Wolff and colleagues present some advantages and drawbacks of available chronostratigraphic parameters to date “old ice records” (stretching back to 1.5 Ma) that should be retrieved from ongoing ice core drilling efforts in Antarctica. Providing reliable chronologies to such ice cores are of utmost importance for climatic interpretations. Dating is indeed sometimes regard as secondary compare to interpretations of elegant proxies or high-tech developments requested to recover such deep-old ices. It is however of primary importance to understand climate dynamics. In this paper, authors discuss synchronization and relative matching between ice core and sediment records using four independent ice core parameters and proxies: water isotopes vs Mg/Ca, dust vs iron acc., CH4, 10Be. The similarities already observed over the last 800 ka between ice core and marine records serve as key assumptions to propose and test matching and synchronizations of old ice records (using these proxies as analogues). Such a discussion is very welcome and this paper certainly contributes to this important goal by suggesting few templates and patterns of relative changes useful for dating. The manuscript is well written, concise and rather clear for a broad audience (not necessarily specialists of one of the presented proxies). I therefore recommend this paper for publication following minor revisions (see comments below).

Comments

Page 2, first paragraph: authors could maybe explain in few sentences the MPT conundrum and why it is particularly “hot” for understanding climate changes mechanisms.

Page 2, third and fourth paragraphs: as this paper is dealing with chronological issues, it could be interesting to read here a bit more about the dating (methods, ages, uncertainties) of the 2 Ma blue ice. For instance: what’s the typical uncertainties associated with these methods? Why is it vital to reduce ages uncertainty to try understanding the underlying causes of the MPT? What is the acceptable minimum uncertainty to interpret the dynamics (amplitude, frequency) of climatic changes over the MPT? Why (and how) is it important (despite uncertainty) to adjunct radiometric ages to fix relative chronologies onto absolute timescale? Etc.

Lines 139-140: please explain how you scaled up the two records. Is maximum stretching still within chronologies uncertainty? This could be interesting since it would somehow quantify the elasticity between two largely used chronologies for two important records, i.e. LR04 and AICC2012 for ODP site 1123 and EDC, respectively.

Lines 151-153: it is interesting to note that these mismatches occur within the Mid-Brunhes transition (MBT). Could it be a problem as well within the MPT where amplitude change drastically?

Line 199: same comment as above, explain the “appropriate scaling” a bit more (how many tie-points, maximum stretching amplitude…).

Lines 225-227: based on which arguments?

Line 269: I’m not sure whether the mention “soon to be published” without any other information is acceptable. It does not tell much to readers. It is a bit annoying since this planktonic isotope and SST records from site MD95-2042 should “serve as a regional template for D-O variability”.

Lines 271-272: I agree, using CH4 requests testing the validity of the underlying assumption first. Is any such East Asian speleothem record already exist or is soon to be published?

Lines 273-277: I would also cite here papers by Giaccio et al., (2015) and Nomade et al. (2019) where millennial events are also clearly identified in records that are also radiometrically date providing potentially important tie-points to fix floating chronologies.

Line 299: add ‘intensity’ to “Earth’s magnetic field”.

Lines 305-306: this is a problem since it invokes some circularity, or at least imply a strong assumption. Why not try using regression method (see Zheng et al. 2020 and 2021 for instance) to normalize old ice 10Be records and minimize climate-related variations? One could for instance use water isotopes or ion concentrations data within multi-linear correction method to obtain climate corrected 10Be record. The idea is essentially to remove the shared variance between 10Be and climatic parameters measured at the same depths. This method has been used in wet deposition environments (Greenland), but could maybe provide interesting results, even imperfects, in such dry deposition settings.

Lines 318-319: this is very unfortunate indeed.

Lines 323-324: authors could emphasis a bit more why 10Be radioactive decay (thus uncorrected 10Be records) could serve as an interesting indicator for long-term age control in case of disturbed or non-continuous ice cores. This was for instance use in Bourles et al. (1989) or to try dating very disturbed (discontinuous) lacustrine records (Lebatard et al. 2010; Simon et al. 2020).

Lines 333-340 and 349-355: one way to try resolving this issue is to use the regression method (see above) and compare such results with estimated accumulation rate (derived from water isotopes) by an iterative approach.

Lines 360-363: I agree the spikiness was bypassed in EDC using statistical method. However, the 10Be measurements density in EDC allowed such treatment. Will it possible to do so within older ice samples considering the amount of material needed to measure 10Be. Further works are indeed needed and maybe 10Be measurements on drill-chips (Auer et al., 2009; Nguyen et al., 2021) will be key to obtain high-resolution 10Be records in old ice records.

References

Auer, M., et al. (2009). Earth and Planetary Science Letters, 287 (3), 453-462.

Bourles et al. (1989). Geochimica Cosmochimica Acta, 53 (2), 443-452.

Giaccio et al. (2015). Geology, 43 (7), 603-606.

Lebatard et al. (2010). Earth and Planetary Science Letters, 297, 57-70.

Nguyen et al. (2021). Results in Geochemistry, 5, 100012.

Nomade et al. (2019). Quaternary Science Reviews, 205, 106-125.

Simon et al. (2020). Quaternary Geochronology, 58, 101081.

Zheng et al. (2020). Earth and Planetary Science Letters, 514, 116273.

Zheng et al. (2021). Quaternary Science Reviews, 258, 106881.

This manuscript’s goal is to provide recommendations for how best to develop age models for new ice cores that are anticipated to retrieve ice from 0.8-1.5 Myr ago.  As significant resources are being invested to retrieve this ice, the thoughtful advance planning for the creation of these age models is commendable. In proposing specific age modeling strategies, the manuscript also generates useful hypotheses about the types of climate responses that are expected to be found in the new ice cores. This manuscript is well written and appropriately aimed at a relatively broad paleoclimate audience. I recommend publication after minor revision. My suggestions for revision are below:

1. Most importantly, it would be useful for the introduction to be more explicit about the practical goals for dating the new ice. Although the manuscript currently introduces the motivation to compare pre-MPT and post-MPT climate responses, it would be useful to list a couple specific hypotheses to be tested that will require good age models and how precise the age models will need to be to evaluate these hypotheses. For example, how well do ages need to be constrained to evaluate shifts in orbital-scale spectral power across the MPT? Do the authors also hope to be able to evaluate leads and lags in the climate system? If so, how will the proposed alignments to sediment core records affect the ability to measure the relative timing of different climate responses?
2. At the end of the manuscript, it would also be useful to propose some criteria that might be used to evaluate the success or uncertainty of the proposed age models.
3. Line 75: Please clarify that Prob-Stack (Ahn et al., 2017) does not have its own age model. Because Prob-Stack uses the LR04 stack as its initial alignment target, it has implicitly inherited the LR04 age model and, thus, has at least as much age uncertainty as the LR04 stack.
4. Line 108: This sentence is unclear. What exactly is “higher” at Dome Fuji?
5. Lines 151-153: Please describe the characteristics of the deep water temperature signal from 450-550 ka that are also found in some of the older sections of the 1123 signal. Are there specific ways to identify in advance which parts of the 1123 signal may be problematic for alignment?

Review of “Stratigraphic templates for ice core records of the past 1.5 million years” by Eric Wolff et al.

Wolff et al. address the challenging problem of dating ancient ice from Antarctica with ages exceeding 800 ka BP – currently the oldest ice in the continuous ice core record. Such dating is challenging in particular when the ice is no longer in stratigraphic order. The authors suggest 4 ice core records (dD, dust, CH4 and 10Be) that may in theory allow for stratigraphic matching to independently-dated marine-sediment records (of benthic Mg/Ca, Iron, Atlantic SST and geomagnetic dipole, respectively).

The paper is well-written and easy to follow. The topic is suitable for Climate of the Past. I have a series of comments for the authors to consider, in particular for an improved discussion of the approach.

(1) I would ask the authors to provide a more in-depth comparison of the methods, including perhaps a ranking of the methods. There is currently no quantitative analysis of the four methods they propose – this could be remedied by adding a table with the correlation coefficients between the ice core record and the marine target record for each of the methods proposed.

Such a table could also include an assessment of how likely the physics that produce the correlation for the last 800kyr is to persist during 0.8 – 1.5 M also. Such an assessment is likely qualitative and somewhat subjective by necessity (e.g. “likely”, “unlikely”, “uncertain” etc).

A third entry into the table could for example be whether the method would allow for value-matching, or only the matching of sequences.

Combined, these elements would allow the reader to assess the relative usefulness of each method. The authors appear to give somewhat of a ranking of the four methods starting on line 366. The authors appear to favor the Dust-Fe matching over the dD-Mg/Ca matching – a choice that I agree with. This ranking is however never stated explicitly, or justified by the analyses.   

(2) I have reservations about the dD to Mg/Ca matching. To me, one of the most exciting prospects of a 1.5 Ma ice core record would be to investigate the Antarctic dD climate record and its spectral properties. This possibility would be lost if it were wiggle-matched to the deep ocean temperature.

A surprising aspect of the 41-ka world is the absence of a precession (21 ka) signal in the benthic d18O. The hypothesis by Raymo et al. (2006), paper cited in the text, is that the NH and SH ice volume each responded to local precession forcing, yet this cancels out in the benthic d18O (and presumably also in benthic Mg/Ca) as the precession forcing is out of phase between the hemispheres. One of the key questions of a 1.5Ma ice core dD record is whether it has power in the precession band in the 41 ka world – the Raymo hypothesis requires that it does. If the dD record is wiggle matched to a benthic record, we would lose the ability to independently assess the differences in spectral content between such records – one of the key scientific objectives.

The authors suggest that the ODP 1123 benthic Mg/Ca resembles Antarctic dD so strongly because both are controlled by SH high-latitude SST. They follow the argument by Elderfield (2012) here (Lines 125-129). Antarctic dD also strongly resembles global benthic d18O (LR04), and it also strongly correlates with mean ocean temperature from ice cores (Shackleton et al., 2021; Fig. 3). I suspect that LR04 may actually give a better correlation to dD than ODP1123 Mg/Ca does (can you check/show?). Since bottom waters formed around Antarctica are at the freezing point, lowering Southern Ocean SST will not cool them further. It seems to me that global mean ocean temperature, the rate of bottom water formation, and circulation may all impact ODP1123 benthic Mg/Ca.  ODP1123 certainly looks identical to LR04 (Elderfield 2012) in d18O, suggesting good connectedness to global ocean conditions.

For these reasons I would ask the authors to reconsider recommending wigglematching dD to ODP1123 as a dating strategy.

(3) could you elaborate on the temporal resolution needed to do the CH4 / planktic d18O matching, and whether this resolution is available in the planktic record (which is promised but unfortunately not shown). For example, planktic d18O misses a DO event around 778ka BP that is clear in the EDC CH4. Could there be DO events that are missed altogether by both records?

(4) if Raymo et al. (2006) is right, the MPT reflects a transition from terrestrial to marine terminating Antarctic ice sheets. If so, could we get local Antarctic dust sources to contribute to the ice core record? Could this impact the dust matching?      

(5) line-by-line comments:

Line 40-44: acknowledge the Japanese and US ice core communities are also working on this in funded projects.

Line 65: other tuning targets are air content and d18O-O2.

Line 115: I think both cited studies suggest dD is a proxy for site temperature – they just disagree as to what the correct calibration is.

Line 128: The temperature of the deep waters formed around Antarctica is probably always close to the freezing point. Could it not reflect the volume of deepwater formation, for example, and the mixture with other deep water masses?

Line 131: You could add Shackleton 2021 here too (new MOT data from MIS 4). In that paper she explicitly plots the very strong correlation between MOT and ice core d18O, which strengthens your argument as you correctly note.

Figure 3: is it possible to also plot the comparison as a Mg/Ca vs. dD scatter plot? That way the reader can assess the correlation better.

Line 146: can you give the correlation coefficient for the comparison?

Line 177-180: I fully agree with this sentiment, but feel the same logic can (should?) be applied to ODP 1123. The Benthic temperature has an imprint of southern high-latitude SST, but also of global ocean temperature, volume of deep water formation, circulation, etc.

Line 203: How good is the match? Can you quantify, e.g. via a correlation coefficient?

Line 255: Perahaps note that DO 2 has no CH4 peak at all.

Line 268: Throughout what? Throughout the record? throughout the Pleistocene?

Line 269: can you give more details than “soon to be published”. Is there a title and author list?

Line 317-318: This is unfortunate (though no fault of the authors). why is the full dataset from a 2013 publication not publicly available at this point? Can you provide more details?

 Line 366: Here the authors appear to suggest a hierarchy with the dust matching being considered more accurate than the other methods. Is this really what you mean to imply?

Line 378: However, wouldn’t the absolute values of the 10Be be challenging to use due to the accumulation uncertainty? Also, the VADM record is really variable, and certain 10Be values are not necessarily very unique.

In anticipation to the new ice core data beyond the Myr that the community will generate, Wolff et al. propose an overview of the possibilities that the community could have to correlate and date such old records. As per request from the handling editor, because the other reviewers have already commented on other points of the MS, and because this is indeed my expertise, I have concentrated my review only on the dust part of the paper.

I have two major issues :

- The authors compare the dust flux in ice on one hand, and an iron concentration on the other, justifying it by the fact that claim dust flux is what they will get in older ice cores, and claiming it is fair to compare it to concentration. I am sorry, but to me it is a bit like comparing apples and pears. And it has been a redundant problem within the dust community working on different archives. As the authors mention, to calculate a flux, one needs ages and accumulation rates, but those accumulation rates would also influence a concentration profile when using it. It is commonly accepted that fluxes or at worse ratios should be priviledged when comparing dust proxies between archives, so it is accept that fact and how to solve this issue if this is problematic for some archives or time intervals. Otherwise, again, it makes any comparison quite speculative despite a match which is indeed surprisingly good.

- What is "appropriate scaling". This is somehow linked to my previous comment as it feels like two curves which are in theory not comparable, can be perfectly "matched". There is no details whatsoever how this scaling is achieved. This is also valid for other matching in the MS, and also more generaly in the literature matching ice and marine records. There is absolutely no details on how it is achieved, and it would be good to provide a detailled explanation on this, including the calculations, how it is done ("stretching-compressing" method?, but how, on which time interval, varying depending on the time interval, etc?), all that in supplementary data, so everybody could understand and more importandly, reproduce it.

Minor comments/questions:

- I am ot a specialist in this, but really no means to date both marine and ice cores beyond 1Myr? That would help generating fluxes... And perhaps this is something the community should concentrate on?

- I think the reference citing dust from Patagonia to Antarctic could be reviewed by adding more recent references than only Delmonte, 2008.