The Yang Chinese composite, after the Mann PC1 and Yamal, had the third-largest hockey stick shape of the proxies illustrated in the IPCC AR4 spaghetti graph. I’d commented on this series on several occasions – see http://www.climateaudit.org/tag/yang
The new emails show that Bradley thought that this series was, to use the technical term preferred by climate scientists, “crap” and should not be used in multiproxy studies – an issue raised by Bradley in connection with Mann et al (EOS 2003) – their attack on Soon and Baliunas 2003.
Needless to say, Bradley did not publish a comment criticizing the use of this series. It has subsequently been used over and over again in IPCC multiproxy studies, commencing with Mann and Jones 2003.
In my post a few years ago, I observed that it was, in fact, “the most heavily-weighted contributor to Mann and Jones  … The Yang composite and the North American PC1 (bristlecones) dominate the Mann and Jones  reconstruction, making other series essentially irrelevant.” It was then used in Moberg et al .
In my earlier post, I observed that “its contribution to Moberg is less marked, but it is one of only 3-4 series that provide a strong 20th century” (the other series include upside-down Bulloides.) Since then, it has been used in Osborn and Briffa 2006, the IPCC AR4 spaghetti graph (Box 6.4 Figure 1), Hegerl et al 2007, Juckes et al 2007, Ljungqvist 2010 and even Loehle and McCulloch 2010.
Here’s what Bradley and the Team said about the validity of this series behind closed doors. The exchange also has an interesting vignette on the speed of peer review when the Team is involved.
The Counterattack against Soon and Baliunas
On June 3, 2003, Mann began drafting the counterattack on Soon and Baliunas for EOS, assisted by Jones, Osborn and Rutherford. Authors who commented at least once were Bradley, Briffa, Crowley, Oppenheimer, Trenberth and Wigley. Ammann and Hughes appear to have been listed as coauthors only as a courtesy, rather than because of actual input.
There are emails through June 13 discussing the drafting of the EOS article. The version as submitted to EOS for peer review did not include the Yang composite. It had the same number of series but used the Briffa 2000 composite instead.
In email 2530 on June 17 at 10:53 GMT, Jones tells Mann that EOS editor Moseley-Thompson has the article for review:
I’ve not heard any more about the EOS piece but Ellen has got it – I got an email from her to Judy….
They had just received notice of acceptance of Mann and Jones 2003 (the article that Bradley later disliked so much). Jones suggested to Mann that they replace the long Briffa series with Yang’s more hockey-stick shaped series (which they had used in Mann and Jones 2003):
I would suggest with EOS we add this series into Fig 1, back to AD200, possibly by replacing the long Briffa series.
In the rest of email 2530, Mann and Jones separately instructed Rutherford to switch the series in EOS Figure 1, with Jones saying that he would check with Moseley-Thompson to see if possible. (None of the other authors appear to have been consulted thus far.)
I’m off home now. Do you want to see if you can switch the two series around as Mike suggested. Replace the long Briffa one with the appended and alter caption accordingly. I’ll email Ellen and Judy to see if possible.
The next morning, (email 2-3637), Jones reported to the Team that the article was with AGU and should go soon to Ellen Moseley-Thompson for reviewing (Moseley-Thompson having participated in the earlier email chain discussing what to do about Soon and Baliunas.) He reported the acceptance of Mann and Jones 2003 and outlined the plan discussed the previous day with Mann to substitute the more hockey-stick shaped Yang series for the long Briffa series:
Here’s a brief update on the EOS article. It is currently with AGU and should go soon to Ellen Mosley-Thompson for assessment/reviewing. Mike and I are trying to co-ordinate its hopeful publication with the attached. This is the GRL paper that Mike has mentioned. Copy is for your info, so don’t pass around. Both reviews were positive and the attached is the resubmitted version. If co-ordination isn’t possible we will still replace the long Briffa et al series (going back over the 2 millennia) in Figure 1 with the blue line from Figure 2a in the GRL article. Text will alter, but only to refer to the new curve.
Jones explained that the substitution “should increase the impact”:
I’m in discussion with AGU and Ellen about co-ordination as this should increase the impact of both pieces. Mike or I will let you know when we hear more.
Later that day (JUne 18 – email 2530), Rutherford reported the completion of the substituted figure, which Jones distributed to Briffa and Osborn.
In the early EDT afternoon (email 2670) of the same day, EOS accepted the article, with a number of minor text changes suggested.
Bradley complains about the proxies
On June 22, four days later, (email 4207), Bradley complained to the team [emphasis added] about the Yang composite that Mann and Jones had unilaterally substituted.
You commented that the Chinese series of Yang et al (GRL 2002) looked weird. Well, that’s because it’s crap–no further comment on what stuff gets into GRL!
You appear to have used their so-called “complete” China record. You really should
consider what went into this –2 ice core delta 18O records of dubious relationship to
temperature (one is cited as correlating with NW China temperatures at r=0.2-0.4), 3
tree ring series, one of which is a delta C-13 record of questionable climatic
significance (to be generous). The other series include two records from a Taiwan
lake–a carbon/nitrogen isotope and a total organic carbon series (interpreted as
high=”warm, wet”) and an oxygen isotope series from cellulose in peat!!! (& don’t ask
about the C-14 based chronology, interpolated to decadal averages!)
I loved this sentence:
“Although a quantitative relationship between the proxy records of the Jinchuan peat,
the Japan tree-ring series and the Taiwanese sediment records with modern climate data
are not given in the original works, the qualitative connectivity with temperature as
the dominant controlling factor has undoubtedly been verified”
Oh, undoubtedly!! And these are 4 of the 9 series going into the “complete China”
Finally, they use another record based on “phenology” and (somehow) this provides a
winter temperature series….
You just shouldn’t grab anything that’s in print and just use it ‘cos it’s there—that
just perpetuates rubbish. This series needs to be removed from Figure 2 in the EOS
forum piece–and if you included it in your GRL paper, I suggest that you reconsider it.
Bradley’s complaint about the loose treatment of warm and wet in Yang had particular resonance because that was the criticism that the Team was levelling against Soon and Baliunas.
Mann replied immediately assuring Bradley and the inside Team that the Yang composite got “moderately low” weight in Mann and Jones 2003 (note- I don’t think that this is true. My notes indicate that it received more weight than any series other than the Mann PC1.) Mann tried to fob off further discussion until he met with Jones in Sapporo (by which time the article would have appeared):
In our GRL article, Phil and I weighted the records we used with respect to their decadal correlations with the instrumental gridpoint surface temperature data for the same region (numbers in parentheses in attached figure 1 from the paper), so if a series is truly crap in an objectively determined sense, it got very low weight. The China series has a reasonable (r=0.22), but not great correlation–and it gets a moderate low weight.
In my opinion, this is a better approach then simply deeming a record crap a priori (and then getting criticized for not considering it). We considered all available records with appropriate resolution that are putative temperature estimates, and weighted them objectively.
We also did careful cross-validation on the resulting reconstruction using independent instrumental data, etc.—so I hardly think we are subject to criticism in how we used the available data, relative to other analyses that have been done…
As for the Eos piece, I think a similar point holds–not showing it at all would seem a conspicuous omission. We could add the local correlation values to each of the panels of Figure 2, and comment briefly–this could be done at the proof stage.
Phil and I can discuss this, if need be, when we meet in Sapporo in a couple weeks,
On June 24 (the article having been accepted), Wigley, in email 4249, asked Osborn to “cover” Bradley’s point about the Yang series:
I think it is *extremely* important to cover Ray’s point about Yang et al. and Mike Mann’s response about weighting. This requires a small addition to the Figure caption.
Briffa, who had thus far been uninvolved, then weighed in (email 5027) to Bradley agreeing with his criticisms of the Yang composite. Briffa anticipated that its inclusion would be criticized by sceptics. However, Briffa argued that trying to cooper up the “suspicious” Chinese series at this stage would delay matters and tried to persuade Bradley to drop his scruples on the basis that they were not endorsing the individual series:
thanks for the communication –
Now to the comments re the EOS piece. I believe you criticised the inclusion of the 2000 (Eurasian ) tree-ring series (since reiterated by Malcolm). Fair enough , though again misguided in my opinion if on the basis of “contains few data ” or ” has weak climate response” . I was perfectly happy to drop it ( I never suggested its inclusion in the first place), but I find it somewhat ironic that it should be replaced with the latest (Mann and Jones) series that contains the same three series plus a mixture of other far more dubious (not to say bad ) series – I agree with the remarks you made re some of these (particularly the Chinese series) in your recent email to someone. I consider that this new series (plus the illustration of the Western US series in the EOS) piece will “stimulate further discussion ” in the field , both between we palaeo-types and the Sceptics.
I and Tim have been left to submit this and the balance of pressure seems to be to submit as is – if we remove the suspicious Chinese series we would have to delay things further (Ellen is hassling for us to submit) and , anyway, it is still contained in the Long series. I am of the opinion that the points made in the piece still stand – and by signing on, we are not individually sanctioning all the curves or data used in the illustrations ( There are genuine problems with ALL of them). We will therefore , add Malcolm’s name and submit the version we now have. Hope this OK with all.
Mann proposed that they deal with things at proof stage, saying that the “low weight” given to the Yang composite dealt with it “appropriately”. (The correlations which Mann said would be added in proof stage were not added.)
I suggest we let Eos size the figures, etc. Then, in the end, we can simply substitute a version of Figure 2 w/ the correlations added at the proof stage. Anything else will slow down the publication of the manuscript unnecessarily, in my opinion. Phil and I have already discussed–we agree that the low weight given to the record in the Mann and Jones composite treats the record appropriately…
Briffa also tried to individually reassure Wigley (2023) on June 24 about the Yang series:
Tim has just told me of your message expressing concern about the China series , and your statement of the necessity to “deal with Ray’s comment” and add in the “small adjustment to the Figure Caption”. .
We (I and Tim) decided to get this off as soon as possible to Ellen (AGU), as we had been asked to do (and as requested by Ellen). Hence it went off earlier today (and before your message arrived). Mike was aware of Ray’s comment and was happy to leave any amendment to the text “until the proof stage” .
In my opinion it is not practical (or desirable) to try to “qualify” any one record in this limited format. It was a majority decision to leave the Mann and Jones 2000-year series in the Figure 1 (as it was to remove the Briffa and Osborn tree-ring based one) , and the details of the logic used to derive the Mann and Jones series is to be found in the (cited) text of their paper. Signing on to this letter, in my mind. implies agreement with the text and not individual endorsement of all curves by each author. I too have expressed my concern to Phil (and Ray) over the logic that you leave all series you want in but just weight them according to some (sometimes low) correlation (in this case based on decadal values). I also believe some of the series that make up the Chinese record are dubious or obscure, but the same is true of other records Mann and Jones have used (e.g. how do you handle a series in New Zealand that has a -0.25 correlation?). Further serious problems are still (see my and Tim’s Science comment on the Mann 1999 paper) lurking with the correction applied to the Western US tree-ring PC amplitude series used (and shown in Figure 2). There are problems (and limitations ) with ALL series used. At this stage , singling out individual records for added (and unavoidably cursory added description) is not practical. We were told to cut the text and References significantly – and further cuts are implied by Ellen’s messages to us.
If you wish to open this up to general discussion , it may be best to wait ’til the proof stage and then we can all consider the balance of emphasis – but we had also better guard against too “selective” a choice of data to present? If you want to get a somewhat wider discussion of this point going in the meantime, feel free to forward this to whoever you wish along with your disagreement, while we wait on the response from AGU.
Osborn also tried to reassure Wigley on June 24 (email 4249):
In Phil’s absence I was just now looked at his PC because I needed some files/emails for a separate matter, and I noticed that you had emailed Phil/Ray/Mike concurring with Ray’s concerns. Until I saw that, I hadn’t realised that anyone else had commented on Yang et al.
Keith and I discussed exactly this issue this morning, and though Keith also had concerns about the record (I haven’t read their paper, so can’t comment) we decided to leave things as they were because: (i) Mike suggested adding correlations to the figure at the proof stage rather than now; (ii) I wasn’t sure how to word a caveat about Yang et al. without making it seem odd that we were including a doubtful record and odd that we hadn’t added caveats about some of the other records.
The current status is that the version I circulated has been submitted back to EOS (because of the reasons given above), and Ellen Mosley-Thompson has approved it. It needs to be reviewed internally at AGU by either Fred Spilhaus or an Associate Editor. It will then be edited to reflect the Eos newspaper style.
I’ve cc’d this to Mike and Phil to see what they want to do. I/we can put a hold on the processing of the current submission and then submit a new version with revised figure and caption. Alternatively we could wait and see what it’s like after EOS have edited it, and then make any final modifications at that stage.
Over to you/Mike/Phil.
The following day (June 25 – email 4712), Wigley clarified that he was not really trying to ensure that the individual series had merit, but only that they chose “sufficiently ambiguous” language in their text to give them a fallback position if they were criticized – a practice that one sees all too often in this field,
It seems that there was a misunderstanding about what I suggested re Yang. To be more specific, I suggest adding the following to the end of the Figure 2 caption:
“….. Note that individual series are weighted according to their quality in forming a composite hemispheric-scale time series.”
The word ‘quality’ here has been chosen carefully — as something that is deliberately a bit ambiguous.
The point here is to have something that we can fall back on if anyone criticizes *any* specific input series (*not* just Yang).
Wigley then responded to Briffa’s individual email with inline comments (539. 2003-06-25):
Tim has just told me of your message expressing concern about the China series , and your statement of the necessity to “deal with Ray’s comment” and add in the “small adjustment to the Figure Caption”. .
We (I and Tim) decided to get this off as soon as possible to Ellen (AGU) , as we had been asked to do (and as requested by Ellen). Hence it went off earlier today (and before your message arrived). Mike was aware of Ray’s comment and was happy to leave any amendment to the text “until the proof stage” .
YEAH, I REALIZE THIS — AND I AGREE THAT IT WAS IMPORTANT TO GET THE DOCUMENT OFF
In my opinion it is not practical (or desirable) to try to “qualify ” any one record in this limited format. It was a majority decision to leave the Mann and Jones 2000-year series in the Figure 1 (as it was to remove the Briffa and Osborn tree-ring based one) , and the details of the logic used to derive the Mann and Jones series is to be found in the (cited) text of their paper.
YOU MISUNDERSTAND ME. OF COURSE IT WOULD BE SILLY TO SINGLE OUT A SPECIFIC ITEM. WHAT IS NECESSARY IS A SENTENCE STATING THE *METHOD* — I.E., THAT ITEMS ARE WEIGHTED BY THEIR CALIBRATION PERFORMANCE.
Signing on to this letter , in my mind. implies agreement with the text and not individual endorsement of all curves by each author. I too have expressed my concern to Phil (and Ray) over the logic that you leave all series you want in but just weight them according to some (sometimes low) correlation (in this case based on decadal values). I also believe some of the series that make up the Chinese record are dubious or obscure , but the same is true of other records Mann and Jones have used (e.g. how do you handle a series in New Zealand that has a -0.25 correlation?) .
IT IS A DIFFICULT CALL — WHETHER TO DUMP SERIES THAT HAVE NO SIGNIFICANT LINK TO TEMPERATURE AND WHICH ARE, AS WELL, DUBIOUS ON A PRIORI GROUNDS; OR TO USE A WEIGHTING SCHEME. IF ONE DID THIS BY SIMPLE MULTIPLE REGRESSION, THEN THINGS WOULD BE WEIGHTED AUTOMATICALLY. HOWEVER, STATISTICALLY ONE SHOULD STILL DUMP THE LOW CORRELATION ONES. I HAVE RESERVATIONS ABOUT WHAT MIKE AND PHIL HAVE DONE — BUT THIS IS SOMETHING WE SHOULD TALK ABOUT FACE TO FACE SOME DAY.
Further serious problems are still (see my and Tim’s Science comment on the Mann 1999 paper) lurking with the correction applied to the Western US tree-ring PC amplitude series used (and shown in Figure 2). There are problems (and limitations ) with ALL series used.
At this stage , singling out individual records for added (and unavoidably cursory added description) is not practical.
I AM NOT SUGGESTING THIS — AS THE ABOVE SHOULD MAKE CLEAR.
We were told to cut the text and References significantly – and further cuts are implied by Ellen’s messages to us. If you wish to open this up to general discussion , it may be best to wait ’til the proof stage and then we can all consider the balance of emphasis – but we had also better guard against too “selective” a choice of data to present? If you want to get a somewhat wider discussion of this point going in the meantime , feel free to forward this to whoever you wish along with your disagreement , while we wait on the response from AGU.
NO — I’M HAPPY WITH KEEPING THINGS AT THIS LEVEL.
The Mann et al EOS article was published on July and was followed soon after by mass resignation of Climate Research editors for the role of the journal in publishing Soon and Baliunas. More on this on another occasion.
Later in the fall of 2003, Briffa and Cook reflected on Mann and Jones 2003 and the Yang composite. In email 435, Cook proposed to Briffa that they try to summarize what was actually known about 1000-year reconstructions, complaining that Bradley’s “air of papal infallibility is really quite nauseating at times” – a point on which both warmist and skeptic could agree. He worried that Jones and Mann were “too personally invested” now:
I am afraid the Mike and Phil are too personally invested in things now (i.e. the 2003 GRL paper that is probably the worst paper Phil has ever been involved in Bradley hates it as well), but I am willing to offer to include them if they can contribute without just defending their past work
On October 17, 2003, Bradley, Hughes and Diaz published their contribution to the rebuttal of Soon and Baliunas. Recall that Bradley had complained that that Jones and Mann had failed to consider what went into the Yang composite “2 ice core delta 18O records of dubious relationship to temperature”. These series were, of course, versions of the Dunde and Guliya series popularized by Lonnie Thompson. In Bradley et al 2003, they were not described as having a “dubious relationship to temperature”, but as “well calibrated” data sets.
As noted above, despite the private misgivings of the various authors expressed prior to publication of Mann et al 2003, Osborn and Briffa 2006 used the Yang composite anyway (justifying its inclusion on the basis that it had been used in Mann et al, 2003.)
Similar rationales were used for its inclusion in the IPCC 2007 Box 6.4 Figure 1 without overturning Bradley’s objection that such re-use simply “perpetuates rubbish.”
Update: Dec 3, 2011. Jean S has drawn attention to the fact that differnt online versions of this article show precisely what changes were made. The earliest version is the one at von Storch’s website here. It has the Yang composite in Figure 2 and the Briffa 2000 reconstruction in Figure 1 (Hughes not an author.) The next June 20) version is online at Stephen Schneider’s website here: it replaced the Briffa 2000 reconstruction in Figure 1 with the Mann and Jones 2003 reconstruction; it still had the Yang composite in Figure 2 and still failed to show Hughes as a coauthor. Hughes was added as a coauthor in the next version (here and as published here), otherwise things appear to be unchanged.