Several readers have asked me to respond to recent comments about me at Real Climate: briefly, Osborn has made one misrepresentation after another and made statements with seemingly complete indifference as to whether he has any basis for making the claims.
In respect to Schmidt’s whinge, as Lucia sagely observed a couple of years ago in connection with Schmidt’s defence of upside-down Tiljander, one cannot assume that people actually asked the questions that Schmidt says they asked or that his answers are adequate because he says so:
I might suggest that you are assuming that people asked the questions Gavin says they asked, and that Gavin’s answer to their questions is adequate because Gavin tells us his answer is adequate.
Schmidt and Osborn’s most recent comments were on July 2, the last day that comments were open on the thread, otherwise I would have attempted to respond at Real Climate. Unfortunately it takes more time to respond to their fantasies than it does for them to make them.
CRU’s Failure to Use or Disclose the Shiyatov Dataset
In a recent CA post, I observed that Shiyatov had crossdated 1021 subfossil trees from his 1968 and 1983 transects plus hundreds of living trees, from which the samples sent to Schweingruber were a minute subset. Although Shiyatov’s comprehensive dataset is by far the most imposing dendro work at Polar Urals, Briffa et al 2013 neither used nor reported the extent of the Shiyatov dataset. In a comment at RC, I asked Osborn to explain their failure to use or even disclose the very large Shiyatov dataset as follows:
Why didn’t CRU use this dataset in Briffa et al 2013 instead of the inadequately replicated dataset that it reported on? Did CRU attempt to obtain access to this data and receive a rejection? And why didn’t CRU report the existence of Shiyatov’s crossdated dataset in its review of previous work at Polar Urals?
It appears that Osborn had no valid answer to these questions as his answers were either unresponsive or evasive or attacks on me. First, here is Osborn’s answer as to why they didn’t use the Shiyatov data:
The dendrochronological data from this recent sampling have not yet been published and it is the prerogative of the Ekaterinburg laboratory to publish the first dendroclimatological analysis of the data that they have spent many years and extensive effort in collecting and processing. …
Your question and other commentary at your blog may give readers the false impression that we have published using an inadequate dataset.
Watch the pea.
While no one would object to Shiyatov and his associates (who by the way, are listed as coauthors of B13) being the first to publish their “recent” work, this is irrelevant to data collected in 1983 and 1968. Osborn’s answer misdirects away from this issue. And, in any event, this has nothing to do with CRU’s failure to even disclose the existence of 1021 crossdated subfossil trees from the two early transects. This should have been reported in the article, together with an explanation of why they didn’t use it, letting reviewers and readers decide for themselves whether CRU was justified in using less than 8% of the subfossil data. Had Briffa reported the existence of the large Shiyatov dataset, a reviewer might well have asked Briffa and coauthors why they weren’t using the best available data and perhaps even asked them to carry out the analysis using comprehensive data. This would have led to a more valuable study. CRU should have asked Shiyatov for the right to use the subfossil data from their 1968 and 1983 transects and, if they were refused, this should have been reported as well.
If I gave the “impression” that CRU used an “inadequate” dataset, so be it. CRU did use an inadequate dataset. Their recommended Polar Urals chronology had only two trees during part of the 11th century.
Osborn also implies that I should have raised the inadequacy of their data in previous criticisms and that this is moving the goalposts on them. However, the Shiyatov transects were not described in Briffa et al 1995 either. Indeed, Briffa et al 2013 is the first CRU article that comes even close to providing technical information on the sites. My first appreciation of the comprehensiveness of Shiyatov’s work came from the relatively recent publications, Shiyatov et al 2011 and Mazepa et al 2011, which I consulted as part of my consideration of Briffa et al 2013, though neither was cited by Briffa. Had I previously realized the comprehensiveness of the Shiyatov dataset, I would have raised the issue earlier.
Surely the responsibility for using the most comprehensive available data rests with CRU and the journal reviewers. If they neglect to do so, they deserve criticism. It is ludicrous for Osborn to attack me because they were negligent.
CRU’s False Claims about “Advocacy” and “Heavy Promotion”
Even my most antagonistic critics have nearly always understood that I have not attempted to offer my own reconstruction. Indeed, this annoys many critics e.g. here. On numerous occasions, I’ve explained that reconciliation of inconsistencies between proxies – especially between seemingly like proxies in the same region – is, in my opinion, a prerequisite prior to any reconstruction being anything other than arbitrary. The inconsistency between various lines of evidence in the Yamal-Polar Urals region is a type case of such inconsistency.
Despite my longstanding policy against offering my own reconstruction under current information, CRU alleged that I had “promoted”, “heavily promoted” and “advocated” versions of the Yamal and Polar Urals chronologies, making these untrue claims both in their blog article itself and in Osborn’s inline comments. In a comment at RC, I rejected these allegations, pointing out that CRU had not provided any quotation or link supporting their assertions. Osborn was unable to provide any backup or source for his allegations.
In Figure 2 of the blog article, CRU had displayed two chronology variations, falsely asserting that I had “heavily promoted” these versions as follows:
Figure 2b compares the new Yamalia chronology with two alternative chronologies heavily promoted by McIntyre and others – the so-called Polar Urals “update” chronology and a Yamal chronology using modern samples from the Khadyta River site.
The Yamal version that Osborn attacked was shown as a sensitivity analysis in my original Sep 27, 2009 post on Yamal and not cited or “promoted” or “advocated” in any subsequent presentation or even blog post. My issue with Briffa-style regional chronologies had been their sensitivity (non-robustness) to inhomogeneity. Under the statistical model necessary for Briffa-style “regional” standardization, results should not depend on which subpopulation is used. However, this was very much not the case with his influential Yamal chronology. To illustrate the inhomogeneity, I showed three chronology variations (scenarios, if you prefer) in my Sep 27, 2009 blog post: (C) the Yamal-11 only (as in Briffa 2000 and Briffa et al 2008); (B) both the Yamal-11 and Schweingruber’s Khadyta River and (A) Schweingruber’s Khadyta River only. I described the calculations as a “sensitivity” study and did not promote or advocate either Scenario A or Scenario B.
To avoid any potential confusion, I stated this non-advocacy as clearly as I could in a post that took notice of Briffa’s initial response to the Yamal controversy:
I did not propose the results of these sensitivity studies as an “alternative” and “more robust” chronology. I am not arguing that the Yamal versions using the Schweingruber data provide the “correct” climate history for the region. I am arguing that the version constructed by Briffa, and relied on so extensively in the literature since then, is not robust in its late-20th century portion to a small and reasonable inclusion of additional data.
We know from Climategate emails that CRU were reading CA posts at the time. Further, the contemporary responses of Briffa and Melvin first conceded that I had raised a “number of valid points that require further investigation” and then, in their second much more detailed response shortly before Climategate, Briffa and Melvin stated that “it is important to examine the robustness of RCS chronologies, involving the type of sensitivity testing that McIntyre has undertaken”. This latter article repeatedly used the term “sensitivity” to describe my analyses.
So Briffa and Melvin clearly understood at the time that I had not “promoted” or ‘advocated” Scenario A (or Scenario B or Polar Urals for that matter), but had raised “valid questions” and that it was “important” to examine inhomogeneity using the sort of sensitivity analyses that I had carried out.
Subsequent to my original September 27, 2009 post, I never employed or even illustrated Scenario A in any presentation or blog article, other than an incidental use in a May 15, 2012 post where I ironically observed that a Yamal chronology incorporating fresh data from Hantemirov was remarkably similar to Scenario B. My position on Esper’s Polar Urals version has been similar: that the inconsistency between proxies need to be reconciled before either can be relied upon, a position that I expressed in my submission to Muir Russell as follows:
In the absence of any explanation of the substitution [Yamal for Polar Urals], there is reason to be concerned about the reasons for using one series rather than the other
In his inline comment, Osborn conceded that this was a reasonable point, but objected that I had raised this example of potential cherrypicking to a panel whose terms of reference including the examination of potential cherrypicking and which had requested submissions. (Not that the Muir Russell panel adequately investigated the matter, but that’s another story.)
When I challenged) Osborn on his untrue claim that I had “heavily promoted” or “advocated” Scenario A and/or Polar Urals, Osborn implicitly conceded that he had no source. Instead of then endeavouring to locate a source, Osborn stated that he would not search through “hundreds of comments” to try to find support for his assertion – though surely this is something that he should have done before making the comment, rather than after.
I will not search through the hundreds of comments you have made on the subject in multiple venues to find ones that might meet some agreed definition of “promotion”
It’s frustrating that a state-funded academic does not take responsibility for his statements. Unfortunately, this sort of recklessness seems to have increased rather than decreased following the inept Muir Russell.
Osborn then asked why “all of this” had happened if I had not “heavily promoted” and/or “advocated” Scenario A and/or Polar Urals arguing that I had left CA readers with the “impression” that these chronologies were “perhaps more reasonable” because I had “claimed they were based on newer or less biased data”.
Again, Osborn provided no citation or references. I never claimed that the data for Scenario A or Scenario B or Polar Urals was “less biased” – I defy Osborn to locate any support for that suggestion.
Nor do I agree with Osborn’s claim that CA readers have the impression that any of Scenario A, Scenario B or Polar Urals were “more reasonable” chronologies. It is my impression that CA readers are far more negative to tree ring chronologies than I am. And that rather than endorsing Scenario A or Polar Urals, the prevalent attitude is: none of the above. It’s not that these scenarios have been “promoted”, but that specialists have been untroubled by the inconsistencies. Briffa et al 2013 at least concedes the existence of the inconsistencies, but unfortunately, CRU are hopelessly incapable of carrying out the requisite statistical analysis (which is not easy).
In the blog article, CRU made another untrue assertion about Climate Audit with reckless indifference:
Over the years, McIntyre has advanced a number of other criticisms of our tree-ring work in northwestern Eurasia. We note here that these too are also wrong.: 1) the original Polar Urals chronology was not wrongly cross-dated as claimed in a 2005 submission to Nature by McIntyre and McKitrick. When we demonstrated this in our response, Nature decided to publish neither their comment nor our response. It is worth noting that this rejection, nor any acknowledgement of his erroneous conclusions, were ever mentioned by McIntyre on his blog. [my bold]
Their claim that I had never mentioned the rejection of this comment at Climate Audit is simply untrue. Osborn stated that he was unwilling to search through “hundreds of comments” at Climate Audit, but how can CRU pretend to have a basis for claiming that I had never “mentioned” the rejection of the comment at Climate Audit if they haven’t thoroughly examined the thousands of posts and comments. If CRU did not carry out the requisite examination, they should refrain from making claims that can only be made responsibly on the basis of a thorough examination. BTW I was able to locate mentions of this rejection (there’s more than one) in a few minutes, but I had an idea of where to look.
Nor do I agree that CRU’s response demonstrated that the observations about crossdating reported in the comment were “erroneous”. The calculations are what they are and yielded the results that I reported. In any event, it is common ground between us that the 11th century portion of the Briffa et al 1995 chronology did not meet quality control standards for number of cores (which implies, in my opinion, that this portion of the chronology should not have been used in Jones et al 1998, of which Briffa was a coauthor. Nor, for that matter, does the 11th century portion of the recommended Polar Urals version of Briffa et al 2013 – which supports the argument against Briffa’s decision to use an “inadequate” version of the Polar Urals data, rather than the comprehensive Shiyatov dataset.
Osborn also complained that I had insufficiently acknowledged their identification of root collar inhomogeneity at Polar Urals. However, in my first article in the recent sequence, I provided a lengthy exposition of their findings in connection with root collar inhomogeneity, complete with long quotations directly from their text.
Not only did I provide a thorough review of their argument on root collar inhomogeneity, I explicitly stated that their concerns about root collar and radial inhomogeneity seemed “entirely reasonable” to me:
These concerns seem entirely reasonable to me.
In my closing comment, I firmly endorsed Briffa’s stand against the use of radially deformed data.
I very much welcome the strong position taken by Briffa and coauthors against the use of radially deformed tree ring data.
I see no basis for any complaint on Osborn’s part that I had insufficiently acknowledged their concern about root collar inhomogeneity. However, root collar inhomogeneity is hardly the end of the story as inhomogeneity in Briffa’s population is pervasive. The purlasi_sc subpopulation appears as inhomogeneous as the root collar samples, but Briffa seem obtusely unaware of the problem. And while Briffa et al have reported one incident of inhomogeneity, their statistical tools for detecting inhomogeneity are hopelessly inadequate – a serious disappointment given that they’ve been working on this issue for over a decade. I observed (continuing my original criticism):
Inhomogeneity between sample populations are a very serious problem in Briffa’s RCS methodology. B13 asserts that need to test subpopulations for homogeneity, but are completely and irredeemably at sea in actually carrying out statistical tests for homogeneity – procedures on which many statistical specialists could have assisted.
In recent comments about Yamal, Gavin Schmidt claimed that he had been misrepresented in my recent post, but, all too characteristically, did not quote the supposed misrepresentation. Or even say explicitly who had made the supposed misrepresentation, though the link to the CA thread implied that I had been responsible.
Here is Schmidt’s complaint:
Unfair” might be misrepresenting my earlier post (criticizing McIntyre’s never-ending accusations of misconduct and his ultimately futile attempts to use FOIA/EIR to get hold of unpublished work) as a criticism of any specific analysis, despite the opening line being “Steve McIntyre is free to do any analysis he wants on any data he can find” and not mentioning his results at all.
“Unfair” might be taking a statement I made in that post (on May 11 2012 – note the date), pointing out that Briffa et al’s results would be different from what McIntyre had put up (on May 6 2012) (as the figure below demonstrates), and then using a calculation made on May 15 2012 to claim I was wrong.
Claiming that my comments were an intemperate response to his results posted 4 days later might, in some circles, also be considered “unfair”.
But when one examines my actual post, it does not contain any statements that contain the assertions about which Schmidt complains. Precisely what Lucia warns against when dealing with Schmidt.
As far as I can tell, the only sentences that have any connection to Schmidt’s complaint are the following:
First here is a comparison from CA in Sep 2009 here of the Briffa 2008 superstick to a version that simply incorporated Schweingruber’s Khadyta River data, applying the method used by Briffa for Taimyr in Briffa et al 2008. Real Climate screeched in fury against this comparison.
In May 2012, I did a quick calculation incorporating more recent Hantemirov data, showing that the resulting modern portion was remarkably similar (TM-climate science) to the green chronology of my September 2009 and had an almost identical discrepancy to the Briffa et al superstick. For reasons that remain unclear, Hantemirov objected to the calculation (see comments within post) and, once again, Real Climate screeched in fury.
The first paragraph seems unarguable to me, though Schmidt would undoubtedly use different words to describe the Real Climate post in September 2009. Eeven Schmidt cannot reasonably dispute that the September 2009 Real Climate post was extremely antagonistic. Indeed, CG-2 clearly demonstrated Schmidt’s intent to be as antagonistic as possible. Even though Briffa recognized that there were substantive technical issues relating to inhomogeneity, on Sep 29, 2009 (CG2-4564), the day before the antagonistic RC post, Schmidt told Jones, Osborn and Mann that the issue was “propaganda”:
This is not a peer review issue – this is a propaganda issue. And right now the good guys have conceded the field.
Schmidt presciently observed that his readers would not “worry about the details”:
The good news is that once something is out there, people will counter with links to that without themselves worrying about the detail.
Even in his most recent whinge, Schmidt does not seriously argue against the strident antagonism of Real Climate’s reaction in September 2009.
The next sentence in my post was the following:
In May 2012, I did a quick calculation incorporating more recent Hantemirov data, showing that the resulting modern portion was remarkably similar (TM-climate science) to the green chronology of my September 2009 and had an almost identical discrepancy to the Briffa et al superstick.
I can’t see anything objectionable in this sentence. I did indeed do the stated calculation (see here.) The calculation involving the Hantemirov data was not the same calculation as the emulation of CRU’s regional chronology that I’d calculated about 10 days earlier (see here) on May 6, 2012. Although Schmidt whinged about the May 6 regional calculation, nowhere is it mentioned in my actual post. Indeed, I haven’t yet commented on the handling of the regional chronology in Briffa et al 2006, though this is an interesting topic that I intend to cover and will discuss briefly in today’s post.
So there is only one remaining sentence in my post to which Schmidt might be objecting:
For reasons that remain unclear, Hantemirov objected to the calculation (see comments within post) and, once again, Real Climate screeched in fury.
Hantemirov’s objection to my calculation was expressed in a threaded comment) to my May 15, 2012, in which Hantemirov complained very angrily about my post as follows:
Steve, I’m horrified by your slipshod work. You did not define what you compare, what dataset used in each case, how data were processed, and what was the reason for that, what limitation there are, what kind of additional information you need to know. Why didn’t you ask me for all the details? You even aren’t ashamed of using information from stolen letters. Do carelessness, grubbiness, dishonourableness are the necessary concomitants of your job? With disrespect…
I disagreed with Hantemirov’s comments in an inline comment. Although this was only a blog article, I provided considerable detail on the calculation, including turnkey code to generate all graphics and results cited in the post. In addition, I have a good record of responding to requests for clarification. So I didn’t and don’t accept Hantemirov’s criticism. That the later Briffa et al 2013 Yamal chronology was similar to my calculation is further support that I had replicated the most important ingredients of their methodology. Be that as it may, the first part of the sentence reported Hantemirov’s intervention and is factually correct.
Hantemirov’s intemperate criticism was immediately seized upon by commenters at Real Climate on the then Yamal thread, first by Steve Metzler here and then by Phil Clarke here. Over the next week or so, RC commenters, abetted by Real Climate moderator Eric Steig, used Hantemirov’s criticism to launched one tirade after another against me: see page 4 and page 5. Comments included the following:
He’s like a delinquent schoolboy who’s left a trail of destruction behind, culminating in burning down the school. And then he’s all defiant, like: “What? What did I do wrong?”
McIntyre doesn’t just use bad (or out of context) data, he also *manufactures his own* if bad data aren’t readily available:
While Real Climate might not characterize their response to Hantemirov’s complaint as “screeching”, it is surely unarguably that Hantemirov’s complaint prompted a large number of furious comments at Real Climate, both by commenters and by moderator Steig.
Thus, I don’t see any factual errors or unfairness in my commentary.
Re-examining Schmidt’s complaint: Schmidt argued that it would be “unfair” to take a statement that he had made on May 11, 2012 “pointing out that Briffa et al’s results would be different from what McIntyre had put up (on May 6 2012) (as the figure below demonstrates), and then using a calculation made on May 15 2012 to claim I was wrong.”
If I had done what Schmidt described, then I would agree that his complaint would be justified. But as Lucia observed, Schmidt saying something is so doesn’t make it so. I did not directly refer to or mention his May 11 criticism of my May 6 emulation of the regional chronology. If Schmidt had directly quoted my words, this would have been evident to RC readers. Nonetheless, in order to avoid any misunderstanding on this point, I added a clarifying sentence to the text of the post itself on July 2 as follows:
the screeching at RC in response to my Hantemirov post occurred in comments (see page 4 on) to their Yamal post of May 11, which chronologically was screeching in fury at my post of May 6 about regional reconstructions.
Although I had not commented on Schmidt’s May 11 criticism of my regional chronology emulation, Schmidt was quoted by a thread commenter here. Again, to avoid any potential misunderstanding, I’ve added a clarifying inline comment to hopefully alleviate Schmidt’s anguish that he might have been misunderstood by a commenter on a thread somewhere on the internet.
This is not to say that I agree with Gavin’s May 11 criticism of my emulation of the regional chronology. Only that I had not raised the issue of the regional chronology, a topic which I had not yet considered in connection with Briffa et al 2013 and which I will now briefly consider.
The Regional Reconstruction
CRU’s 2006 regional chronology has been a longstanding interest at CA and was the topic of a partly successful, partly unsuccessful FOI request. In April 2012, I was successful in obtaining a list of sites used in the unreported 2006 regional chronology. On May 6, I reported the results of my emulation as shown below:
Figure 1. Emulation of CRU regional chronology, as shown in CA post http://climateaudit.org/2012/05/06/yamal-foi-sheds-new-light-on-flawed-data/
In his May 11 article, Schmidt sneered at this “insta-reconstruction” and stated that he could “guarantee” that Briffa’s eventual reconstruction would have different results than my emulation:
The irony is of course that the demonstration that a regional reconstruction is valid takes effort, and needs to be properly documented. That requires a paper in the technical literature and the only way for Briffa et al to now defend themselves against McIntyre’s accusations is to publish that paper (which one can guarantee will have different results to what McIntyre has thrown together).
I was just as confident as Schmidt that CRU’s planned publication would feature a chronology that would be more HS-like than the 2006 regional chronology. After all, CRU had ferociously contested production of their calculation, even under FOI. Osborn argued that release of the 2006 regional chronology would result in:
adverse reputational consequences for the individual scientists involved in this work and the University itself if disclosure had been effected.
On the specific issue of whether the planned chronology of Briffa et al 2013 would differ from the 2006 regional chronology, Schmidt and I could not have been more in agreement.
However, it seems to me that Schmidt’s May 11 post additionally implied that little credence could be given to my emulation of the 2006 CRU regional chronology, which Schmidt described as an “insta-reconstruction” and said that I had “thrown together”. The next graphic compares my insta-reconstruction with information provided in the SI to Briffa et al 2013. Despite Schmidt’s sneering, I think that even my severest critic must concede that there is a very close resemblance between my emulation and the actual 2006 regional chronology.
Given the remarkable similarity between my emulation and the actual 2006 regional chronology, one wonders whether CRU might have been better off merely producing the regional chronology. CRU purported to justify its refusal on the grounds that they had been “working” on the regional chronology since 2006 and were still working on it. Schmidt purported to justify their regional chronology work as follows:
Regional reconstructions are generally more worthwhile than reconstructions from a single site because, if there is shared variance, the regional result is likely to be more robust and be more representative – and that makes it more valuable for continental and hemispheric comparisons. The key issues are whether all the trees (or some subset of them) share a common signal (are they mostly temperature sensitive? are some localities anomalous? etc.). It isn’t as simple as just averaging all the trees in a grid box or two. The history of such efforts follows a mostly standard path – local chronologies are put together, different ‘standardisation’ techniques are applied, more data is collected, wider collations are put together, and then regional reconstructions start to appear.
To the extent that Briffa et al 2013 is supposed to deliver on this expectation, it is a complete failure. Supplementary Material 9 is the only discussion of the regional chronology. It begins with a discussion of what appears to be the 2006 regional chronology, though this is not explicitly stated. On the contrary, they introduce the (2006) chronology as having been constructed to illustrate a statistical tradeoff between random sampling error and systematic error:
In many statistical applications, there is a balance between minimising random sampling error and minimising systematic error (i.e. a trade-off between variance and bias). This issue arises in tree-ring chronology construction too, balancing the inclusion of more data to reduce the noise (i.e. the sampling error) against the inclusion of data from too large an area such that the signal becomes ambiguous or even incompatible. In this section we briefly evaluate some of the key aspects associated with this balance, as they apply to the Yamalia chronologies.
We have constructed four different chronologies to illustrate some of the issues associated with chronology sampling error and bias, and to compare these between a single-site chronology and a chronology developed from a much larger region.
Whatever CRU’s original purpose in constructing the 2006 regional chronology, one feels confident that it was not constructed merely as an illustration of the tradeoff between variance and bias. SupMat 9 then more or less concedes that the sites in the 2006 regional chronology meet the common signal test of Briffa et al 2008, but adduce arguments against their inclusion that are nowhere in evidence in Briffa et al 2008:
For the current study of the Yamalia region, the problems of a biased RCS curve and a temporally inconsistent chronology clearly outweigh the benefits of reduced sampling error, and therefore this strategy is not adopted in the main part of this study.
Thus, despite “working” on regional chronology methodology for nearly two decades, CRU is intellectually no further ahead than when they started: arm-waving ad hockery. But that’s a story for another day.