P
pixelsmith
Guest
The IPCC Trick
Mann’s IPCC trick is related to the Jones’ trick, but different. (The Jones trick has been explained in previous CA postshere, here and consists of replacing the tree ring data with temperature data after 1960 – thereby hiding the decline – and then showing the smoothed graph as a proxy reconstruction.) While some elements of the IPCC Trick can be identified with considerable certainty, other elements are still somewhat unclear.
The diagram below shows the IPCC version of the Briffa reconstruction (digitized from the IPCC 2001) compared to actual Briffa data from the Climategate email of October 5, 1999, smoothed using the methodology said to have been used in the caption to the IPCC figure (a 40 year Hamming filter with end-point padding with the mean of the closing 20 years).
Figure 3. Versions of the Briffa Reconstruction in controversy, comparing the original data smoothed according to the reported methodology to a digitization of the IPCC version.
Clearly, there are a number of important differences between the version sent to Mann and the version that appeared in the IPCC report. The most obvious is, of course, that the decline in the Briffa reconstruction has, for the most part, been deleted from the IPCC proxy diagram. However, there are some other frustrating inconsistencies and puzzles that are all too familiar.
There are some more technical inconsistencies that I’ll record for specialist readers. It is very unlikely that that the IPCC caption is correct in stating that a 40-year Hamming filter was used. Based on comparisons of the MBH reconstruction and Jones reconstruction, as well as the Briffa reconstruction, to versions constructed from raw data, it appears that a Butterworth filter was used – a filter frequently used in Mann’s subsequent work (a detail that, in addition, bears on the authorship of the graphic itself).
Second, the IPCC caption stated that “boundary constraints imposed by padding the series with its mean values during the first and last 25 years.” Again, this doesn’t seem to reconcile with efforts to replicate the IPCC version from raw data. It appears far more likely to me that each of the temperature series has been padded with instrumental temperatures rather than the mean values of the last 25 years.
Finally, there are puzzling changes in scale. The underlying annual data for the Jones and Briffa reconstructions are expressed in deg C (basis 1961-1990) and should scale simply to the smoothed version in the IPCC version, but don’t quite. This may partly derive from errors introduced in digitization, but is a loose end in present replication efforts.
The final IPCC diagram (2.21) is shown below. In this rendering, the Briffa reconstruction is obviously no longer “a problem and a potential distraction/detraction”and does not “dilute the message”. Mann has not given any “fodder” to the skeptics, who obviously did not have a “field day” with the decline.
IPCC Third Assessment Report Figure 2.21: Comparison of warm-season (Jones et al., 1998) and annual mean (Mann et al., 1998, 1999) multi-proxy-based and warm season tree-ring-based (Briffa, 2000) millennial Northern Hemisphere temperature reconstructions. The recent instrumental annual mean Northern Hemisphere temperature record to 1999 is shown for comparison. Also shown is an extra-tropical sampling of the Mann et al. (1999) temperature pattern reconstructions more directly comparable in its latitudinal sampling to the Jones et al. series. The self-consistently estimated two standard error limits (shaded region) for the smoothed Mann et al. (1999) series are shown. The horizontal zero line denotes the 1961 to 1990 reference period mean temperature. All series were smoothed with a 40-year Hamming-weights lowpass filter, with boundary constraints imposed by padding the series with its mean values during the first and last 25 years.
Contrary to claims by various climate scientists, the IPCC Third Assessment Report did not disclose the deletion of the post-1960 values. Nor did it discuss the “divergence problem”. Yes, there had been previous discussion of the problem in the peer-reviewed literature (Briffa et al 1998) – a point made over and over by Gavin Schmidt and others. But not in the IPCC Third Assessment Report. Nor was the deletion of the declining values reported or disclosed in the IPCC Third Assessment Report. [Dec 11.- IPCC TAR does contain a sly allusion to the problem; it mentions "evidence" that tree ring density variations had "changed in their response in recent decades". Contrary to claims of realclimate commenters, this does not constitute disclosure of the deletion of the post-1960 values in the controversial figure or even of the decline itself.] The hiding of the decline was made particularly artful because the potentially dangling 1960 endpoint of the Briffa reconstruction was hidden under other lines in the spaghetti graph as shown in the following blow-up:
Figure. Blow-up of IPCC Third Assessment Report Fig 2-21.
To my knowledge, no one noticed or reported this truncation until my Climate Audit post in 2005 here. The deletion of the decline was repeated in the 2007 Assessment Report First Order and Second Order Drafts, once again without any disclosure. No dendrochronologist recorded any objection in the Review Comments to either draft. As a reviewer of the Second Order Draft, I asked the IPCC in the strongest possible terms to show the decline reported at CA here:
They refused, stating that this would be “inappropriate”, though a short discussion on the divergence was added – a discussion that was itself never presented to external peer reviewers.
Returning to the original issue: climate scientists say that the “trick” is now being taken out of context. The Climategate Letters show clearly that the relevant context is the IPCC Lead Authors’ meeting in Tanzania in September 1999 at which the decline in the Briffa reconstruction was perceived by IPCC as “diluting the message”, as a “problem”, as a “potential distraction/detraction”. A stone in their shoe.
Update (Dec 11) : Some of the follow-up comments on this post do shed light on this sequence and enable a more precise interpretation of the emails. With the benefit of these comments, there are a couple of points on the chronology that I need to modify, particularly in respect to the role of the October 5 revision of the Briffa reconstruction in respect both to the Arusha meeting and to the hide the decline.
The Arusha meeting objected to the Briffa reconstruction “diluting the message” and reducing confidence in the multiproxy reconstructions. And, of course, it is the overstated confidence that has been the primary objection here. However, I agree with critics who observe that the proximate objection to the Briffa reconstruction at Arusha was not that the decline per se diluted the message, but the Briffa reconstruction overall diluted the message and interfered with a “tidy story”. The stone in the shoe was that the Briffa reconstruction prevented a “tidy story”; the “decline” as a separate problem came a bit later.
After the Arusha meeting, Briffa hurriedly re-did his chronology and the new version was delivered to Mann on Oct 5, 1999 – it was this version that had the big decline. In the First Order Draft of Oct 27, 1999, IPCC author Mann deleted the post-1960 portion of the Briffa reconstruction plus other things that I don’t yet quite understand. Jones’ trick, as observed in the post, is a little different. (The post-1960 portion of the Briffa reconstruction was also deleted from the NCDC archive and the Climategate Letters, as previously noted, was the first digital “archive” of the post-1960 Briffa reconstruction used in TAR.)
As of Oct 5, 1999, the revised Briffa reconstruction had not been presented in any peer-reviewed literature but nonetheless was adopted by IPCC. The hasty recalculation of the Briffa reconstruction resulted in a big decline in the late 20th century – this is the decline illustrated in the graphic in my post.
In the First Order Draft of late October 1999, IPCC did not show the decline. In the Jones trick email two weeks later, as noted above, Jones hid the decline in a slightly different way.
Another issue raised by readers pertains to quotations. The post was already long and I tried to keep the quotations relatively concise. Some readers have criticized the ellipsis. I’ve accordingly amended the quotations (amendments in square brackets.)
Mann’s IPCC trick is related to the Jones’ trick, but different. (The Jones trick has been explained in previous CA postshere, here and consists of replacing the tree ring data with temperature data after 1960 – thereby hiding the decline – and then showing the smoothed graph as a proxy reconstruction.) While some elements of the IPCC Trick can be identified with considerable certainty, other elements are still somewhat unclear.
The diagram below shows the IPCC version of the Briffa reconstruction (digitized from the IPCC 2001) compared to actual Briffa data from the Climategate email of October 5, 1999, smoothed using the methodology said to have been used in the caption to the IPCC figure (a 40 year Hamming filter with end-point padding with the mean of the closing 20 years).
Figure 3. Versions of the Briffa Reconstruction in controversy, comparing the original data smoothed according to the reported methodology to a digitization of the IPCC version.
Clearly, there are a number of important differences between the version sent to Mann and the version that appeared in the IPCC report. The most obvious is, of course, that the decline in the Briffa reconstruction has, for the most part, been deleted from the IPCC proxy diagram. However, there are some other frustrating inconsistencies and puzzles that are all too familiar.
There are some more technical inconsistencies that I’ll record for specialist readers. It is very unlikely that that the IPCC caption is correct in stating that a 40-year Hamming filter was used. Based on comparisons of the MBH reconstruction and Jones reconstruction, as well as the Briffa reconstruction, to versions constructed from raw data, it appears that a Butterworth filter was used – a filter frequently used in Mann’s subsequent work (a detail that, in addition, bears on the authorship of the graphic itself).
Second, the IPCC caption stated that “boundary constraints imposed by padding the series with its mean values during the first and last 25 years.” Again, this doesn’t seem to reconcile with efforts to replicate the IPCC version from raw data. It appears far more likely to me that each of the temperature series has been padded with instrumental temperatures rather than the mean values of the last 25 years.
Finally, there are puzzling changes in scale. The underlying annual data for the Jones and Briffa reconstructions are expressed in deg C (basis 1961-1990) and should scale simply to the smoothed version in the IPCC version, but don’t quite. This may partly derive from errors introduced in digitization, but is a loose end in present replication efforts.
The final IPCC diagram (2.21) is shown below. In this rendering, the Briffa reconstruction is obviously no longer “a problem and a potential distraction/detraction”and does not “dilute the message”. Mann has not given any “fodder” to the skeptics, who obviously did not have a “field day” with the decline.
IPCC Third Assessment Report Figure 2.21: Comparison of warm-season (Jones et al., 1998) and annual mean (Mann et al., 1998, 1999) multi-proxy-based and warm season tree-ring-based (Briffa, 2000) millennial Northern Hemisphere temperature reconstructions. The recent instrumental annual mean Northern Hemisphere temperature record to 1999 is shown for comparison. Also shown is an extra-tropical sampling of the Mann et al. (1999) temperature pattern reconstructions more directly comparable in its latitudinal sampling to the Jones et al. series. The self-consistently estimated two standard error limits (shaded region) for the smoothed Mann et al. (1999) series are shown. The horizontal zero line denotes the 1961 to 1990 reference period mean temperature. All series were smoothed with a 40-year Hamming-weights lowpass filter, with boundary constraints imposed by padding the series with its mean values during the first and last 25 years.
Contrary to claims by various climate scientists, the IPCC Third Assessment Report did not disclose the deletion of the post-1960 values. Nor did it discuss the “divergence problem”. Yes, there had been previous discussion of the problem in the peer-reviewed literature (Briffa et al 1998) – a point made over and over by Gavin Schmidt and others. But not in the IPCC Third Assessment Report. Nor was the deletion of the declining values reported or disclosed in the IPCC Third Assessment Report. [Dec 11.- IPCC TAR does contain a sly allusion to the problem; it mentions "evidence" that tree ring density variations had "changed in their response in recent decades". Contrary to claims of realclimate commenters, this does not constitute disclosure of the deletion of the post-1960 values in the controversial figure or even of the decline itself.] The hiding of the decline was made particularly artful because the potentially dangling 1960 endpoint of the Briffa reconstruction was hidden under other lines in the spaghetti graph as shown in the following blow-up:
Figure. Blow-up of IPCC Third Assessment Report Fig 2-21.
To my knowledge, no one noticed or reported this truncation until my Climate Audit post in 2005 here. The deletion of the decline was repeated in the 2007 Assessment Report First Order and Second Order Drafts, once again without any disclosure. No dendrochronologist recorded any objection in the Review Comments to either draft. As a reviewer of the Second Order Draft, I asked the IPCC in the strongest possible terms to show the decline reported at CA here:
Show the Briffa et al reconstruction through to its end; don’t stop in 1960. Then comment and deal with the “divergence problem” if you need to. Don’t cover up the divergence by truncating this graphic. This was done in IPCC TAR; this was misleading. (Reviewer’s comment ID #: 309-18)]
Returning to the original issue: climate scientists say that the “trick” is now being taken out of context. The Climategate Letters show clearly that the relevant context is the IPCC Lead Authors’ meeting in Tanzania in September 1999 at which the decline in the Briffa reconstruction was perceived by IPCC as “diluting the message”, as a “problem”, as a “potential distraction/detraction”. A stone in their shoe.
Update (Dec 11) : Some of the follow-up comments on this post do shed light on this sequence and enable a more precise interpretation of the emails. With the benefit of these comments, there are a couple of points on the chronology that I need to modify, particularly in respect to the role of the October 5 revision of the Briffa reconstruction in respect both to the Arusha meeting and to the hide the decline.
The Arusha meeting objected to the Briffa reconstruction “diluting the message” and reducing confidence in the multiproxy reconstructions. And, of course, it is the overstated confidence that has been the primary objection here. However, I agree with critics who observe that the proximate objection to the Briffa reconstruction at Arusha was not that the decline per se diluted the message, but the Briffa reconstruction overall diluted the message and interfered with a “tidy story”. The stone in the shoe was that the Briffa reconstruction prevented a “tidy story”; the “decline” as a separate problem came a bit later.
After the Arusha meeting, Briffa hurriedly re-did his chronology and the new version was delivered to Mann on Oct 5, 1999 – it was this version that had the big decline. In the First Order Draft of Oct 27, 1999, IPCC author Mann deleted the post-1960 portion of the Briffa reconstruction plus other things that I don’t yet quite understand. Jones’ trick, as observed in the post, is a little different. (The post-1960 portion of the Briffa reconstruction was also deleted from the NCDC archive and the Climategate Letters, as previously noted, was the first digital “archive” of the post-1960 Briffa reconstruction used in TAR.)
As of Oct 5, 1999, the revised Briffa reconstruction had not been presented in any peer-reviewed literature but nonetheless was adopted by IPCC. The hasty recalculation of the Briffa reconstruction resulted in a big decline in the late 20th century – this is the decline illustrated in the graphic in my post.
In the First Order Draft of late October 1999, IPCC did not show the decline. In the Jones trick email two weeks later, as noted above, Jones hid the decline in a slightly different way.
Another issue raised by readers pertains to quotations. The post was already long and I tried to keep the quotations relatively concise. Some readers have criticized the ellipsis. I’ve accordingly amended the quotations (amendments in square brackets.)
