This post will be a sticky top post for a day or two, new stories will appear below this one.
By Andrew Montford & Harold Ambler
May 24, 2012 4:00 A.M. in the National Review – reposted here with permission
Climategate, the 2009 exposure of misconduct at the University of East Anglia, was a terrible blow to the reputation of climatology, and indeed to that of British and American science. Although that story hasn’t been in the news in recent months, new evidence of similar scientific wrongdoing continues to emerge, with a new scandal hitting the climate blogosphere just a few days ago.
And central to the newest story is one of the Climategate scientists: Keith Briffa, an expert in reconstructing historical temperature records from tree rings. More particularly, the recent scandal involves a tree-ring record Briffa prepared for a remote area of northern Russia called Yamal.
For many years, scientists have used tree-ring data to try to measure temperatures from the distant past, but the idea is problematic in and of itself. Why? Because tree-ring data reflect many variables besides temperature. Russian tree growth, like that of trees around the world, also reflects changes in humidity, precipitation, soil nutrients, competition for resources from other trees and plants, animal behavior, erosion, cloudiness, and on and on. But let’s pretend, if only for the sake of argument, that we can reliably determine the mean temperature 1,000 years ago or more using tree cores from a remote part of Russia. The central issue that emerges is: How do you choose the trees?
It was the way Briffa picked the trees to include in his analysis that piqued the interest of Steve McIntyre, a maverick amateur climatologist from Canada. The Climategate e-mails make it clear that McIntyre earned the public scorn of the most powerful U.N. climatologists, including James Hansen, Michael Mann, and Phil Jones, while simultaneously earning their fear and respect in private.
McIntyre noticed a few problems with the way Briffa chose the sampling of Russian trees, and he wrote to Briffa requesting the data Briffa used in a published tree-ring paper. Briffa declined. And so began a four-year saga involving multiple peer-reviewed journals, behind-the-scenes maneuvering by Briffa and his closest confidants, and a Freedom of Information Act request on the part of McIntyre that appears to be on the verge of being granted. Even without the final set of data, however, McIntyre has shown beyond the shadow of doubt that Briffa may have committed one of the worst sins, if not the worst, in climatology — that of cherry-picking data — when he assembled his data sample, which his clique of like-minded and very powerful peers have also used in paper after paper.
It was already known that the Yamal series contained a preposterously small amount of data. This by itself raised many questions: Why did Briffa include only half the number of cores covering the balmy interval known as the Medieval Warm Period that another scientist, one with whom he was acquainted, had reported for Yamal? And why were there so few cores in Briffa’s 20th century? By 1988, there were only twelve cores used in a year, an amazingly small number from the period that should have provided the easiest data. By 1990, the count was only ten, and it dropped to just five in 1995. Without an explanation of how the strange sampling of the available data had been performed, the suspicion of cherry-picking became overwhelming, particularly since the sharp 20th-century uptick in the series was almost entirely due to a single tree.
The intrigue deepened when one of the Climategate e-mails revealed that, as far back as 2006, Briffa had prepared a much more broadly based, and therefore more reliable, tree-ring record of the Yamal area. But strangely, he had decided to set this aside in favor of the much narrower record he eventually used.
The question of Yamal had rightly come up when Briffa was questioned by Climategate investigators. He told them that he had never considered including a wider sample than the one he went with in the end, and hadn’t had enough time to include a wider one. However, the specific issue of the suppressed record appears to have largely been passed over by the panel, and Briffa’s explanation, like so many others given to the Climategate inquiries, appears to have been accepted without question.
But the ruse has now been shot to pieces, by the recent decision from the U.K.’s information commissioner that Briffa can no longer withhold the list of sites he used in his suppressed regional record for the Yamal area. The disclosure of these sites has allowed McIntyre to calculate what the broad series would have looked like if Briffa had chosen to publish it. He has shown that it has no hint of the hockey-stick shape that Briffa’s cherry-picked data indicated.
Two and a half years after the initial revelation of the Climategate e-mails, new controversies, on the part of the scientists and the investigators involved, continue to emerge. Many of the players involved are desperate to sweep the scandal under the rug. However, their machinations have only succeeded in bringing renewed attention to their questionable science and ugly behind-the-scenes shenanigans, reigniting hope that more complete and more independent investigations — on both sides of the Atlantic — will yet be performed.
— Andrew Montford is the author of The Hockey Stick Illusion and the proprietor of the Bishop Hill blog. Harold Ambler is the author of Don’t Sell Your Coat and the operator of the blog talkingabouttheweather.com.
First sentence: “Climategate, the 2009 exposure of misconduct at the University of East Anglia, was a terrible blow to the reputation of climatology, and indeed to that of British and American science.”
It is appropriate to make distinction between science and scientists. I frequently hear the argument that those who don’t believe in CAGW don’t believe in science when it is scientists misrepresentation of science that is at question. That distinction needs to be made more often and more forcefully.
Latitude says:
May 25, 2012 at 9:47 am
“…is there anyone that believes you can get temperatures from tree rings?”
To take a tree ring core the ‘dendroclimatologist” has to drill a hole in the tree and extract the core. Maybe then he – or she – would be better off just throwing away the core and sticking a thermometer into the tree and measuring the temperature inside. At least that would be a true value.
Well actually, the full Yamal dataset may yet turn out to be a hockey stick. Depending on how the past few years of ringing resolve, it may prove to a hockey stick with the blade pointing DOWN!
lazyteenager;
along the way because 
tells you whether or not there is any sort of credible linear relationship to be found.
I was under the impression that the trees chosen for analysis were geographically situated to avoid these kinds of extraneous influences and to be especially sensitive to temperature only.>>>>
Could you please describe the conditions and/or evolution that would result in a tree being insensitive to changes in precipitation? Pestilence? Disease? Animals foraging? Late frost?
And also, could you explain how the sensistivity to winter temperatures is incorporated into the tree rings?
David’s point is well-made, Lazy. The fundamental problem is that you, or anyone, would use the words chosen for some property regarding data that is the basis of a statistical study!
A very short lesson on statistical analysis. You have a dataset. You hypothesize “Widgets are a (linear, quadratic, logistic, gaussian, smooth) function of Frobbits” within the dataset (which is highly multivariate — Widgets might depend on Cogs as well in nontrivial ways, we just don’t know).
The default action — one that any scientist or analyst departs from only if they are in peril of their soul or are Bayesian divinities in possession of ex dei-quality Bayesian priors — is to take the goddamn data, put Frobbits on one axis, Widgets on another, look at the distribution of the scatter plot, and use regression to fit a (linear, quadratic… whatever) relationship, taking care to compute
What you do not do — again, unless you are in league with demonic forces — is pick a subset of the data to use when doing it. And here’s why: Because there are many other possibly confounding influences that might have affected Widgets, and you do not know what they are! Indeed, to know them (from the Widget dataset itself) you have to have already solved a more difficult problem than the one you are trying to solve. To correct for Cogs (which also affect Widgets) you have to have obtained things like the probability of Widgets given Cogs, and you really can’t get more information than now just plotting and analyzing Widgets as a multivariate probability distribution over Cogs and Frobbits. The only place where confounding can be corrected at the level of Widgets and Frobbits directly is to use exterior priors, exterior meaning that they come from an analysis using something (well, conditional probabilities) that you profoundly trust and that don’t come from the data you are already working with.
And that’s the rub. The entire tree-ring PCA statistical voodoo involves matching a “signal” evident in the data relating ring width to supposedly known temperatures and then locking in that signal in such a way that it permits past temperatures to be reconstructed. We know the late 20th century had a “linear” warming trend of thus and such — if we pick out tree ring series that exhibit the same linear trend over the same period, the idea is that these trees are the ones so situated that confounding influences “don’t work”. They are the “safe” trees to use, the ones that presumably encode the desired past signal with minimum “noise”. Mann and Briffa’s methodology already heavily weights such trees, which is why it disproportionately produces hockey stick shapes — it amplifies trees with a strong linear late 20th century blade trend, while leaving all of the flat uninteresting trees to produce the long stick.
Even so, if one applied this algorithm symmetrically all would be — not “good”, but perhaps still honest. That’s because trees with a strong linear trend up would be weighted the same as trees with a strong linear trend down across the proxy match region, and once again the laws of iid samping and unbiased inclusion of data would save the day as the ups and downs partially cancel, leaving (from enough samples, subject to a zillion unlikely hypotheses — I don’t think much of this approach) at least an unbiased result. Sadly, we have the twin peaks of “hide the decline” and now Yamal — not even this flawed algorithm with its largely unjustifiable steps can get an unbiased result when one deliberately leaves out series with a strong negative trend, and no algorithm is going to work if you choose the goddamn data you feed into it on any basis but “unusable” or the failure of statistical common sense (data way out of the normal bounds of nearly all of the other samples).
I “do” random number generator (hypothesis) testing and Monte Carlo. Suppose you have a source of perfectly random digits in the range 1-5. You turn it on and out pops the five-digit sequence 1 2 3 4 5. That doesn’t look too random to you, so you frown and set that number aside. Next out is 42122. Better, although having three 2s worries you. 32541 — now that looks random, we’ll put this in a pile to definitely use. 33322 — only two of the five numbers? Doesn’t look that random to me — reject.
After a few minutes of doing this sort of thing — applying any accept/reject heuristic to the data — you will have taken a perfectly good sequence of random digits, one where strings like 1 2 3 4 5 must occur with an easily computable probability (on average) and transformed it into something that looks “more random” to you but which would almost instantly fail statistical tests of randomness. And this, my friend, is the bete noir of both medicine and climate science — confirmation bias influencing data selection. If you reject all pieces of wood that don’t look like an elephant from a dataset of wood, eventually you can safely conclude that all trees are shaped like elephants. It won’t be a valid conclusion — a glance at a forest confounds it — but somewhere out there there are likely enough to be at least a few trees that do, and that’s all that you need.
This is why (I think) Steve Mcintyre twigged to MBH in the first place. If you just take the damn normalized/scaled tree ring data and average the regionally coarse-grained and surface-projected timeseries, you get what is probably already your best possible statistical result. Straight-up averaging of data means “I don’t know what might confound the result I hope to find, and using the (fairly weighted) average of everything averages in particular over my ignorance.” But when you do that with the original tree ring data (or the newer Yamal series plotted above) the noise leaves you with little discernible trend. Which is then, most likely (in a literal sense) the right answer. Little discernible trend.
Taking (really) trendless data and extracting a trended signal that just happens to agree with your prior beliefs and has the endless benefit of an eternity of grant renewals, pieces of Nobel Prizes, and a certain amount of notoriety if not fame — only the Devil could offer a deal like that one. Hence my initial observation — a scientist or statistician that departs from the rule of use all of the data including the part that disagrees with your personal biases or confounds your expected results does so only in peril of their soul.
rgb
Lazy Teenager says
“I was under the impression that the trees chosen for the analysis were geography situated to avoid these types of extrenous influences….”
* * *
Now now, let’s not be too hard on the LT. In fact, I’ve found exactly the tree he was talking about!
Its plastic and sits in my living room.
not for this thread but wanted to share my experience with the website “Climate Progress”, they have deleted some of my comments which in fact were more cut and pastes from NOAA about the seas and the correct term not becoming more acidic but “less base” and from the US water board about the PH of the waters around the US .
edited by JOE ROMM- enough said!!!!
I may be stupid, but can someone please explain how tree rings from one tiny area of the world show global warming (or cooling). Surely the tree ring data only shows what happens locally and then only during the growing season. I appreciate that the rings will reflect the length of the growing season, but they will also indicate other factors, like rainfall, CO2 concentration in the air, nitrates and phosphates etc. The temperature during winter could have fallen drastically, which would reduce the average annual local temperature significantly, so even if the width of the tree rings only demonstrated temperature, it would still not be a guide to average temperatures all year round.
Is there something I am missing??
Just a few corrections:- Briffa did not deline the data – it was not his to share and so he politely informed McIntyre that he would pass the request onto the Russian scientists, one of whom, Rashit Hentemirov, later wrote in response to McIntyre whining about 10 years of ‘stonewalling’:
Steve has an amnesia. I had sent him these data at February 2, 2004 on his demand.
Ooops. The climategate mail did not reveal that ‘Briffa had prepared a much more broadly based, and therefore more reliable, tree-ring record of the Yamal area.’, it was a list of areas and grid boxes in which a reconstruction may be attempted. Ooops.
The graph, of course, cannot be ‘what the broad series would have looked like if Briffa had chosen to publish it.’ as it is based on more modern data supplied to McIntyre by Hantemirov.
Indeed, when Hantemirov, a professional dendrochronologist, saw what McIntyre had done with his valued data he pronounced himself ‘horrified’ by the ‘slipshod’ work, accused McIntyre of not performing due diligence, and characterised him as careless, grubby and dishonourable. Somehow this detail didn’t make it into Montford’s account, I can’t think why.
Remove these inaccuracies, the baseless speculations (‘Many of the players involved are desperate to sweep the scandal under the rug.’) and the weasel words (‘McIntyre has shown beyond the shadow of doubt that Briffa may have committed one of the worst sins’) etc and there’s not a lot left. Clearly anyone turning to Montford for a factual, complete and balanced account is making a category error. But we knew that.
The idea that the media will be going anywhere near this smelly slice of tripe is delusional.
Poems are made by fools like me, but only Mann can fake a tree
If this were a just and honest world, McIntyre would win a Nobel Prize and the thanks of every person.
rgbatduke – “… does so only in peril of their soul.”
Nicely done, you knocked that one out of the park!
[my bold]
The problem is Lazy how do we know? Suppose YAD061 was under a much larger neighbouring tree which suddenly died. Would YAD061 experience increased growth? Do the same trees in the same are respond differently to temperature? YAD061 sure did. Does this not strike you as odd given your statement?
As usual the authors of the article in the newspaper did NOT put up THE GRAPH which you have put up here (ie Yamal versus rest). The story is in the graph, without the graph no one will read the rest except some diehard skeptics like us. In the future when talking about Yamal ALWAYS put the graph there its irreplaceable
LazyTeenager says:
May 24, 2012 at 11:12 pm
(Quoting the article) “Because tree-ring data reflect many variables besides temperature. Russian tree growth, like that of trees around the world, also reflects changes in humidity, precipitation, soil nutrients, competition for resources from other trees and plants, animal behavior, erosion, cloudiness, and on and on.”
—————
I was under the impression that the trees chosen for analysis were geographically situated to avoid these kinds of extraneous influences and to be especially sensitive to temperature only.
Is Montfort unaware if this or is he just being sneaky?
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
If the only thing that influenced the trees’ growth was temperature (no water, not nutrients, etc.), wouldn’t you have a bunch of dead trees?
Are Briffa and Mann weather-tree channelers or weather-tree mediums?
Phil C: translation::”There really worried about this story beacuse of DE facto evidence.
Another thing if you followed the Astronauts statements at the ICCC conf please note that its was emphasized to not use the words “Climate Change” ever but “Human caused global warming”, which was the ORIGINAL theme changed by the AGW ers when they realized it may not warm. We are palying into their game by using CC.
LazyTeenager says:
May 24, 2012 at 11:12 pm
Because tree-ring data reflect many variables besides temperature. Russian tree growth, like that of trees around the world, also reflects changes in humidity, precipitation, soil nutrients, competition for resources from other trees and plants, animal behavior, erosion, cloudiness, and on and on.
—————
I was under the impression that the trees chosen for analysis were geographically situated to avoid these kinds of extraneous influences and to be especially sensitive to temperature only.
Is Montfort unaware if this or is he just being sneaky?
********************************************************************************************************************
I’m just curious; how could those influences have been avoided? Any ideas at all?
REPLY: Well, he’s “lazy”, what do you expect? Since he’s too lazy to do the work to figure this out himself, I’ll point out that the issue has already been covered on WUWT, and there’s no such thing supporting “lazy’s” assertions about “the trees chosen for analysis were geographically situated to avoid these kinds of extraneous influences and to be especially sensitive to temperature only.” Liebigs law makes this impossible.
http://wattsupwiththat.com/2009/09/28/a-look-at-treemometers-and-tree-ring-growth/
“lazy” is just an anonymous troll from Australia whom I think works for John Cook in a paid disinformation position to post here. Ignore him, his opinions are meritless – Anthony
Lazy Teenager says
“I was under the impression that the trees chosen for the analysis were geography situated to avoid these types of extrenous influences….”
=========================
It doesn’t matter where they are sitting….
Trees grow in the middle band of temperature…..too cold, they don’t grow….too hot they don’t grow
All a tree can tell you is how long or short that middle band was……
Say a tree grows between 50 and 80 degrees…….can it tell you the range of temp was -10 – 80?
or was it 50 – 110? was it an ice age and temps ranged from -10 – 80? or was it warming and temps ranged from 50 – 110?
@ur momisugly Phil Clarke….References please??
So you agree that the data McIntyre requested was NOT the same as the Hantemirov data. Therefore, McIntyre was never supplied with the data used by Briffa.
Thanks for playing.
So essentially there was –
One Tree Ring to rule Yamal,
One Tree Ring to find them,
One Tree Ring to bring Yamal,
and in the darkness bind them.
Ed_B says:
May 25, 2012 at 7:18 am
Where are the lawsuits? If British Industry that paid for carbon offsets were to band together and sue the government and Briffa et al, for billions in fees paid out on fraudulent science.. would that not bring the issue to a head? The process of discovery, getting at the very basis of the goverment decisions(science used), should be enlightening.
*
I agree with Ed B. Where are the lawsuits? If the MSM won’t drag this ugliness into the open, lawsuits will. British industry, American industry, Australian industry, European industry – I don’t care who starts it. We need to see big industry step up to the plate.
Anthony:
You say of Lazy Teenager:
A purchaser is entitled to get his money back if the purchase is ‘not fit for purpose’. If you are right then John Cook would be foolish if he did not ask to get his money back.
Richard
Tilo Reber sayvers:
May 25, 2012 at 9:34 am
Everyone read this.
If he is right then it should be shouted from the rooftops.
And he is right.
I moved to my present property 14 years ago. The previous householder had planted several walnut tree saplings. Three of them are close to the perimeter of the property. They are about 15 feet apart. The one on the left is now about 20 feet tall. The one in the middle is about 35 feet tall, and the one on the right is about 10 feet tall. Just an observation.
PS I never get any walnuts because the ******* squirrels get them first.