The Origin and Propagation of the Antarctic Centennial Oscillation
Round 1
Reviewer 1 Report
This paper attempts to argue that an atmospheric oscillation mode in the Southern Hemisphere, here termed the Antarctic Oscillation/Antarctic Circumpolar Oscillation, can be tracked through 11 Antarctic ice cores and that the offsets in timing of this cycle can be used to understand Southern Ocean regional climate and global climate through teleconnections.
The paper is incredibly long and dense and pretty much unreadable. This isn't a knock on their grammar or anything like that, which is frankly the best I've seen in a paper I've gotten to review. I just can't imagine anyone but people who've spent a couple of decades doing statistics and atmospheric science being able to make it more than a few pages in. That's certainly one way to go, but if you ever want a student to read this paper all the way through, I don't think it's going to happen.
The authors argue that they can compare asynchronous ACO events at different sites by using "signpost" events associated with major transitions like the start of the deglaciation, so that the same diachronous event can be identified in different ice cores. I confess that I don't see obvious similarities between the signpost events at Law Dome and Vostok in figure 2 that would convince me that yes, this is the same exact cycle happening at a thousand year lag in Vostok. Cycle 63-62 in figure 2 at Law Dome, for instance - why identify it at 14-8-15 ka when there's another little dip right before it at 15-15.2 ka or so that immediately follows an enrichment, as the (teeny) proposed 63-62 event at Vostok does? Also, who's to say that the deglacial warming at one site didn't start slightly behind another one, making that not a useful way to lock timelines from different cores?
I also don't see much discussion of the resolution of the data at different times or different sites: looking at the Vostok data table the authors cite, it looks like the sampling resolution of the deuterium data is in the 80-90 year range for the LGM and last deglaciation, which is lower than the Nyquist frequency for a 143-year cycle.
The stated motivation for this work is problematic: the introduction argues that because previous work to reconstruct CO2 levels in the atmosphere over hundreds of millions of years, with many uncertainties in the proxies at that age remove, hasn't shown a correspondence between CO2 and temperature at these extremely long timescales therefore means that CO2 cannot be responsible for climate change at decadal/centennial timescales like anthropogenic climate change. I find this unconvincing; one of their cited papers, Rothman 2002, states that "Superficially, this observation would seem to imply that pCO2 does not exert dominant control on Earth’s climate at time scales greater than about 10 My. A wealth of evidence, however, suggests that pCO2 exerts at least some control [see Crowley and Berner (30) for a recent review]. Fig. 4 cannot by itself refute this assumption." And that's for very long timescales. I don't think this is relevant to AGW timescales, or even necessarily Milankovitch timescales. I notice that the authors do not cite Shakun et al. 2012 on the influence of CO2 on hemispheric and global average temperature during Termination 1, and that's a pity. Shakun et al. argue that CO2 does play a major role in driving temperature increase at that time, in the Northern Hemisphere and globally, contrary to the authors' assertions about CO2. I have no problems with authors trying to find natural climate cycles that may be contributing to modern climate change, I think that's great, but trying to argue that this cycle is responsible for all of climate change since 1880 and throwing out all of AGW based on data with, like, million year resolution strikes me as waaaaaaaay overselling things.
The identification of the AAO with the Southern Oscillation is strange - the Southern Oscillation works in a 2-7 year band, not centennial. If the authors want to argue that the 140-something-year AAO modulates the SO, they can talk about Abrams et al., or cite to paleo-ENSO work by Kim Cobb et al. to try to show variations in ENSO over the last millennium to match up to their AAO cycles? And anyway, that's modulating and teleconnecting to the SO, not *being* the SO. I therefore question quoting Halpert and Ropelewski's words about the SO and subsituting [AAO] (lines 1378-1380) for SO. They weren't talking about the AAO.
When the authors mention teleconnections to the North Atlantic as shown by North Atlantic sediment cores, they first cite to a paper about Patagonia (line 82) and then (line 1384) to a paper about Southern Chile, neither of which strikes me as North Atlantic.
Author Response
Point-by-point responses to Reviewer 1
Reviewer: This paper attempts to argue that an atmospheric oscillation mode in the Southern Hemisphere, here termed the Antarctic Oscillation/Antarctic Circumpolar [D: sic; the term is Centennial, not Circumpolar] Oscillation, can be tracked through 11 Antarctic ice cores and that the offsets in timing of this cycle can be used to understand Southern Ocean regional climate and global climate through teleconnections.
The paper is incredibly long and dense and pretty much unreadable. This isn't a knock on their grammar or anything like that, which is frankly the best I've seen in a paper I've gotten to review.
Davis et al.: We appreciate the reviewer's kind words about the grammar of the paper. We acknowledge that this paper is longer than usual. The length of the paper arises because the topic is complex and transdisciplinary, and the results do not parcel coherently into smaller packages.
That the paper is "pretty much unreadable" despite the best grammar this reviewer has seen is definitely a source of concern to us. We have pored through the paper to find ways to improve and shorten the presentation, and implemented as many as we could. We hope our efforts have made the paper more readable.
R: I just can't imagine anyone but people who've spent a couple of decades doing statistics and atmospheric science being able to make it more than a few pages in.
D: We can only plead that the statistical methods are the oldest and simplest available (regression and correlation analysis) or standard in the field of climate science (spectral analysis, time series analysis). These methods are, in our view, straightforward compared to the techniques that are sometimes used in climate science, such as Wavelet Analysis (which is based on the Fourier transform), Empirical Mode Decomposition, Hilbert-Huang transform, etc. etc.
We acknowledge that our paper does span a broad reach of oceanic and atmospheric science, but such a transdisciplinary approach is unavoidable and essential to understand the Antarctic Centennial Oscillation (ACO) at the depth we seek in this paper. We can only suggest that readers either take our word for the representations and interpretations of transdisciplinary science, or confirm for themselves the accuracy of our representations and interpretations by looking up and studying the cited references, as we have done.
R: That's certainly one way to go, but if you ever want a student to read this paper all the way through, I don't think it's going to happen.
D: We definitely hope that students will be able to read this paper all the way through. We respect the reviewer's opinion on the matter, appreciate the feedback, and can only hope that our extensive revisions have made the paper more comprehensible even at the student level.
We do feel that students will be incentivized to read this paper, because understanding the ACO/AAO is central to contemporary global climate. Additionally, this paper contains a plethora of prospective Ph.D. dissertations, a veritable treasure chest for graduate and postdoctoral students looking for a project. We believe that with the data included in the SM and elementary spreadsheet competence, the average graduate student will be able not only to understand the paper, but also replicate its findings in a few weeks of effort.
R: The authors argue that they can compare asynchronous ACO events at different sites by using "signpost" events associated with major transitions like the start of the deglaciation, so that the same diachronous event can be identified in different ice cores.
D: We are not certain that the reviewer fully appreciates our methodology. Please note that we do not define "signpost events," but rather "signpost [ACO] cycles" that are quantitatively associated in time with consensually-identified climate events.
Our rationale for this methodology, which is detailed in our previous paper on the topic (reference 11), is simply that any identified (numbered) centennial cycle that occurs with fixed temporal relationship to a consensually-accepted climate event in different paleoclimate records is the same ACO cycle (a "homolog"). This methodology has been well-vetted by previous peer review and endorsed by the editors of Climate by virtue of the earlier paper pioneering this methodology and published in this journal. Re-adjudication of a peer-reviewed methodology that has been published previously in Climate and therefore endorsed by the editors seems to us unnecessary and inappropriate.
R: I confess that I don't see obvious similarities between the signpost events at Law Dome and Vostok in figure 2 that would convince me that yes, this is the same exact cycle happening at a thousand year lag in Vostok.
D: The similarity is clear to us, as noted in the paper. The corresponding identified signpost cycles occur in the different records at exactly the same temporal position in relation to known climate events. Please have a look also at Figure 4, where every ACO cycle over the last 21000 y is labeled across each of the 11 paleoclimate records used in this study.
R: Cycle 63-62 in Figure 2 at Law Dome, for instance - why identify it at 14-8-15 ka when there's another little dip right before it at 15-15.2 ka or so that immediately follows an enrichment, as the (teeny) proposed 63-62 event at Vostok does?
D: The basis for assigning cycles 63-62 as signpost cycles is straightforward: these identified ACO cycles occur in all eleven paleoclimate records we evaluated precisely at the start of the Antarctic Cold Reversal (ACR). The ACR is in this case the independent climate event that enables unambiguous identification of these ACO signpost cycles as homologs.
This is clear to us in Figure 2 referenced by the reviewer. There ACO cycles 63-62 at Law Dome occur exactly at the beginning of the Antarctic Cold Reversal (ACR), while cycles 63-62 at Vostok also occur exactly at the beginning of the ACR. It is the ACR that is the "independent climate event" associated with cycles 63-62 at both Law Dome and Vostok, and in every other Antarctic paleoclimate record we analyzed (Figures 3, 4 c).
The reviewer describes cycles 63 and 62 in the Vostok record as "teeny." We respectfully refer the reviewer to our earlier paper (reference 11) for detailed description of the size (amplitude) criteria used to categorize ACOs and for validation of the methods used in this paper, which are identical to those used previously, as indicated throughout the present paper.
R: Also, who's to say that the deglacial warming at one site didn't start slightly behind another one, making that not a useful way to lock timelines from different cores?
D: It appears that the reviewer may here be misinterpreting this central aspect of our results. Warming at any particular site does start at a very (not "slightly") different time from warming at any other site, up to a millennium, as shown quantitatively in Figures 2 and 3. That is the point of Figure 2 and is reflected both in our earlier paper (reference 11, including particularly the Supplementary Materials) and throughout this paper.
R: I also don't see much discussion of the resolution of the data at different times or different sites:
D: We respectfully refer the reviewer to our Methods section, and to the Supplementary Materials (SM) of reference 11 which is cited in the Methods. As confirmed in both places, these issues have been dealt with exhaustively. For details we refer readers, including this reviewer, to the sources we cite repeatedly in the present paper, which quantify sampling resolution and relate it to the Nyquist frequency for ACO cycles. This is covered thoroughly in the cited reference 11 and the associated SM.
R: looking at the Vostok data table the authors cite, it looks like the sampling resolution of the deuterium data is in the 80-90 year range for the LGM and last deglaciation, which is lower than the Nyquist frequency for a 143-year cycle.
D: Kindly consult our analysis of sampling resolution as detailed in reference 11 and the associated SM. In particular, the issue of artifact from frequency aliasing (Nyquist error) has been dealt with exhaustively for the Vostok data in those Supplementary Materials. The resolution is generally twice what this reviewer has estimated, as shown in Figure 2 of the referenced SM, and well above the Nyquist frequency across the full timespan of records analyzed. Please see Figure 2 c of the SM to reference 11, which answers this concern about Nyquist frequency. In the interest of space in an already long paper, we do not think it is advisable to repeat what has been published previously on this matter.
The sampling resolution is at least twice the 80-90 years interpreted by the reviewer. The issue of sampling resolution is fully discussed and validated in the SM that accompanies our previous paper (reference 11, Supplementary Materials), which is in turn referenced and cited in this paper. In particular, the Nyquist frequency is well below the sampling frequency across the full range of Vostok paleoclimate data for this time period, and therefore frequency aliasing is excluded rigorously from this analysis. Kindly revisit the paper cited (reference 11) and its Supplementary Materials for confirmation.
R: The stated motivation for this work is problematic: the introduction argues that because previous work to reconstruct CO2 levels in the atmosphere over hundreds of millions of years, with many uncertainties in the proxies at that age remove, hasn't shown a correspondence between CO2 and temperature at these extremely long timescales therefore means that CO2 cannot be responsible for climate change at decadal/centennial timescales like anthropogenic climate change.
D: We do not wish, nor is it appropriate or necessary, to re-adjudicate our previous Climate paper on atmospheric carbon dioxide. That paper has been peer-reviewed, and endorsed implicitly by the editors of Climate.
However, neither do we wish to distract attention from the main purposes of this paper by over-emphasizing a secondary motivation. Our primary purposes in conducting this study are, as stated, to establish the internal dynamics of the ACO and compare the ACO quantitatively to the Antarctic Oscillation.
Therefore we have revised this section of the Introduction and Motivation extensively, eliminating all references to atmospheric carbon dioxide that the reviewer found problematic. We hope and trust these changes will satisfy this reviewer on this point.
R: I find this unconvincing; one of their cited papers, Rothman 2002, states that "Superficially, this observation would seem to imply that pCO2 does not exert dominant control on Earth’s climate at time scales greater than about 10 My. A wealth of evidence, however, suggests that pCO2 exerts at least some control [see Crowley and Berner (30) for a recent review]. Fig. 4 cannot by itself refute this assumption."
D: Acknowledged. This is addressed by the deletion of most of the objectionable material.
R: I notice that the authors do not cite Shakun et al. 2012 on the influence of CO2 on hemispheric and global average temperature during Termination 1, and that's a pity.
D: Acknowledged. This is addressed by the deletion of most of the objectionable material.
R: Shakun et al. argue that CO2 does play a major role in driving temperature increase at that time, in the Northern Hemisphere and globally, contrary to the authors' assertions about CO2.
D: Acknowledged. This is addressed by the deletion of most of the objectionable material.
R: I have no problems with authors trying to find natural climate cycles that may be contributing to modern climate change, I think that's great, but trying to argue that this cycle is responsible for all of climate change since 1880 and throwing out all of AGW based on data with, like, million year resolution strikes me as waaaaaaaay overselling things.
D: Just to be clear, we have no desire at all to "sell" anything to anyone. Our sole purpose is to present evidence for and against worthy hypotheses so that we can better understand the natural phenomena at play here and convey that understanding to our readers.
We acknowledge the reviewer's difficulty with these points, however, and have eliminated all of the problematic material in our major revision of this section. This renders the reviewer's critique of the carbon dioxide paper moot.
R: The identification of the AAO with the Southern Oscillation is strange - the Southern Oscillation works in a 2-7 year band, not centennial.
D: The reviewer is correct. That claim was based on our misinterpretation of Halpert and Ropelewski, which we have corrected. We are grateful to this reviewer, for pointing out this mistake.
R: If the authors want to argue that the 140-something-year AAO modulates the SO, they can talk about Abrams et al., or cite to paleo-ENSO work by Kim Cobb et al. to try to show variations in ENSO over the last millennium to match up to their AAO cycles? And anyway, that's modulating and teleconnecting to the SO, not *being* the SO. I therefore question quoting Halpert and Ropelewski's words about the SO and substituting [AAO] (lines 1378-1380) for SO. They weren't talking about the AAO.
D: We agree and apologize for the error, which is now corrected. We are grateful to this reviewer for catching it.
R: When the authors mention teleconnections to the North Atlantic as shown by North Atlantic sediment cores, they first cite to a paper about Patagonia (line 82) and then (line 1384) to a paper about Southern Chile, neither of which strikes me as North Atlantic.
D: The reference to the North Atlantic Core is contained within the references to the Moreno et al. paper. This paper includes data on the NA core, in the form of Fe concentration, which we re-digitized and evaluated as part of Table 2. We have revised the text to make this clear.
In summary, this reviewer has provided an invaluable perspective and pointed out significant distractions as well as factual and interpretational errors. We have made extensive revisions to accommodate the comments of this reviewer, and corrected the errors. We thank this reviewer most sincerely for the careful review of our work.
Reviewer 2 Report
"Origin and propagation of the Antarctic centennial oscillation" by W. Jackson Davis et al. for Climate - MDPI discusses the evolution of the ACO based on paleo-climate data collected at different sites around the Antarctic continent.
The methods appear to be very similar to those presented by the authors on a previous paper on the same journal.
In figure 5 the robustness and the meaningfulness of the trends should be further verified. Data clustered and derivation of linear trend is not physically justified. The scatter does probably not warrant a robust statistical fit. The same applies to Fig 11, 15.
In Fig 20 the upwelling is attributed to a region off the Australian Bight and the authors claim that this is the region of the strongest winds. The region they highlight is mostly between 30 and 45, not in the furious fifties. Moreover, the area of strongest winds is circumpolar and not localised to the South West of Australia.
In the conclusion and hypotheses the authors make strong claims, some of which are not fully supported by their analysis. I argue that the analysis is based on biased statistical regressions and some are purely descriptive.
I also dislike the Future Research section in a manuscript.
I do not believe the results are fully supported by the analysis, and suggest rejection
Author Response
Point-by-point responses to Reviewer 2
Reviewer: "Origin and propagation of the Antarctic centennial oscillation" by W. Jackson Davis et al. for Climate - MDPI discusses the evolution of the ACO based on paleo-climate data collected at different sites around the Antarctic continent.
The methods appear to be very similar to those presented by the authors on a previous paper on the same journal.
Davis et al.: As stated throughout this paper, and in response to this comment now re-emphasized repeatedly, the methods and rationale are not "similar," but identical to those we used in our previous paper. We have simply extended the analysis in this paper from four drill sites synchronized on the AICC2012 chronology (reference 11) to 11 drill sites including more than one well-verified chronology (this paper), with the result that we are able to significantly extend our earlier conclusions.
R: In figure 5 the robustness and the meaningfulness of the trends should be further verified. Data clustered and derivation of linear trend is not physically justified. The scatter does probably not warrant a robust statistical fit. The same applies to Fig 11, 15.
The reviewer is correct that a linear trend in some of these graphs is difficult to establish with such a small sample size, and especially in the few cases of clustered datapoints. Moreover, the parametric Pearson correlation analysis that we used indeed assumes linearity, which we did not test. The Pearson correlation coefficient also assumes independence of observations, normal distribution, and equal population variance, likewise untested. The limited number of drill sites available to this analysis (maximum of 11) means that the sample size is unavoidably small, but still within the bounds considered by statisticians to be acceptable for correlation analysis (statisticians say the acceptable criterion n > 6, we use n = 9 as a limit criterion). We would respectfully note that the same approach was published in 2000 by Masson et al. (reference 40), with similar results, as referenced in our paper.
Given the reviewer's reservations, however, the best resolution may be to perform a non-parametric correlation analysis on these same data, using the Spearman Rho correlation coefficient. We have done so, and obtained identical conclusions to the parametric Pearson correlation coefficient, as now reported in the caption to Figure 5. Note that the Spearman Rho does not require the underlying assumptions of the Pearson, and so the findings are distribution-free.
The referenced figures contain, in the plot area, the Pearson correlation coefficient (r), the probability of Type I error (alpha, or p), and the sample size (n, the number of data points), which enables any reader to re-compute the p value reported. The finding that many p values are highly significant despite the small sample size, confirmed now by the non-parametric analysis, gives us confidence that the trends demonstrated by the data, whether or not rigorously linear, are statistically discernible.
R: In Fig 20 the upwelling is attributed to a region off the Australian Bight and the authors claim that this is the region of the strongest winds. The region they highlight is mostly between 30 and 45, not in the furious fifties. Moreover, the area of strongest winds is circumpolar and not localised to the South West of Australia.
D: In response to this comment, we re-examined our sources on the issue of winds in the Southern Ocean, namely the oceanographic textbook by Tomczak and Godfrey (our reference 95). Those authors describe the strongest oceanic wind stress across any body of water on Earth as occurring between 40-60o S and 30-120o E (their figure 1.4). We have repositioned the upwelling zone in Figure 20 to correspond precisely to these coordinates. This revision significantly improves the presentation. We are grateful to this reviewer for this comment.
We note further, in response to this reviewer's comment, that the circumpolar winds that are part of this Southern Ocean wind regime vary systematically with longitude and latitude, and the winds are strongest by a factor of 2-10 in the aforementioned region now highlighted accurately in Figure 20. The area of strong winds is indeed circumpolar, but the strongest wind stress in the Southern Ocean by an order of magnitude is localized to the area now shown in Figure 20. We urge the reviewer to confirm this by consulting Tomczak and Godfrey, Figure 1.4, which is available online.
The reviewer is correct, however, that neither are the strongest winds located in the Australian Bight as shown in previous Figure 20. We have corrected the position of strongest winds in the revised Figure 20 and thank the reviewer for pointing out this error.
Incidentally, the position of strongest winds is easily confirmed in real time by visiting the fascinating online NullSchool global wind and current map, which typically shows the strongest winds in the region noted above and now correctly identified in Figure 20.
R: In the conclusion and hypotheses the authors make strong claims, some of which are not fully supported by their analysis.
D: We agree that of the 12 conclusions made, some are more direct and rigorous in their support than others. One way to handle this variance is to include the phrase "with various degrees of confidence" so that the introductory sentence to the section now reads: "The empirical evidence reported in this paper supports, with various degrees of confidence, twelve conclusions about the ACO."
R: I argue that the analysis is based on biased statistical regressions and some are purely descriptive.
D: We have never encountered the phrase "biased statistical regressions, and do not know what it means in this context. This reviewer has not defined the term. The data used throughout this study, including Figures 5, 11 and 15 cited by the reviewer, are accepted consensually in climate science. They cannot be fairly characterized as "biased," and therefore we cannot see how their distribution can be considered biased.
Indeed, the data we use is one of the major strengths of our study. We use data collected and developed by previous investigators, particularly ice core data but also geophysical data as used in Figures 5, 11 and 15, that are accepted by consensus in numerous climate science studies we cite. These data have been used in dozens, probably hundreds, of previous studies. The data are beyond reproach.
The reviewer describes our study as sometimes "purely descriptive," presumably as opposed to "analytic." First, we don't see why that would be problematic. Descriptive findings are common and often essential to subsequent quantitative analysis. Indeed, sound descriptive analyses are central to almost every scientific discipline. They are the observations on which hypotheses are based, and all quantitative analysis is anchored first on descriptive reports. Would Milankovitch have bothered to analyze orbital cycles without earlier published descriptive reports of regularized horizontal layers in sedimentary rock?
Second, the assertion that some of the findings reported in this paper are purely descriptive appears inaccurate to us. Of 20 figures presented in the paper, 17 are purely analytic and quantitative, while the remaining three (Figure 3, 6, 20) are partially descriptive but based ultimately on quantitative data. Even the descriptive elements of these three figures are based on quantitative data, i.e. the cartography in Figures 6 and 20.
If the data are acceptable, which they most certainly are in this case as the corresponding data are universally used in paleoclimate science, we don't understand how their distribution could be considered "biased." Other than Figure 5 and the related Figures 11 and 15, addressed above, the reviewer does not indicate which part of our analysis he/she believes are based on "biased statistical regressions."
Therefore the claim cannot be addressed without further elaboration by the reviewer.
R: I also dislike the Future Research section in a manuscript.
D: We appreciate that opinion, and such a section is indeed atypical for original scientific papers, though not unprecedented. Indeed some journals encourage authors to elaborate on future research, inasmuch as authors are in the best position to see future possibilities.
We include the section on future research in this paper mainly because we know that this paper opens a large field of future research on a seminal climate phenomenon, the Antarctic Centennial Oscillation, and we also know also that we shall not revisit this research territory. Our intent is providing this section this section to give future researchers of the Antarctic Centennial Oscillation a roadmap paved with insights we have gained in our discovery and initial elaboration of this natural climate cycle. On balance, we think the field benefits from inclusion of this section.
R: I do not believe the results are fully supported by the analysis, and suggest rejection.
D: This is a puzzling claim, inasmuch as to our understanding the results are the analysis. The terms are synonymous; therefore, how could they be anything but mutually supportive? We will proceed under the assumption that the reviewer means that the conclusions are not fully supported by the analysis.
This reviewer identifies only a single case in which he/she thinks the conclusion is based on biased regression and therefore not supported by the analysis," namely Figure 5 (and similar Figures 11 and 13). We have answered that criticism above and have made appropriate revisions as detailed above.
In this case, the criticism that the conclusions are not supported by the analysis cannot be taken seriously until the reviewer states explicitly which conclusions he/she thinks are problematic. It is obviously impossible to address the reviewer's concerns unless they are stated. If the reviewer wishes to elaborate which conclusion he/she thinks is not supported by our analysis, we will be more than happy to address the concern.
We thank this reviewer for his/her comments, which have helped to improve the presentation.
Reviewer 3 Report
A general observation is the systematic omission of the article "the" throughout the text. One example is at lines 301-302: "The finding that spectral density peaks...shows that temperature-proxy cycles oscillate..."; this could read as "The finding that the spectral density peaks...shows that the temperature-proxy cycle oscillates...".
Another observation is at line 42: "...135-150 million years (y)..."; this could read as: "...135-150 million years (My)...". Therefore, wherever in the text there appear "million years", they can be symbolized as My.
A third observation lies in line 252: "...0.05>p<0.10..." should be written "...0.05<p<0.10...". Also, at line 301: +-<3.6% should be written as <+-3.6%.
In Figure 1, the x-axis (period in years) does not match the period 21-9 kyb1950 mentioned in its legend.
Finally, the conclusion in 1081 that "the ACO encompasses all of Antarctica" is a little bit dangerous for such a generalization.
Author Response
Point-by-point response to Reviewer 3
Reviewer: A general observation is the systematic omission of the article "the" throughout the text. One example is at lines 301-302: "The finding that spectral density peaks...shows that temperature-proxy cycles oscillate..."; this could read as "The finding that the spectral density peaks...shows that the temperature-proxy cycle oscillates...".
Davis et al.: The topic of articles in English grammar, and specifically when to use or omit them, is complex and subtle. Although all three authors of this paper are native English speakers, we are still not sure we understand the article rule. Strunk and White, in their iconic grammar guide "Elements of Style," omit the topic entirely, perhaps wisely.
According to several websites on English grammar, the "zero article" rule (omission of an article before a noun or noun phrase) applies when addressing things in general, as opposed to a specific example, or with non-countable nouns, in front of proper nouns (names of people or things), etc. We refer the reviewer to a URL we consulted: http://www.icaltefl.com/zero-article-in-english-grammar.
We have combed through the paper, line by line, and applied these article rules as best we are able, starting with the title of the paper (now preceded by the article "The") and hope the effort suffices. In our understanding of the zero-article rule, the sentence cited by the reviewer above is appropriately unchanged because "spectral-density peaks" and "temperature proxy oscillations" correspond to "things in general."
R: Another observation is at line 42: "...135-150 million years (y)..."; this could read as: "...135-150 million years (My)...". Therefore, wherever in the text there appear "million years", they can be symbolized as My.
D: We agree, and made this change. Thank you.
R: A third observation lies in line 252: "...0.05>p<0.10..." should be written "...0.05<p<0.10...".
D: Thanks also for pointing out that typo, which we fixed.
R: Also, at line 301: +-<3.6% should be written as <+-3.6%.
D: Agreed, thank you. We fixed it.
R: In Figure 1, the x-axis (period in years) does not match the period 21-9 kyb1950 mentioned in its legend.
D: That's because the x-axis portrays the frequency of dominant oscillations shown over the period 21-9 Kybp, not their time of occurrence. Spectral analysis is of course devoid of timing information, and limited only to frequency.
R: Finally, the conclusion in 1081 that "the ACO encompasses all of Antarctica" is a little bit dangerous for such a generalization.
D: We had thought that documenting the same ACO cycle at each of eleven drill sites distributed widely all over Antarctica in vastly different environments (coastal and inland, sea level and mountainous) comprised robust evidence that the ACO encompasses all of Antarctica. The only way this could not be true would be if this climate cycle appeared only at drill sites, but not at intervening locations, which seems physically implausible.
However, out of respect to this reviewer's concerns, we have used more temperate language to connect the evidence with the conclusion. We revised the abstract, for example, to read: "Spectral and time series analyses reveal that ACOs occurred at al 11 sites over all time periods evaluated, indicating suggesting that the ACO envelops all of Antarctica." We do agree that this is a more cautious statement of this finding, and we are happy to error on the side of caution.
We made comparable changes elsewhere in the paper where this particular finding is discussed.
We thank this reviewer for his/her constructive comments, which have significantly improved the presentation.
Reviewer 4 Report
Comments on “Origin and Propagation of the Antarctic Centennial Oscillation” by Davis et al.
Based on stable water isotopes frozen in ice cores from 11 Antarctic drill sites, this study discussed the internal climate dynamics of Antarctic Centennial Oscillation over the last 21 millennia. It is revealed that the Antarctic Centennial Oscillation cycles propagated on a multi-centennial timescale from the Last Glacial Maximum to the Last Glacial Termination. The route was clockwise around Antarctica in the streamline of the Antarctic Circumpolar Current, originating from the East Antarctic coastline. The potential mechanisms of teleconnection were further investigated. The results of this study are interesting and explained the origin of Antarctic Centennial Oscillation and Antarctic Oscillation. The datasets used in this study are reliable, the method adopted by this study is objective and reasonable, and the manuscript are well organized. Therefore, I recommend acceptance of this paper after some minor revisions.
1. Lines 54-55: Antarctic Oscillation and Southern Oscillation are two different climate systems, how could the Southern Oscillation Index be used to quantify the Antarctic Oscillation?
2. Line 75: “… AAO – then termed the Southern Oscillation ….”, the Southern Oscillation describe the fluctuation of opposed pressure anomalies in the eastern and western hemispheres (Troup 1965: The ‘southern oscillation’), however, the AAO describe the fluctuation of atmospheric mass between the mid-latitudes and the Antarctic (Gong and Wang, 1999: Definition of Antarctic oscillation index). How could it be possible that the AAO was then termed the Southern Oscillation?
3. Line 318: A long term time lag (1040 y) was identified between the ACO cycles at Vostok and their homologs at LD. Could authors give more discussion on the physical explanations?
4. Line 906, “teleconnection velocity” appears many times in this study, what is the definition of teleconnection velocity and how to quantify it?
Author Response
Point-by-point response to Reviewer # 4
Comments on “Origin and Propagation of the Antarctic Centennial Oscillation” by Davis et al.
Reviewer: Based on stable water isotopes frozen in ice cores from 11 Antarctic drill sites, this study discussed the internal climate dynamics of Antarctic Centennial Oscillation over the last 21 millennia. It is revealed that the Antarctic Centennial Oscillation cycles propagated on a multi-centennial timescale from the Last Glacial Maximum to the Last Glacial Termination. The route was clockwise around Antarctica in the streamline of the Antarctic Circumpolar Current, originating from the East Antarctic coastline. The potential mechanisms of teleconnection were further investigated. The results of this study are interesting and explained the origin of Antarctic Centennial Oscillation and Antarctic Oscillation. The datasets used in this study are reliable, the method adopted by this study is objective and reasonable, and the manuscript are well organized. Therefore, I recommend acceptance of this paper after some minor revisions.
Davis et al.: We very much appreciate the constructive and helpful comments.
R: 1. Lines 54-55: Antarctic Oscillation and Southern Oscillation are two different climate systems, how could the Southern Oscillation Index be used to quantify the Antarctic Oscillation?
D: The reviewer is correct. We have rectified the error throughout the paper. Thank you very much for pointing it out.
R: 2. Line 75: “… AAO – then termed the Southern Oscillation ….”, the Southern Oscillation describe the fluctuation of opposed pressure anomalies in the eastern and western hemispheres (Troup 1965: The ‘southern oscillation’), however, the AAO describe the fluctuation of atmospheric mass between the mid-latitudes and the Antarctic (Gong and Wang, 1999: Definition of Antarctic oscillation index). How could it be possible that the AAO was then termed the Southern Oscillation?
D: Again, our mistake. We are grateful that this reviewer caught it. We have corrected the error throughout the paper.
R: 3. Line 318: A long term time lag (1040 y) was identified between the ACO cycles at Vostok and their homologs at LD. Could authors give more discussion on the physical explanations?
D: We intended section 4.2, second paragraph, to provide a physical explanation. We made significant revisions and adjustments to this section which we hope improve it to the extent that the physical explanations as we understand them are clear.
R: 4. Line 906, “teleconnection velocity” appears many times in this study, what is the definition of teleconnection velocity and how to quantify it?
D: We define it now in the Methods section right before section 2.4 as "the speed of movement of the advancing ACO wavefront). It can be calculated with the equation velocity = distance/time, where time = latency, as now stated explicitly in the text. We continue to use latency as an inverse proxy of teleconnection velocity over a fixed distance, as now also explicitly stated in the text.
We are deeply grateful to this reviewer for his/her cogent and constructive comments and excellent suggestions, most of which have been implemented.
Round 2
Reviewer 1 Report
The authors have clearly put a lot of work into improving the manuscript, which I appreciate.
Reviewer 2 Report
The authors went through great lengths to address issues raised on the first review.
I believe their answers and the revised manuscript to be satisfactory for publication.
I encourage to carefully check the formatting on the submitted manuscript.
