Let me first clearly confess that the following is at the limits of my competence as an amateur, so most of the following is my opinions and impressions, rather than an objective analysis of the relative cases.
‘Mr. McIntyre’ has indeed spoken, both on his blog and in the ‘open discussion’ section of CP, on the subject of the Juckes et. al. millennial temperature reconstructions. Both CP (Goosse) and Juckes responded. The paper was eventually published.
The discussions are interesting in a number of ways.
First, my impression of the critiques and comments made by Willis Eschenbach and Steve McIntyre. They appear to have made several legitimate and important points about the methodology and the choice of proxies used by the team. I have no idea which method of analysis is ‘best’, so won’t argue the toss on this, but I was left with the impression that the selection of proxies was not well handled. McIntyre legitimately points out that the reasoning behind the a priori selection of proxies was not followed in all cases, and the exclusion, in particular, of the Indigirska proxy was not adequately explained.
It is quite possible to argue that the Juckes team chose not to use the Indigirska data because this ‘belonged’ to another group of scientists who are using it as a ‘work in progress’, and to do so would be unfair to them. But if this is the case, it would have been simple and clear enough to simply say so, which the Juckes team does not do.
It is also possible to argue that the Indigirska data is not used because it materially changes the appearance of the reconstruction, specifically, by adding to earliest temperature estimates and taking away from more recent ones. However, as I do not know enough to decide whether McIntyre’s version of the ‘revised’ reconstruction, shown on his website, is a correct analysis, I’m reluctant to suggest either that he has shown this to be the case, or conversely that the Juckes team has shown that it is not. At the moment, my nose is telling me that McIntyre may be ‘winning’ this one.
Hugues Goosse’s final comments are interesting in their own right. The impression they gave me was that a second and the a third draft had to be produced, to respond to some (if not all) of the criticisms offered by Eschenbach and McIntyre, and also by the reviewers, that this was in itself a difficult and sensitive matter, and, as James Annan pointed out, may have been the stage at which Zorita’s involvement came to an end. There seems little value in speculating why, however; maybe Zorita is willing to offer an answer?
So far, so good. The original paper as presented to CPD was in need of revision. Several important points were eventually revised, though there remained one or two matters on which McIntyre, and probably Eschenbach, would have remained dissatisfied. Examples might be the selection of some proxy datasets with unspecified correlations with temperature, such as the bristlecone series and a couple of the aquatic series, and also the choice of analytic methodology.
The final paper is certainly the better for the revisions which have been included. But is the ‘Union’ reconstruction presented in it a ‘better’ reconstruction that what has appeared previously, or is it little more than a ‘revised Moberg’ with a new method added? The answer to this may well be to do with the area about which I know and understand least, the analysis method, so I’m not going to pretend to know. I’m willing to accept that it is at least an open question, and the burden should probably be on the Juckes team to justify its use.
Then the wheels come off. It stops being a discussion of the technical details of the paper, and becomes a different creature entirely, when McIntyre posts on his blog. I have no objection to him criticising the Juckes reconstruction, nor to him expressing his opinion about the validity of the paper, but McIntyre’s language goes far beyond this.
The tenor of the comments on CA is distinctly aggressive and provocative. The implication is clearly intended, that The Juckes et. al. reconstruction is deliberately deceptive, that it chooses to misuse or ignore data when this doesn’t suit, and that, all in all, it represents an attempt to make the data match a predisposition determined by an agenda. Such imputations are not likely to induce a sympathetic or reasonable response in anyone, least of all a group of highly respected and dedicated scientists. Nor do they serve to illuminate the questions which should matter, and which were suitably expressed in the comments on CPD.
I don’t mind Steve McIntyre pushing science teams to be open and accurate in their work. I don’t mind him, or others, finding and voicing criticisms of science, even if I don’t especially want to agree with him. But I also don’t see the value in using what amounts to intimidatory and inflammatory rhetoric in a public forum; I don’t think it serves the science, or McIntyre himself, particularly well.
What about the actual matter in dispute: What was the global temperature in the past? Which proxies should or shouldn’t be used in building reconstructions? How the hell are you supposed to build any kind of a picture of temperatures in the last millennium, with all the uncertainties that both the papers on the subject, and the critics of the those papers, have raised in the past ten years? And, finally, if the acceptability or otherwise of a reconstruction is a function of an individual’s predisposition towards the question, what is the point of even having an argument in the first place; there will be no chance of a mutually acceptable result.
The paper mentions 23 existing proxy datasets. There may be more, who knows? Has anyone produced a definitive cross comparison of all 23? If not, why not? Sure, there are going to be contradictory signals and large error bars, and some sets will start part-way through, and others will look like outliers, but at least there would be, somewhere, a complete picture of all the available proxy records. Would anyone like to do this?
I’d also like to say, as an endnote, that producing proxy reconstructions of temperature must be about as hard as science gets. I have enormous respect for the people who do this work. I am also aware that it is much easier to pick holes in someone else’s work than it is to originate the research. I don’t wish my opinions expressed above to be read as denigratory of the Juckes team, even though, in the end, I suspect that McIntyre has a point. I also hope that Steve, Willis and all others involved recognise that I am trying to be as fair as I can, and any misunderstanding is entirely due to my own inadequacy.
3 comments
Comments feed for this article
October 18, 2007 at 1:12 pm
Ian Hopkinson
We seem to be getting awfully tied up here with “audit” as implemented by ClimateAudit, perhaps it would be more helpful to think about other types of audit and from there construct the type of additional audit we might require of climate science.
In my view RealClimate is at it’s best when it is giving a fairly straight description of science. Similarly ClimateAudit is at it’s best when someone is doing something interesting with data i.e. JohnV’s OpenTemp analysis or Steve McI’s acquisition of his own dendro records (a work in progress). I believe Judith Curry (as reported in Storm World – which I can recommend) made similar comments over the hurricanes dispute – things went down hill when discussion left the fairly strictly scientific arena.
Obviously here everything is always at it’s best because it’s generally terribly polite 😉
October 20, 2007 at 2:34 am
Steve Reynolds
Fergus, it looks like your analysis here is so good that no one can find anything to say about it.
October 20, 2007 at 8:17 am
fergusbrown
Or so bad that nobody thinks it’s worth the effort… 🙂
One thing which gave me pause was the Esper et. al. (2007) paper, which I reported on last month. I think both papers used this ‘new’ technique of analysis, in the Esper case, for recalibration. The effect in the Esper paper was to lower estimates for the MWP by 0.2C. When I get time I’ll look at it more closely.