Climate Audit

March 15, 2013 was the IPCC deadline for use in AR5 and predictably a wave of articles have been accepted. The IPCC Paleo chapter wanted a graphic on regional reconstructions and the PAGES2K group has obligingly provided the raw materials for this graphic, which will be published by Nature on April 21. Thanks to an obliging mole, I have information on the proxies used in the PAGES2K reconstructions and will report today on the Gergis reconstruction, of interest to CA readers, which lives on a zombie, walking among us as the living dead.

The PAGES2K article has its own interesting backstory. The made-for-IPCC article was submitted to Science last July on deadline eve, thereby permitting its use in the Second Draft, where it sourced a major regional paleo reconstruction graphic. The PAGES2K submission used (in a check-kited version) the Gergis reconstruction, which it cited as being “under revision” though, at the time, it had been disappeared.

The PAGES2K submission to Science appears to have been rejected as it has never appeared in Science and a corresponding article is scheduled for publication by Nature. It sounds like there is an interesting backstory here: one presumes that IPCC would have been annoyed by Science’s failure to publish the article and that there must have been considerable pressure on Nature to accept the article. Nature appears to have accepted the PAGES2K article only on IPCC deadline eve.

The new PAGES2K article contains reconstructions for all continents and has an extremely long list of proxies, some of which have been discussed before, but some only now making their first digital appearance. Each regional reconstruction is a major undertaking and deserving of separate peer review. It seems impossible that these various regional reconstructions could themselves have been thoroughly reviewed as re-submitted to Nature. Indeed, given that the PAGES2K coauthor list was very large, one also wonders where they located reviewers that were unconflicted with any of the authors.

Of particular interest to CA readers is the zombie version of the Gergis reconstruction. Previous CA articles are tagged gergis.

CA readers will recall that Gergis et al 2012 had stated that they had used detrended correlations to screen proxies – a technique that seemingly avoided the pitfalls of correlation screening. Jean S pointed out that Gergis et al had not used the stated technique and that the majority of their proxies did not pass a detrended correlation test – see CA discussion here (building on an earlier thread) reporting that only 6 of 27 proxies passed the stated significance test.

Senior author David Karoly asked coauthor Neukom to report on correlations and, after receiving Neukom’s report, wrote his coauthors conceding the validity of the criticism:

Thanks for the info on the correlations for the SR reconstructions during the 1911-90 period for detrended and full data. I think that it is much better to use the detrended data for the selection of proxies, as you can then say that you have identified the proxies that are responding to the temperature variations on interannual time scales, ie temp-sensitive proxies, without any influence from the trend over the 20th century. This is very important to be able to rebut the criticism is that you only selected proxies that show a large increase over the 20th century ie a hockey stick .

The same argument applies for the Australasian proxy selection. If the selection is done on the proxies without detrending ie the full proxy records over the 20th century, then records with strong trends will be selected and that will effectively force a hockey stick result. Then Stephen Mcintyre criticism is valid. I think that it is really important to use detrended proxy data for the selection, and then choose proxies that exceed a threshold for correlations over the calibration period for either interannual variability or decadal variability for detrended data. I would be happy for the proxy selection to be based on decadal correlations, rather than interannual correlations, but it needs to be with detrended data, in my opinion. The criticism that the selection process forces a hockey stick result will be valid if the trend is not excluded in the proxy selection step.

Unfortunately, as coauthor Neukom immediately recognized a big probleM:

we don’t have enough strong proxy data with significant correlations after detrending to get a reasonable reconstruction.

Mann and Schmidt immediately contacted Gergis and Karoly advising them to tough it out as Mann had done with his incorrect use of the contaminated portion of the Tiljander data, where Mann’s refusal to concede the error had actually increased his esteem within the climate community. Nonetheless, Gergis and Karoly notified Journal of Climate of the problem. Despite Karoly’s concerns about substantive problems, Gergis hoped to persuade Journal of Climate that the error was only in their description of methodology and to paper over the mistake. However editor Chiang’s immediate reaction was otherwise, advising Gergis:

it appears that you will have to redo the entire analysis (and which may result in different conclusions), I will also be requesting that you withdraw the paper from consideration.

Upon receiving advice from Mann, Gergis tried to persuade Journal of Climate that the error was not one of methodology, but one of language only. Bur Chief Editor Broccoli was not persuaded, responding:

In that email (dated June 7) you described it as “an unfortunate data processing error,” suggesting that you had intended to detrend the data. That would mean that the issue was not with the wording but rather with the execution of the intended methodology.

Editor Chiang added:

Given that you had further stated that “Although it was an unfortunate data processing error, it does have implications for the results of the paper,” we had further took this to mean that you were going to redo the analysis to conform to the description of the proxy selection in the paper.

After further lobbying form Gergis, Chiang reluctantly permitted Gergis to re-submit as a “revision” by the end of July, but insisted that they show the results of both methods, describing this as an “opportunity” to show the robustness of their work:

In the revision, I strongly recommend that the issue regarding the sensitivity of the climate reconstruction to the choice of proxy selection method (detrend or no detrend) be addressed. My understanding that this is what you plan to do, and this is a good opportunity to demonstrate the robustness of your conclusions.

Gergis didn’t meet the July 31 deadline and Journal of Climate reported that the paper had been “withdrawn” by the authors.

The article was apparently resubmitted to Journal of Climate by the end of September, where, according to Gergis’ current webpage, it remains “under review”.

Nonetheless, the Gergis reconstruction has already been incorporated into the PAGES2K made-for-IPCC composite. CA readers will recall the Mole Incident in 2009. Once again, I am in possession of the proxy list used in the zombie reconstruction and can report that it has had only negligible changes.

On the left is a list of the 28 proxies used in the disappeared Gergis version, highlighting the proxies re-used in the zombie version. On the right is the list of proxies in the new version, highlighting the additions. 21 of 27 proxies are re-used. Six proxies have been excluded, while seven have been added. Remarkably, Gergis has kept the numbering as close as possible to the original list, so that the first 20 re-used proxies appear in the same order as in the original table.

The medieval portion of their reconstruction only has two proxies – as observed at CA very early here, where it was also pointed out that these two proxies did not constitute “new” information, as claimed in an IPCC draft, since they had not only been available for AR4 but illustrated in it.

Excluded from the original list are a tree ring series(Takapari), two ice core series (both from Vostok) and three coral series (Bali, Maiana and Fiji 1F O18), replaced by a speleothem (Avaiki), three tree ring series (Baw Baw, Celery Top West, Moa Park), two coral luminescence series (Great Barrier Reef, Havannah) and a coral O18 series (Savusavu). None of the excluded or included series is particularly long.

Obviously Gergis et al have not “redone the analysis to conform to the description of the proxy selection in the paper” as they continue to use many of the proxies that failed the original significance test – see the graphic below from last June.

Nature reviewers obviously didn’t have the concerns about robustness that were expressed by Journal of Climate editors last summer, as the new article doesn’t demonstrate any such “robustness”. It will be interesting to see whether Journal of Climate editors will themselves adhere to the scruples that they showed last summer and require Gergis et al to demonstrate the robustness of their reconstruction.