date: Thu, 20 Jul 2006 16:26:31 -0400 from: "Wahl, Eugene R" subject: RE: confidential to: "Keith Briffa" Hi Keith: I have been travelling in a remote area of southern Colorado without internet access from Monday until just now. I'm still several hours from Boulder, and need to get there before evening (it is 13:45 local time). I will look over your materials tonight and get back to you then, but please find copied below a message I sent to a contact at the Pew Trust for Global Climate Change in Washington on Monday, before I left for S. Colorado. It concerns perspective requested in relation to US House hearing that occurred yesterday. I am very glad to converse with you on this, as I had decided for my part to initiate communication for the reasons I will now outline. From my perspective, the skeptic perspective is actually getting scientifically uncritical FOSTERING lately. I say this in regard to the US National Academy of Science (NAS) report released last month, and also the Energy and Environment Committee hearings in Washington, yesterday. In both cases the NAS spokespersons highlighted the Mann et al. way of calculating PCs of ITRDB proxies from N. America (centered and standardized in relation to the calibration period mean and SD rather than the full time series length) as a significant mistake in method, related to the"hockey stick" shape of the reconstruction. However, in neither case, and also not in the NAS report, did they cite the Wahl-Ammann (WA) results that this alteration from "standard" practice, when put into the actual WA reconstruction algorithm, actually makes extremely little difference in the actual reconstruction--0.05 degrees over 1400-1449 (the period in which this is relevant). Any other effect on the reconstruction is due not to the reference period for centering/standardizing, but rather whether the data are standardized (MBH) or not (MM). It seems that this examination has been entirely overlooked by the NAS (it was again yesterday by Gerald North in front of the House committee), and this is essentially due to lack of reading WA fully. I know this for a fact from one of the NAS authors, with whom I spoke about this issue directly on June 22. The NAS authors did cite WA in several other regards, almost entirely focusing on the most MBH-critical portions of WA, and ignoring important contextualizing we made of some of these criticial results. So there is a brief synopsis until I can look over what you sent carefully. I think this is a key issue that needs to be CORRECTLY dealt with by the IPCC. [In my mind there is no justification for the NAS overlooking it, and then continuing to do so a month later, even after I made the direct author explicitly aware of this ommission.] The skeptical perspective is valuable and important, but it is not scientifically valid to give it weight that, objectively, it does not merit. PLEASE LOOK OVER THE COPIED TEXT BELOW...IT ALSO HAS THE EXACT REFERENCES IN THE WAHL-AMMANN PAPER THAT DEAL WITH THESE ISSUES, AND THE "IN PRESS" TEXT IS ATTACHED. Peace, Gene Dr. Eugene R. Wahl Asst. Professor of Environmental Studies Alfred University ********************* START OF COPIED MESSAGE ***************** DELETED IS A PORTION OF THE MESSAGE THAT IS NOT RELEVANT IN THIS CONTEXT ... 2) ...continuing...MBH used the calibration period (1902-1980) to get the means and SDs to center and normalize all their data because they were using 11 different lengths of proxy data to make their final product. As I understand it, they decided it was better to use one common period for this purpose, so that everything was centered/scaled according to the same standard, rather than 11 different sets of standards. That this was therefore a rational response to a perplexing issue is often lost in all the noise about their standardization practice. 3) The PC1 of the NAm ITRDB data is only one of 22 proxies used in the 1400-1449 reconstruction segment, and thus not the only information that went into the reconstruction. That it had a hockey-stick shape did mean that it projected strongly onto the instrumental PC1 that was the reconstruction target. [This reconstructed instrumental PC1 was then plugged back into the UDV' (singular value decomposition, SVD) formula as the U matrix, to yield the temperature field reconstruction, from which the N Hemisphere average was calculated]. 4) HERE IS THE REAL MEAT ON THIS ISSUE...Wahl-Ammann (attached) have shown that the entire issue of the N. American ITRDB proxy PCs retained and differing standardization conventions leads to NEARLY IDENTICAL reconstructions in ALL situations, as long as the reconstructions are taken to convergence. That is, we started with 2 NAm ITRDB PCs across the differing conventions--plugging them into the MBH reconstruction--and then did the same up to five PCs. In all cases (including the main MBH emulation itself), the reconstructions essentially converged. This is just what should be expected from first principles--the underlying data is the same in all cases, and the different centering and PC calculation conventions (using variance/covariance or correlation matrix to do SVD) should only have the impact of translating the underlying patterns ACTUALLY in the data differently across the rank ordering of PCs. Thus, by making sure to check for convergence [a "due dillegence" procedure, to utilize MM-type usage], one can make sure all the patterns in the data are being utilized. MOST IMPORTANT RESULT IS IN THIS PARAGRAPH In the case (a) of MBH with (correlation matrix with calibration period standardization), PCs1-2 are enough to get the converged result. Using the stated method (b) from MM 2005 Energy and Environment (correlation matrix with full-period standardization), PCs1-2 yield the converged result. Using the method (c) MM 2005 EE actually used in their code (variance/covariance matrix with full-period centering but NO standardization), PCs1-4 lead to the converged result. Note that there actually is an offset of 0.05 degrees (MBH is cooler, by the way) between (a) and (b) over 1400-1449, which is the true BIAS that the MBH standardization convention ACTUALLY ADDS. (c) requires more PCs for convergence because using the var-cov matrix rather than the correlation matirx on unstandardized data means that the first one or two PCs will be capturing the differences in variance across the proxies input into the process, which in fact vary by a factor of about 13. In effect, the PC algorithm in this case is first standardizing, and only then capturing signal. MM have made a big deal about the hockey stick pattern being shifted to the 4th PC in this case, but in fact it is a necessary result of their procedure, not some true degrading of this signal's importance. All of this is set forth on pp 22-25 (outline of analysis) and 30-32 (results) of WA. Not one word of it is mentioned in the Wegman report, which is highly misleading, as it has been publically available since March!! WA is accepted/in press and is publically available. Note that it has yet to go through final galley corrections. ********************* END OF COPIED MESSAGE *************** ________________________________ From: Keith Briffa [mailto:k.briffa@uea.ac.uk] Sent: Tue 7/18/2006 10:20 AM To: Wahl, Eugene R Subject: confidential Gene I am taking the liberty (confidentially) to send you a copy of the reviewers comments (please keep these to yourself) of the last IPCC draft chapter. I am concerned that I am not as objective as perhaps I should be and would appreciate your take on the comments from number 6-737 onwards , that relate to your reassessment of the Mann et al work. I have to consider whether the current text is fair or whether I should change things in the light of the sceptic comments. In practise this brief version has evolved and there is little scope for additional text , but I must put on record responses to these comments - any confidential help , opinions are appreciated . I have only days now to complete this revision and response. note that the sub heading 6.6 the last 2000 years is page 27 line35 on the original (commented) draft. Cheers Keith -- Professor Keith Briffa, Climatic Research Unit University of East Anglia Norwich, NR4 7TJ, U.K. Phone: +44-1603-593909 Fax: +44-1603-507784 http://www.cru.uea.ac.uk/cru/people/briffa/ Attachment Converted: "c:\eudora\attach\wahlammann_climaticchange2006.pdf"