it is a high-res model for the plateau driven by ERA-interim as you did in (Servettaz et al. 2020). 

Figures

Fig. 6. Remove the DRI acronym or use it throughout and define it at the first mention (L231).

My personal preference would be that you call the core “shallow core” instead of short core. Like you did in your previous paper (Servettaz et al. 2020).

Fig. 7. Only orange shading is shown in the plot.

Fig. 8. The line is gray, not black.

Fig. 11. Why is the SAM summer index displayed? Display annual index values instead.

Caption L620 ‘show’.

Author contribution

L928. Say something like ‘contributed to with comments on the initial manuscript’, as otherwise, it sounds like the coauthors were reviewers.

References

Jones TR, Cuffey KM, Roberts WHG, et al (2023) Seasonal temperatures in West Antarctica during the Holocene. Nature 613:292–297. https://doi.org/10.1038/s41586-022-05411-8

Markle BR, Steig EJ (2022) Improving temperature reconstructions from ice-core water-isotope records. Clim Past 18:1321–1368. https://doi.org/10.5194/cp-18-1321-2022

Marshall GJ, Thompson DWJ (2016) The signatures of large-scale patterns of atmospheric variability in Antarctic surface temperatures. J Geophys Res Atmos 121:3276–3289. https://doi.org/10.1002/2015JD024665

Servettaz APM, Orsi AJ, Curran MAJ, et al (2020) Snowfall and Water Stable Isotope Variability in East Antarctica Controlled by Warm Synoptic Events. J Geophys Res Atmos 125:e2020JD032863. https://doi.org/https://doi.org/10.1029/2020JD032863

OVERVIEW

In this manuscript, Servettaz et al produce a 2000-year temperature reconstruction from an ice core drilled in the Aurora Basin region using two independent methods. The first is the widely applied reconstruction from water isotopes. The second is a reconstruction that combines measurement of argon and nitrogen isotopes in ice core air bubbles with borehole temperature measurements in an inversion to solve for temperature history. When the authors compare their water isotope-based and borehole/gas isotope-based reconstructions, they find substantial differences. In seeking to explain the discrepancies between the two reconstructions, they explore a range of possibilities.

This study represents a substantial amount of work. Not only are the described gas measurements time consuming and analytically challenging, but the authors perform a number of complex analyses. This manuscript is worthy of publication, but I ask the authors to address several specific comments here.

SPECIFIC COMMENTS

 Consider separating into paired (or complementary) publications

As this manuscript fits so much into a single paper, the reading of it was quite an undertaking. I think the authors have done some really nice work here that is worthy of publication, but I also worry that - in its current form - the paper is a bit cumbersome. This may not be appropriate to suggest here, but the authors might consider separating the work into complementary studies. There are several aspects of this work that require detailed description and include clever/novel ways of probing the data. However, they may distract from the main message/motivation of the study.

These include (in my opinion) 1) establishing the ice and gas age models (and delta-age) for the core, 2) the inferred ice flow history (and advection-based temperature correction) from reconstructed accumulation, and 3) the comparison of reconstructed lock in depth from gas isotope and delta-depth. On a related note – this work has constraints on temperature (from water isotopes) and accumulation (from annual layer thickness/density) that may also be used to independently evaluate lock in depth and delta age, which might be quite interesting to compare to their gas isotope-based and delta-depth based reconstructions.

There’s clear reason why this content is included in the manuscript, but there’s also plenty of novel content in this portion of the work that may merit its own discussion. The authors should use their discretion in evaluating this suggestion.

 Smoothing of gas isotope data

I may be mistaken here, but it appears that the authors apply a smoothing using 5m moving average of d15N and d40Ar before calculating d15Nexc. They argue this is due to the spatial heterogeneity of bubble trapping, citing Orsi, 2013. However, the cited work suggests that the spatial heterogeneity is mass dependent. If this is the case, shouldn’t it be possible to calculate the thermal gradient from d15Nexcess without smoothing the data?

The influence of advection/ice flow on borehole temperature

The borehole temperature profile is clearly dominated by the ice flow/advection signal. However, the authors argue – in section 4.2 – that the borehole thermometry supports a more variable temperature history than is suggested by water isotopes. Can this really be seen over the strong, advection-based signal?

 Discussion sections

Regarding the discussion sections, I think a more quantitative assessment of the potential hypotheses seems reasonable here. Of the potential variability seen in the gas isotopes (but absent in the water isotopes) there are two apparent explanations. The first is that the gas isotope data are capturing true climate variability that the water isotopes do not, which may be related to water isotopes potentially misrepresenting cold intervals. The second is that the gas isotope data are reflecting more local (or ice flow-induced) phenomena, such as the slope modulated katabatic wind strength and its influence on the strength of the inversion layer and/or convective zone.

As currently written, the latter hypothesis appears quite plausible, but is only examined on a relatively cursory basis. Additionally, the abstract and conclusions of the manuscript don’t mention this possibility at all, and instead imply that the former hypothesis (that the gas isotope-based reconstruction shows real climate variability) is well supported. However, this doesn’t seem to be supported by any statistical analysis.

This is new, hard-earned data, and as such I think discussion of a more speculative nature is reasonable. However, it does not appear – based on the current analysis – that the hypothesis that the gas isotopes are showing a climatological (rather than advection) signal is favored here. Therefore – the authors should either 1) show convincing evidence that this is the case or 2) in the abstract and conclusions include space for both possibilities.   

SAM specific

The first reviewer had plenty of comments here, so I’ll abstain from adding any.

MINOR COMMENTS

 Line 81: Not totally clear what is meant here.

Line 85-86: … is the main source of (temporal?) variation…

Line 90-92: I’m not clear on the meaning of this sentence.

Lines 105-108: This sentence is confusing. Also – is there a reference to cite for the ideal accumulation range for gas isotope-based temperature reconstructions?

Line 122: Can you include a site mean annual temperature here?  

Line 153: Approximately how much ice was shaved off?

Line 164: … laboratory standard of (combined?) N2, Ar, (and) Kr.

Line 172: Which elemental ratios were measured? If none are shown, maybe not a necesary detail to include. 

Line 180: Please include the original and improved pooled standard deviations.

Line 195: … on thinly closed (pores?) may have…

Line 201-202: This is a bit strongly worded here - I would suggest adding some caveats to this statement.

Line 211: How long was the probe equilibrated for?

Lines 213-215: Unclear how this would work.

Line 260: How many tiepoints?

Figure 5: Could the last part of figure A3 also be shown here? This is a neat (but somewhat complicated) way to reconstruct ice flow, so it may make sense to either add some of the details from FigA3 here or move everything to the appendix.

Line 394: It might make it more clear if you use ‘pooled standard deviation’ rather than ‘difference’ here. Also – it would be informative to see d15Nexcess plotted in figure 2 (as a subplot) with both the individual values and moving means.

Figure 7: There is no blue shading in this figure (only red). Also – the authors may have flipped the y-axis firn column height to show similarities with the temperature profile but this isn’t clear in the caption and isn’t how this information is normally presented. One last suggestion – given that the authors suggest that the lock in depth is a function of ice accumulation (line 423), it might make sense to show reconstructed accumulation (as shown in A3) in this figure.

Lines 470-472: For those unfamiliar with inversions, it’s unclear what this means or why it’s being stated here.

Figure 9 (and others): I might have missed it, but I’m not sure the ‘way out’ and ‘way back’ on this figure was ever explained.  

Line 595-597: The slope still appears relatively steep during this interval.

Figure A7: What are the units on the x-axis here? Millivolts?

Figure A1: Looks like the legend has a typo – offsets are larger than what was actually applied.