Self-described “Hansen bulldog” Tamino, writing at NASA’s realclimate blog hosted by Hansen’s other bulldog (Gavin), wrote:
As another example, Montford makes the claim that if you eliminate just two of the proxies used for the MBH98 reconstruction since 1400, the Stahle and NOAMER PC1 series, “you got a completely different result — the Medieval Warm Period magically reappeared and suddenly the modern warming didn’t look quite so frightening.” That argument is sure to sell to those who haven’t done so. But I have. I computed my own reconstructions by multiple regression, first using all 22 proxy series in the original MBH98 analysis, then excluding the Stahle and NOAMER PC1 series.
As always with the Team, you have to watch the pea under the thimble. Tamino has totally misrepresented and misinterpreted Montford on this point. Neither Montford (nor I) ever made such an assertion. The only person to do so, as I’ll show below, was Mann himself.
In our 2005 (EE) article where we analyzed the various permutations and combinations, our concern wasn’t with the Stahle-NOAMER PC1 pair, but with the Gaspe-NOAMER PC1 pair. In our 2005 article, we closely examined both the Gaspe and bristlecone proxies – believing then, as now, that if these series had unique capability to interpret world climate fields, then readers should be enabled to as much as possible about the unique characteristics of these groves.
The Stahle-NOAMER PC1 combination wasn’t mentioned in our articles. Nor did we (or Montford) present the particular sensitivity combination that Tamino now purports to rebut. This combination arose not in our analyses, but in Mann’s own analyses.
In their November 2003 response to MM2003, Mann presented a graphic that showed elevated early 15th century values with variations of three series (1) no NOAMER PC1; (2) no Stahle PC1 and (3) the shorter archived version of Twisted Tree, rather than the longer grey version used in MBH98 (in which early portions did not have the usual minimum numbers of trees.)
Mann et al 2003, Figure 1. The three datasets in the caption are the NOAMER PC1, the Stahle PC1 and Twisted Tree.
As Montford accurately reports in HS Illusion, I was extremely interested in this particular graphic because Mann et al themselves, in effect, conceded that the differences arose out of only a few series. Montford described this as follows:
Mann may well have felt that he had done enough to fend off McIntyre’s criticisms but McIntyre’s perspective was quite different. Without realising that he’d done it, Mann had inadvertently shone a little light on another murky corner of his famous paper. To McIntyre, what made Mann’s response most interesting was not the fact that Mann had used an undisclosed methodology, but the fact that if you left out just two of the proxy series – the Stahle and NOAMER PC1s – you got a completely different result – the Medieval Warm Period magically reappeared and suddenly the modern warming didn’t look quite so frightening. What this meant was that Mann’s result – that the Medieval Warm Period didn’t exist – seemed to rest on just a tiny fraction of his data. The rest of the series were just ‘noise’. Mann may well have been justified in using a stepwise procedure, but if his conclusions depended on just two PC series, then they could hardly be considered robust.
Note Tamino’s selective quotation from Montford’s book. Montford was describing my reaction to Mann’s 2003 response to MM2003. Neither Montford (nor I) claimed that Mann’s calculation in his November 2003 response were correct. Montford described the impact of Mann’s calculation on me. At the time, we hadn’t isolated the precise difference between our calculation and Mann’s calculations. While Mann attempted at the time – mostly successfully in the climate science community – to distract attention to replication details involving unreported aspects of their methodology (an experience which informs some of my present procedures in dealing with these guys), his own diagram showed me that the differences arose from only a few series. Which we proceeded to analyse in detail.
The Twisted Tree series was quickly seen to be moot as it did not come into play in the AD1400 step. (Mann’s defence of his version was unconvincing to say the least. Our comparison used the archived version which did not go back as far as the grey version. The early portion of the grey version included periods with less than the minimum number of cores for a chronology under Jacoby-d’Arrigo methods.) Given that MBH claimed to have screened chronologies to ensure a minimum number of cores, their insistence in this instance of using a chronology portion that did not meet their reported QC standards seemed odd, to say the least. Needless to say, no one in the “community” cared.
Our analyses also quickly showed that the presence/absence of the Stahle PC1 didn’t matter as it didn’t have a HS shape. This was a non-issue in our own presentations – Tamino’s mention of Stahle was therefore a red flag to both Montford and me. Needless to say, the (bristlecone) NOAMER PC1 did “matter”, as did another series (Cook’s 1983 Gaspe chronology), even though it hadn’t been mentioned in the 2003 MBH response.
It’s hard to say precisely what Mann did in his 2003 diagram showing such a large impact from the NOAMER PC1-Stahle PC1. Our own calculations yielding a high early 15th century also involved Gaspe. At the time of MM2003, we had not fully appreciated the important role of the unique and unreported extrapolation of the Gaspe series, but became aware of it very quickly in late 2003 and were fully aware of the issue when we submitted out 2004 Nature articles. (There is a later unpublished version of Gaspe that doesn’t have a HS shape – an issue that is avoided by the Team.)
I presume that Mann’s 2003 diagram inadvertently used the actual Gaspe data (rather than the version with the unique and unreported extrapolation) and this led to a more dramatic result than he might have intended – but this is only speculation.
Mann re-visited this calculation in a graphic in the unpublished 2004 Mann et al submission to Climatic Change shown below. This has a different result than the 2003 diagram, showing high early 15th century results from a Gaspe-NOAMER PC1 combination – a point on which we were and are in agreement with them.
A point that is little understood because of constant disinformation from the self-appointed bulldogs is that our results and those of Wahl and Ammann (or Mann) are in close agreement with sufficiently well-defined calculations – a point that we made in MM2005 (EE) as follows:
We emphasize the consensus between ourselves and Mann et al. on the results of sufficiently well-defined calculations. The PC calculations themselves are replicated between parties to complete accuracy. Differences remain in the emulations of NH temperature (given the PC series), but Mann et al.  showed a calculation with high early 15th century results if the North American PC1 were unavailable; the comments in Mann et al. [2004b] about the effect of the PC4 confirm this overall agreement if assumptions are sufficiently well defined.
In December 2005, as I’ve reported on many occasions, recognizing this point, I proposed to Caspar Ammann that we attempt to write a joint paper accurately setting down points of empirical agreement e.g. the results of 2 covariance PCS versus 5 covariance PCs; the impact of the presence/absence of bristlecones; verification r2’s, etc. Ammann refused, saying that this would be “bad for his career”. Whether or not it would have been bad for Ammann’s career, I think that the “community” would have benefited, if Ammann had accepted my proposal. (The offer was made in writing and including a proviso that the parties could go back to square one if they were unsuccessful in achieving a joint paper; it was a very fair offer.)
Instead, the Team’s approach has always been one of misdirection, Tamino’s post and Gavin’s commentary being only the most recent examples. Tamino’s realclimate post totally misrepresented Montford’s paragraph and purported to rebut a claim that neither Montford (or I) ever made. It wasn’t Montford (or I) that presented the Stahle-NOAMER PC1 combination that Tamino purports to refute. It was Mann himself.
Reasonable people can disagree as to whether the post should be entitled “Mann Bites Dog” or “Dog Bites Mann”, but surely no one can dispute that They are The Gang That Can’t Shoot Straight.