date: Thu, 29 Nov 2007 11:49:28 +0100 from: Bo Vinther subject: [Fwd: 2007GL032450 (Editor - Fabio Florindo): Decision Letter] to: Tom Melvin , Keith Briffa , Phil Jones , Lars Berg Larsen , "Henrik B. Clausen" , cuh@gfy.ku.dk, Marie-Louise Siggaard-Andersen , Rashit Hantemirov , Kurt Nicolussi , Bjorn.Gunnarson@natgeo.su.se, Håkan Grudd , Matti.Eronen@helsinki.fi, mukhtar@forest.akadem.ru Dear Colleagues Good news: Our paper on the A.D. 536 event has now been reviewed an we have been asked to submit a response to the comments made by the reviewers and a revised paper by December 12. I will circulate drafts before the weekend - so you will be able to have a weeks time to come with suggestions/additions..... The reviews can be seen below - but let me just correct one misunderstanding in the review: Reviewer #2 makes a lengthy review partly focused on discrediting Greenland ice core dating - claiming that the dating uncertainties are "guesstimates"! A cornerstone in his argument is that a volc. signal previously reported to have been dated to A.D. 572 now has been shifted to A.D. 567/68 - this is, however, not the case as the A.D. 572 signal and the A.D. 567/68 signal are two separate volc. signals in the ice core (A.D. 572 is a small signal containing significant amounts of fluoride and chloride - suggesting a local eruption - most probably Icelandic). Anyway I just wanted to point out that the fact that the A.D. 572 and A.D. 567/68 have nothing to do with each other mutes the claim of reviewer #2 with respect to uncertainties in ice core dating....and therefore we have no scientific basis for moving the Greenland ice core dating 7 years as suggested by reviewer #2. Cheers......Bo -------- Original Message -------- Subject: 2007GL032450 (Editor - Fabio Florindo): Decision Letter Date: Wed, 28 Nov 2007 20:47:01 UT From: [1]grlonline@agu.org Reply-To: [2]grlonline@agu.org To: [3]bo@gfy.ku.dk Dear Dr. Vinther: Your paper, 2007GL032450, entitled "New Ice Core Evidence for a Volcanic Cause of the A.D. 536 Dust-veil" will require moderate revision before publication in Geophysical Research Letters. Attached below are the review comments. We will need to receive the following: (1) a Response to Reviewer letter addressing all of the identified problems and (2) a copy of the manuscript with the changes noted (e.g., highlighted, "track changes," italics or bold changes) [PLEASE BE SURE TO UPLOAD YOUR ARTICLE WITH TRACKED/HIGHLIGHTED CHANGES AS A RESPONSE TO REVIEWER FILE] and (3) a copy of the manuscript with same changes incorporated by December 12, 2007. While you are preparing your files, keep in mind that publication is contingent upon your preparation of a publication-ready version that corrects these and meets GRL's length restrictions. GRL has a strict length requirement. For information, please see our Web site: [4]http://www.agu.org/cgi-bin/wc/length_tool.cgi Articles that exceed the length requirement cannot be processed. The Editor will review these, in coordination with the reviewers. Please consult the reviewers' comments (below) for more information. When you are ready to submit your revision, please use the link below. [5]http://grl-submit.agu.org/cgi-bin/main.plex?el=A2K6DcLf5A3BHNB4I6A9zUPbdtKao6DloCwqjSfaA Z I look forward to receiving your revised manuscript. Sincerely, Fabio Florindo Editor, Geophysical Research Letters ------------------------------------------------------------------------------ Reviewer Comments Reviewer #1 Evaluations: Science Category: Science Category 2 Presentation Category: Presentation Category A Annotated Manuscript: No Anonymous: Yes Referrals: No Highlight: No Reviewer #1 (Formal Review): Comments to the paper "New Ice Core Evidence for a Volcanic Cause of the A.D. 536 Dust-veil" of Larsen and co-authors. The paper regards the interpretation of the depositions of volcanic material on polar ice sheets in a climatic context. Interesting the authors compare data from ice cores drilled in both hemispheres and temperature reconstructions from tree rings in a common time scale, in order to link temperature anomalies and volcanic emissions. Considerations about the actual climatic impact of volcanic emissions not only are addressed to the case of the 536 A.D. event, but useful notes also regard other events, providing precious information toward the understanding of the actual climatic significance of volcanic depositions in ice cores. The data sets used and/or produced by the authors are highly reliable, as well as the synchronization between glaciological time-series from Greenland and Antarctica. Similarly, most of the conclusions inferred from the data provide interesting and strong evidences for people interested in the climatic impact of volcanic activity, i.e. for people working in different scientific fields: glaciology, tree rings studies, climate modelling, atmospheric processes, dating of paleo-records. The paper is well written, following a clear scheme that helps in understanding the conclusions proposed by the authors. For all these reasons the paper deserves to be published on GRL, even if in some sections there are some aspect that in my opinion should be discussed and/or considered in a different optic in order to increase the scientific value of the paper. I suggest the authors to consider the comments here below listed, modify, in case, the text accordingly and/or discuss the points here below raised. Detailed comments: Lines 28-30: this sentence is not exact since sulfate data from ice cores here presented can't indicate the existence of a dust veil, but only of an acidic veil: eliminate the word "dust" in line 30 or re-word the sentence. Lines 81-83: it looks to me that, while the mean of the concentrations of the GRIP, NGRIP and Dye-3 signatures of the 529 events is quite similar to the mean of the two 17th century Haruna events, the single depositions do not show a clearly similar shape, except for the GRIP signature. In any case, all this discussion is in my opinion too qualitative and descriptive and an explanation of why three different eruptions from a same source should show this similar pulsed shape, is missing. Are indications of multiple eruptions for the two events historically known presented in Figure 3c present in literature? Alternatively, is it possible to imagine mechanisms of atmospheric transport at the base of this common shape? Line 83: please specify in the text where the events presented in Figure 3c are measured (NGRIP, I think). Line 91: this sentence is really misleading and should be deleted. At least the signatures assigned to the Haruna eruptions show a very typical pulsed shape, but the other events (533-4 and Tambora) do not show nothing particular that permit to define their shape "very similar". Line 114: The peak in the EDC ice core corresponding to the 542 event is listed in Severi et al. 2007 and Ruth et al. 2007 supplementary data, but not in the table provided by Castellano et al., 2005, JGR. This means that there is a small signal, but below the threshold between volcanic and non volcanic signals. The authors should mention that actually the signal is not detected in the EDC ice core. This of course could/should be due to loss of the signal at Dome C, likely because of post-depositional effects. Line 120: Interestingly the two couples of events lined up after the synchronization proposed by the authors are well defined in all the cores (the shallow cores drilled at DML and the two EPICA cores) and show quite high deposition fluxes (you can associate the events in the different cores by comparing the tables presented in Traufetter et al, 2004, Ruth et al., 2007 supplementary data and Castellano et al., 2005), close to the Tambora flux. A very interesting point to rise here is that these two events, very likely produced by tropical eruptions (considering that their signatures are recorded in both hemispheres) did not produce any significant effect on Northern Hemisphere climate unlike the 536, the Tambora and the 1601 event. This point should be here emphasized, discussed or commented even if rises complications to the interpretation of the climatic impact of volcanic eruptions, and is, in my opinion, an important point to consider. Of course it still remains the possibilities of two couples of synchronous volcanic events, i.e. four distinct eruptions, occurred in the two hemispheres, but, as the authors suggest, it is hard to believe that this good accord between the two records is coincidental. Line 133: I do not think that a difference of only 400 km between the two sources can explain a so marked difference in deposition at the two poles, i.e. data provided by the authors cannot be used to discriminate between two sources located relatively so close in space. Stochastic processes of atmospheric transport of single event's clouds could dramatically affect depositions. This not means that the association of the 536 signal to the Rabaul eruption is not plausible. Reviewer #2 Evaluations: Science Category: Science Category 1 Presentation Category: Presentation Category B Annotated Manuscript: No Anonymous: No Referrals: No Highlight: Yes Reviewer #2 (Formal Review): Review of Larsen et al. Introduction No need to pussy-foot around on this review. Given that I know most of the authors there is not much point in trying to be anonymous, especially as one aspect of the paper is the suggested elimination of my "AD 540 cosmic impact" ideas. However a few words of explanation are necessary at the beginning. Recommendations are at the end. 1) The event that some people, including the current authors, refer to as the "AD 536 event" has been described by me as the "AD 540 event, i.e. the 536-545 event" because it is much more than a single event and the aftermath of a single event. It is clearly two stage (more detail below) and the real question relates to what caused the second part of the event in the years 540, 541, 542 and even 543. It is interesting that in the current paper a single cause around 535 is seen to initiate effects that last until at least 550. 2) The event appears to have been global because its effects are seen in tree-ring chronologies from the Old World and from both N and S America 3) Why was a cosmic (comet) vector ever suggested, by me, for the broader AD 540 event? It was suggested because the various ice cores (Dye3, GRIP, NGRIP and GISP2) failed to identify a clear (large and environmentally effective) volcanic acid spike in Greenland records in the 536-545 time window indicated by tree-rings. With no volcanic signal it was reasonable to suggest that the environmental effects had been caused by loading of the earth's atmosphere from space. 4) The current paper is interesting because it pits ice cores (claimed, but not proven, to be near calendrical in character) against tree-rings that are known, and proven, to be exactly dated to the calendar year. The following review looks at several aspects of the paper and suggests areas that would benefit from re-working. Some of the information may not be welcomed by the authors. Overview The current paper makes some bold claims. Re-evaluation of the three main Danish ice cores Dye3, GRIP and NGRIP across the 6th century indicates that there are three spaced volcanic eruptions at or around 529, 533.5 and 567.5. Of these, the second is claimed to have been the cause of the well-known 536 dust-veil event, and its effects are claimed to have run on until at least 550. This would of course make the 533.5+/-2 volcanic event the most environmentally effective in recent millennia. A subtext of the paper is that, by making the whole 536-550 environmental downturn a response to a single massive volcanic eruption, there is no need to allow the suggestion that the second component of the environmental downturn - that around 540-543 - might have an extraterrestrial cause, i.e. a cause due to loading of the atmosphere from space, most probably by comet debris. Main points Problem 1 The authors appear not to recognise that the focus of the extraterrestrial-loading scenario (put forward by Baillie 1999) is not AD 536 but relates to the more notable downturn in Irish oaks and other world tree-ring chronologies in the time window 540-542. For example, Baillie (cited by the authors) states: "So 536 is a year of very widespread reduced growth, but it is notably separate from the real downturn which affects the oaks at AD 540" (Baillie 1999; 66) Again, D'Arrigo et al. building on the same issue, state (source cited by the authors): "In AD 536 the Index value is 0.645, relative to the long-term mean of 1.0. Standard deviation (SD) is 0.204 over the full length of the chronology. This low growth value signals the onset of an unusually cold decade (AD 536-545) in which the mean ring-width index is 0.670 (SD 0.240), with a minimum of 0.37 in AD 543. AD 538 shows a brief recovery with an index value of 1.223. As noted, this two stage pattern is also evident in the European oak and other tree-ring series and may signify a delayed climatic response to one event (as is typical for many volcanic eruptions - e.g., Stothers (2000) or possibly two separate events (Baillie, 1994, 1999b)"(D'Arrigo et al. 2001, 241) So, not only do the authors ignore any previous comments relating to the two-stage nature of the environmental event, they ignore the evidence of their own Figure 1, where the two-stage nature is clear, especially in the lower panel. Moreover, Briffa, Grudd and others have observed this two stage nature in Fennoscandian chronologies where 541 is only slightly less bad than 536. Salzer and Hughes indicate similar clusters of narrow and frost damaged rings at 536 and in the early 540s. I do not think the two-stage nature of the downturn can realistically be ignored. Problem 2 A knock-on effect of ignoring the two-stage nature of the 536-550 event is evident in the authors' text. They are forced to make a weak statement: "...summer cold lasting from 536 to at least 550. ...These results suggest an eruption in 535 (or successive eruptions in and shortly after 535) of unparalleled magnitude in the last two millennia" (current paper page 5) Thus it is implicit in their text that their newly discovered massive eruption "around 535" gave rise to fifteen years or more of environmental effects. This leads to problem 3. Problem 3 Two of the current authors are well known as having undertaken a superposed-epoch analysis of the effects of known volcanoes wherein they concluded that the main effects of eruptions were over within a few years. (Jones PD, Moberg A, Osborn TJ, Briffa KR (2004) Surface climate responses to explosive volcanic eruptions seen in long European temperature records and mid-to-high latitude tree-ring density around the Northern Hemisphere. In: Robock A, Oppenheimer C (eds) Volcanism and earth's atmosphere. AGU Geophysical Monograph, Washington, DC, pp 239-254). Recently this has been up-dated by Fischer et al. (2007) who conclude that "The average influence of 15 major volcanic eruptions over the last 500 years is a distinct European summer cooling during two post-eruption years" (Fischer et al. Geophysical Research Letters, Vol. 34, L05707, doi:10.1029/2006GL027992, 2007). Now, in the current paper we are led to believe that a 6th century eruption can have up to 15 years of effects without any explanation as to how this is possible. Problem 4 If there were large eruptions at 529, 533.4 and 567.5 (errors are hardly relevant as the layers of acid are spaced) how is it that only the middle one has any obvious environmental effects? It seems not to worry the authors that one eruption can cause 15 years of environmental effect while two others appear to have no effects whatsoever. I'll come back to this below. Problem 5 Moving to the ice cores. The whole impact of the paper rests on the dating accuracy of the three Danish ice cores. Only if the suggested date ranges are realistic can the comparisons be made with the tree-ring records which are undoubtedly absolutely dated. So, what does 533-34 +/-2 actually mean? It means only that the ice scientists, studying the internal detail of three parallel ice cores (Dye3 GRIP and NGRIP) have satisfied themselves that their dating is almost absolute back in the 6th century AD. The main problem with this situation is that there is no independent replication. In dendrochronology the final truthing of a long chronology is when it agrees with a chronology constructed independently by independent workers. This is known as 'tertiary replication'. In the case of the ice cores the only hope of truly independent replication rested with the American GISP2 core. In this core 14 metres of ice were lost in the 6th century AD, thus no independent replication of the Greenland results is currently possible. Studies on the GRIP and GISP2 ammonium records show that these two series are in good agreement back to AD 1030; earlier than this they increasingly diverge so that by AD 700 the offset is around 19 years; the GRIP dates being older than the GISP2 dates. This means that the internal procedures used by the European ice-core workers have to be internally very robust. We now know (the current paper) that there were initially problems with the Dye 3 chronology. In 1980 Hammer et al. noted a first result from the Dye 3 core in a footnote. "Note added in proof: In a new ice core from South Greenland, a strong acidity signal was found and preliminarily dated to AD 540+/-10". In 1982 Herron published more details which included statements about a "535 acid layer". To any reader of the Herron paper this would have been interpreted as a more refined date for the 540+/-10 layer mentioned in 1980. However doubts about this could be generated from information given by Hammer in a paper in 1984. In Herron's 1982 paper a visible dust layer was given the date AD 196. Hammer (1984) mentions this same layer but with a new date, namely 174-5; in addition he does not mention 540 or 535 but instead 516. To an interested reader it could be assumed that a whole section of Dye3 had been moved back in time by around 19 years (535 to 516) or 21 years (196 to 175). The current authors confirm that Herron's AD 535 date was indeed moved back in time to AD 516 sometime between 1982 and 1984. This 19/21-year movement of a section of the Dye3 core is particularly interesting. In his 1982 paper Herron dated the 10th century Eldgja eruption in the Dye 3 core to 934. That date (variously ECM signal 932; acid signal 934) is still used in the major ice-core papers, e.g. Clausen et al. 1997. In addition, Herron pointed out an acid signal at around 40 BC. By 1984 Hammer records this date as 50 BC which is still the current dating for this acid layer. So now we can see the comment in the current paper in proper context. In Dye3 between 1982 and 1984: Acid layer at AD 932/4 remained at same date Acid layer at AD 535 moved back to AD 516 Dust layer at AD 196 moved back to AD 174/5 Acid layer at around 40 BC moved back to 49/50 BC So we now know that a major section of the Dye 3 core had dating problems somewhere between AD 932 and AD 516. The interesting thing being that by 1984 all the key (AD 932, AD 516, AD 175 and 50 BC) dates were fixed in time, purely on the basis of the Dye 3 core. The existence of the GRIP and NGRIP cores has not moved these dates. If we go to Clausen et al. 1997 the Dye 3 and GRIP date lists are padded out: Dye 3 GRIP 932 934 895 898 889 - 871 871 755 757 743 744 674 675 - 645 - 622 (in Crête core as 623) 585 - 572 572 534 532 530 527 516 514 We are interested in errors. Note that in 1997 the key 6th century ice-core dates are given as 572, 534/32, 530/27 and 516/14. In the current paper these dates are given as 529+/-2, 533-34 +/-2 and 567-568 +/-2. Let's look at these dates: Dye 3 GRIP Current paper 674 675 674/75 +/-2 572 572 567/68 +/-2 534 532 533/34 +/- 2 530 527 529 +/-2 An interested observer might ask why the replicated 572 date has become 567.5 +/-2 in a period when everything else stays the same, i.e. are these quoted errors anything other than guesstimates? This may be seen as minor quibbling but it is being used to show that the ice core dates are not 'set in stone' they are somewhat flexible. As an outside observer what intrigues me is how the very first cores got the dates so exactly correct. Subsequent cores have not really altered the dates since the time the Dye 3 error was corrected. Thus the possibility has to exist that the errors in the Dye 3 record in the 6th century AD are larger than those quoted by the authors. Because the ice-core workers do not have external ice-core replication, is there anything that might provide such replication? One possibility has to be the frost-ring record in Californian bristlecone pines. LaMarche and Hirschboeck (1984) originally raised the issue of frost rings sometimes being the result of large explosive volcanic eruptions. This work has more recently been elaborated by Salzer and Hughes (2006). Here is a list of all the AD frost rings in bristlecone pines (not narrowest rings, just frost rings) between 484 and 681, and the best available list of ice-core acid dates. Frost Ice-acid 484 514/16 522 529 532 533.5 536 567.5 541 585 574 622.5 627 645 674 674.5 681 If we re-jig this list a pattern emerges: Frost Ice-acid 522 515 (ice 7 years too old) 532 - 536 529 (ice 7 years too old) 541 533.5 (ice 7.5 years too old) 574 567.5 (ice 7.5 years too old) 627 622.5 (ice 4.5 years too old) 674 674.5 (ice correct) OR 681 674.5 (ice 6.5 years too old) So in the 6th century there are four consecutive acid layers whose spacing is exactly the same as the spacing between four bristlecone frost events...with the ice-core dates consistently 7 years too old. Given that the original problem with the Dye 3 core seems to have been in the 6th century, is it possible that the 535 to 516 move was simply larger than it should have been by about 7 years? This seems like a perfectly reasonable question, the alternative being that the similar spacings are simply 'coincidence'. Moving to European tree-rings. Pentti Zetterberg supplied me with the 6th century data for his (presumably temperature sensitive) Finnish pine chronology. It is plotted in Figure 1. Similarly Hakan Grudd and Keith Briffa supplied me with their temperature sensitive Fennoscandian pine chronology; plotted in Figure 2. It is clear that both chronologies show extreme negative growth departures in 536, and in 541/42 and in 574/75. These growth events mimic the spacing of the ice acidity layers at 529, 533/34 and 567/68 almost exactly given the quoted +/-2 year error on the ice core dates. Viz. Pines 536...5 or 6 years...541/42...33 years...574/75 Acid 529...4 or 5 years...533/34...34 years...567/68 So, as with the bristlecone pine frost ring data, temperature sensitive European pines also suggest that the ice core chronology in the 6th century be moved forward in time by around 7 years; 529 becoming 536, 533.5 becoming 540.5 and 567.5 becoming 574.5. Again, the only alternative is to invoke 'coincidence'. Fig 1 Fig 2 Conclusion: In the current paper, as presented, the authors have two major difficulties: 1) Their 529 and 567.7 acid layers produced no environmental effects, while their 533.5 acid layer had massive effects. This seems illogical. 2) Currently their 533.5 volcano has to produce 15 years of environmental effect - a decade longer than the effects of any other volcano. By moving their acid dates forward in time by seven years the effects would become: 1) consistent, in that each volcano produces clear environmental effects in both Europe and America. 2) the move could explain the long-drawn-out effects as follows. The second eruption around 541/42 hits before the effects of the first eruption are over. This has always been a possible explanation for the really big environmental events such as the "540 event"; i.e. it is caused by multiple big volcanoes. Overall, such a new dating scenario would remove the necessity of invoking a cosmic vector to explain the effects around AD 540. I have no problem with this scenario, however some explanation will have to be found as to why so many veiled references to comets and sky gods occur just at that time (see for example McCafferty and Baillie 2005 The Celtic Gods: comets in Irish mythology. Tempus). Recommendations: 1) The authors need to consider providing proper detail of the two-stage nature of the 536-550 environmental event. 2) The authors need to explain how an eruption can have up to 15 years of environmental effect unless they accept some of the later recommendations in this list. 3) A clearer statement on the nature of the glitch in the original Dye 3 dating (relating to the first half of the first millennium AD) would help to clarify the issue of dating error. 4) The authors have to decide whether the spaced tree-ring events at 536, 541/42 and 574/75 (as detailed above) are in fact related to the effects of the volcanoes that produced the acid layers in Greenland ice at "529, 533/34 and 567/68". That is, should they move the ice core dates forward by about 7 years? 5) If recommendation 4 is accepted then the authors should attempt to delineate exactly where the ice-core problems (giving rise to the 7-year offset) actually occur. Presumably these relate to years of multiple snowfall which would be important to know about. 6) If recommendation 4 is not accepted obviously I will be at liberty to present the above evidence in detail in a future publication.,..also 7) If recommendation 4 is not accepted it leaves open the question of the cause of the 540-543 secondary downturn. In fact, if the recommendation is not accepted the clear absence of a volcanic signal in the years around 540-543 would actually reinforce the need for an alternative cause (probably a cosmic one) for the second stage environmental effects. On a personal note. If the ice-core dates have to be shifted by seven years obviously that is 'the sting in the tail' for the current authors. The 'sting in the tail' for me is the demise of the inherently interesting comet scenario. What is most interesting about this current paper is the revelation that all the initial ice-core work missed the cause of the largest environmental event in the last two millennia. As it was only found by detailed re-analysis, it leaves open the question of what other major events have been missed. Reviewer #2 (Highlight): This is a very important paper because it presents data relevant to a major research question, namely what caused the global environmental event across the period AD 536 to AD 545 and beyond. HOWEVER The paper CANNOT be published as it stands for a series of reasons given in my review below and attached. It requires significant re-thinking by the authors and quite possibly revision of their chronology. All relevant detail is given in the review. I would encourage the authors to take on board my recommendations and to revise the paper accordingly. Attachment Converted: "c:\eudora\attach\2_reviewer_attachment_1_1196255893.pdf" Attachment Converted: "c:\eudora\attach\2_reviewer_attachment_2_1196255893.pdf" Attachment Converted: "c:\eudora\attach\2_reviewer_attachment_3_1196255894.pdf"