the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Apr 2026
Fidelity and stratigraphy of the Antarctic Allan Hills old ice archive from Continuous Flow Analysis
Abstract. The Allan Hills blue ice area, East Antarctica, offers a unique opportunity to extend the ice core record beyond 800 thousand years (kyr), with ice as old as 6 million years recently recovered. The ice in this area demonstrates several peculiarities—such as strong layer thinning and folding—that warrant an in-depth investigation of its stratigraphy and the fidelity of the climate record it contains. Here, we present a high-resolution Continuous Flow Analysis (CFA) of two shallow ice cores from the Allan Hills (ALHIC 2201 and ALHIC 2302), spanning the upper 69 m and 46 m, respectively.
Our CFA analysis includes methane, water stable isotopes, and particle concentrations, allowing us to characterize their variability and assess how well geochemical measurements are recorded and preserved in Allan Hills ice. Dating of the ice (using the 40Ar chronometer) revealed ages ranging from ~150–1200 kyr with substantial age discontinuities and folding, highlighting the complex stratigraphy of ice in this region. To assess glacial-interglacial variability, we conduct descriptive statistical analyses of each measurement relative to the EPICA Dome C (EDC) deep ice core record, which served as a reference benchmark over the last 800 kyr. Relative to EDC, we find data representing warm climate states are overrepresented at Allan Hills, and the Allan Hills records typically exhibit a narrower distribution compared to the EDC record. These observations confirm previous suggestions of interglacial bias and glacial cycle averaging in some sections of Allan Hills records. Differences between ALHIC 2201 and ALHIC 2302 further suggest that coring location within the Allan Hills region affects signal preservation in the record. Our high-resolution investigation of this ice is a critical step toward improved interpretation of the discrete, multi-million-year records from the Allan Hills.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Climate of the Past.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(3084 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2045', Anonymous Referee #1, 27 May 2026
-
RC2: 'Comment on egusphere-2026-2045', Anonymous Referee #2, 31 Jul 2026
This paper describes the analysis by continuous flow techniques of two upper sections of ice core taken from the Allan hills region of Antarctica. Ice cores from this region have sampled by far the oldest dateable ice in Antarctica, with ice as old as 6Ma reported. The significance of this is the access to climate and atmospheric signals far older than the stratigraphic ice cores such as from EDC, Dome Fuji and Vostok. Their value then is enabling data from, and insights into, earlier significant climate epochs such as the mid-Pleistocene transition when the glacial cycle changed pacing from ~40ka to ~100ka, and the earlier mid-Pliocene warm period.
The authors, and indeed the wider COLDEX community would recognise that dating these older cores is problematic, and that the proxy signals are difficult to attribute and understand due to the accumulation and complex flow regime of the ice itself, with stagnation, overturning and upwelling of the ice as it closes on to the Transantarctic mountains.
This paper takes some novel approaches to the analysis of climate proxy data from two fairly shallow sections of cores (46 and 69m – the upper, least damaged, sections of deeper ice cores recovered from the Allan Hills region) which nevertheless include ice that may be as old as 1.2 Ma. It examines evidence for biasing of the proxy data towards warm interglacial periods that has previously been reported.
The paper is well written, and interesting, and certainly appropriate for Climate of the Past. It does tend towards descriptive statistics and sometimes simplistic explanations of likely causes of the observations, but there are some interesting approaches to looking at the data that others will likely take up with other cores.
Generally, the paper is fine, though there are a mixture of minor and perhaps more substantial comments below that the authors may wish to address. There is a tendency to write off anything that doesn’t fit the general theme of the paper as ‘contamination’ (e.g. in discussing methane data), or ‘stratigraphic inconsistencies’ which probably captures the main reason why these cores are difficult to characterise! And when the two cores show different characteristics, then ‘differing flow regimes’ can be brought to bear.
The major observations of biasing of the ALHIC cores towards interglacials might be more robust if the choice of data sampling from the EDC core was better justified. The EDC data is sampled by taking 1000 random mean data points from one meter depth intervals. The use of data on a depth scale already introduces a bias of EDC data points taken from the glacial period given that the last interglacial spans a significant portion of the (depth) record. A quick histogram of the EDC isotope data in binned 1m intervals, and an eyeball analysis suggest about two thirds of the record is glacial. So, it might not be surprising to find the ALHIC records biasing towards the interglacial given that they have (presumably) a quite different age-depth distribution. It is also worth pointing out that the variance of the EDC data differs across the climate states, which would impact on some of the analysis here.
The CFA (continuous flow analysis) technique applied to the two cores is very well described, with a clear explanation of the analytical procedures and methods of calibration, and good figures illustrating the methodology. In terms of modern ice core CFA techniques, particularly those of the European groups, the range of analytes is limited, with only methane, microparticles (dust) and the stable water isotopes delD and del18O measured. But, given the later scope of the manuscript, which examines whether Allan Hills cores can reliably capture long-term trends in paleoclimate, this is not a limitation, but a sensible selection of primary proxies with which to test whether blue ice cores from the Allan Hills adequately record climate data.
There are a few specific remarks on the analytical section:
L66: the sentence “….and glacial ice to the average are yet to be fully” is incomplete.
L195: in describing the validation of the methane calibration, comparison is made between the CFA methane, and measurements on discrete samples taken from the two cores. This showed a mean offset of -0.29 ppb, with a precision of +/- 53.2 ppb. I can’t help feeling that this almost perfect minimal offset is either luck or judicious rejection of data that do not compare well. This can anyway be seen in the figure 3, where the comparison is poor in the upper 10m of the 2302 core (and rejected from the comparative statistics). I’d like to see a little more written detail on the comparison between discrete and CFA data – actually figure 3 looks pretty good for the 2201 core, so I’m surprised at the poor precision figure rather than any offset.
L206: Note is made that the calibration of the stable isotopes was carried out at the start and end of each day of analysis, but that only an overall campaign calibration was used to calibrate the data. I’ve no objection to this, but it would have been useful to be given the number of days and the overall period of the campaign. Instrument stability over a few consecutive days of analysis would be more convincing than if the campaign had continued over several weeks. Could you include details of the campaign period in the paragraph L221 et seq?
While commenting on the analytical technique, I’d remark that the figures A1 and A2 are quite convincing. I’d remove the connecting lines across the data gaps in the dust data (e.g. around 42 and 44m) in A1, and for all three species plotted in A2. The dust measured at the two institutions is remarkably similar, given the nature of ice selection and preparation for CFA analysis. Figure A2 is also convincing too; there are small offsets in the methane which I’d expect given the nature of sample preparation (removing contaminated ice from the ends of short CFA sections) and the difficulty of precise depth registration.
Figure A3 and the relevant sections of text give a good explanation of the response times for each analyte in the CFA flow path and provide good evidence for the claimed spatial resolution of the data.
Section 2.3 describes the dating of the ice sections. I’ll accept that dating via the 40Ar technique is always going to be difficult and, unlike for example the ~800ka European and Japanese cores, will never have the precision possible where the ice is stratigraphically preserved. But, the ice obtained in the Allan Hills is so much older, so offers climate/gas data that is just never going to be possible from the stratigraphically preserved cores. So, dating the ice is critical to placing the climate data in time. The age uncertainty from the 40Ar method is noted as +/- 11% (L271) or +/- 60ka (L272). However, although the age uncertainty is given, often in the text the ages seem to be taken with more confidence than might be justified.
The clearest evidence that the dating of the ice must be wrong in places is shown in figure A4. There are several places where the methane and del18O indicate the ages of the ALHIC cores must be wrong. Given that methane is globally mixed (as mentioned in the text, L370), there are dated methane values that do not match the methane of the same age in the EDC core (examples at around 150, 350, 390, 550ka in 2201, and 270, 360, 405ka in 2302) with a similar mismatch in the del18O at the same age points. I accept that the dating precision of +/- 60ka could move many of these points laterally in time to match with EDC, but the very fact you have clusters of points of similar age and methane/del18O suggests that the age precision is better than +/-60 ka.
The worry is that errors in ages of the ALHIC cores impact some of the data analysis across the remainder of the manuscript. As an example, figure 5 describes a section of potential folding in core 2201 with an age of 619 ka bracketed by ages 393 and 399 ka – sections that already look incorrectly dated when looking at the methane, d and del18O data around 390 ka in figure A4.
L 274: 15cm of adjacent ice – can you clarify this sentence? I assume it is horizontally adjacent, rather than adjacent in the vertical sense. Given that elsewhere in the manuscript (Lines 409 et seq.) the dipping of the ice layers is given some significance, is it not fair to say that even a horizontally adjacent piece of ice might have a different age? But, then you could point to Figure A2 to show how reproducible are the analyses taken from presumably two adjacent CFA sections of the same core. Though this might then undermine your argument about signal smoothing due to the dipping layers described in L417 et seq.
L357: you state all geochemical measurements except methane in 2201 have lower variance than EDC. Yet table 1 shows higher variance for 2201 dust and d than EDC.
L426: rather than “over three times narrower than that of” you could say with more clarity “is less than a third of”.
L458: states ice cores only record environmental conditions only during snowfall. This is not correct: dry deposition of many species is likely, especially in low accumulation areas. Specifically, dust can be deposited independently of accumulation (as the dust on the hood of my car during a dry spell attests). Of the proxies described in this paper, methane is trapped independent of snowfall – it is the depth of pore close off and trapping rather than the magnitude of the methane signal that changes with accumulation rate.
Section 3.3.1 takes a decile-based approach to testing the biasing in the climate capture in the ALHIC cores. It’s quite a simple, but clever approach and figure 7 does seem to justify the observation in the text that, for example, the majority of the ALHIC methane data fall above the median EDC methane value, biasing ALHIC methane towards the interglacial climate states. However, as previously noted, this might arise from the fact that EDC was sampled on 1m depth intervals, and the glacial periods are over-emphasised in EDC data on a depth scale.
Section 3.3.2 and Appendix B is a quite clever analysis of the data distribution and smoothing. It is more robust in statistical technique than some of the more descriptive statistics appearing in the main body of the paper. While I could not follow all the detail (what is a Wasserstein difference – this could be explained), it was an interesting (new?) way of using a remote dataset (the EDC core) with resampling to examine biasing and smoothing in the ALHIC cores. It may identify the differences between ALHIC and EDC cores, but surely does not explain the climatic/glaciological origin of the differences. And how could it? The age models of ALHIC cores are not well constrained, and flow regimes are so strikingly different. Nevertheless, a very interesting approach.
Citation: https://doi.org/10.5194/egusphere-2026-2045-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 337 | 134 | 21 | 492 | 31 | 28 |
- HTML: 337
- PDF: 134
- XML: 21
- Total: 492
- BibTeX: 31
- EndNote: 28
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
In their manuscript Hudak et al. present CFA analysis and Ar40 dates of two Allen Hills cores alongside a description of the CFA setup at OSU. The CFA data is than used to investigate differences in variance of the Allen Hills data in comparison to existing EDC data. The application of continuous flow analysis to the discontinuous AL ice is certainly a challenge in terms of data analysis opening interesting avenues of investigation. Overall, the manuscript is well written and argued and fits the scope of Climate of the Past.
I do, however, have a request and some suggestions that should be taken into consideration prior to publication:
All the data analysis that the authors present is based on re-sampled versions of the existing EDC data. Because most of the analysis is dependent on the choices made during the resampling these choices need to be very well justified and compared to the ALHIC data, i.e. its age density. This justification and its implications for the analysis is probably best placed in the EDC data section or in an extra section that describes the statistical analysis. I understand that the age density of the ALHIC data might be ill defined but as so much of the discussion hinges on the differences in the data distributions to the EDC data it is well spent time/space to argue this carefully to take the reader along.
The authors put a lot of focus on purely descriptive statistics (mean, variances, simple correlations). The quantitative analysis is hidden in section 3.3.2 and mostly on the appendix. This is unfortunate as I think the analysis of the variance reduction and the representation of the climate states is clever and provides a lot of substance to the observations. For my taste, most of the descriptive statistics tables/figures could be moved to the appendix instead of the variance reduction analysis.
Specific Remarks
Section 2.2.2: The information about the precision of the water isotope and CH4 data would be a valuable addition to this section to gauge the differences between the discrete and the continuous data.
L192ff: I think all is fine with the solubility correction, but with the large error on the average offset it is not possible to make any quantitative statements. What is also missing is the magnitude of the solubility correction that was applied. If that is on the same order or smaller than the uncertainty of the offset than it is even less advisable to make any quantitative statements. I suggest rewriting this paragraph.
Section 2.2.3+2.2.4: The information about how the delay times for each analyser were determined (and how large/variable they are) seems to be missing. What is also missing is how the missing data at sample breaks was handled. Depending on the measurement resolution a significant amount of data will represent a mixture of non-continuous ice. In the application to the discontinuous ice here, that might be okay, but a few sentences should be added for explanation.
L274: The final sentence here is probably better placed as the second sentence of the paragraph.
Section 2.4: This section is missing the age-scale information for the EDC data. Later AICC is mentioned, but I think it would be better to state that here. What is also missing here is how the original EDC datasets where (sub/re)-sampled for the statistical analysis. It is important to be very explicit with this because especially the variance of the data and its range is very much dependent on the choices made here. The choices made here also need to be justified carefully considering the age density of the ALHIC data presented. Ideally the age densities should match as much as possible.
L295: should be “are due to depths not analysed”
L321ff: I think the comparison Figure A4 should find a place in the main part of the manuscript. Especially if sections are discussed in detail here, as in the final sentence.
L389: Should probably be “The average ALHIC water isotope values…”
Section 3: From my point of view there is an over emphasis, both in text and figure on the simple descriptive statistics of the data. I feel that it does not bring a lot to the arguments the authors are trying to make and could mostly be moved to the appendix, to shorten the main part of the manuscript.
Section 3.3.2: I think the approach used here is quite clever and deserves to not be hidden in the supplement: This is the data analysis that goes beyond simple descriptive statistics and adds quite a bit of value to the paper, from my point of view. I would suggest presenting it a more prominently and extend the section.
Section 3.3.1: I think the decile plots are probably best presented with the deciles as the x-axis labels not the data ranges. That would make the plots cleaner and would also make it easier to look at a specific decile without having to count the labels. Alternatively, the deciles could be added as an additional axis.
Appendix A: Both figures A1 and A2 include linearly interpolated values over data gaps, these should be removed.
Fig A2: It is quite surprising to see so large depth offsets (>10cm) between the replicate measurements. I suggest the authors add some detail on the precision of the depth assignment to the discussion.