Example #2 Of Work That Needs To Be Replicated: Dendroclimatology

For anyone who has paid attention, the dendroclimatology field has been rife with bad practices for years – cherry picking data sets, hiding modern data that shows “the wrong answer,” using bizarre statistical approaches, flipping data sets upside down, and utter resistance to data requests and any attempts at replication.   Most of the really damning CRU emails are about various dendroclimatology studies, and Keith Briffa, lead author of this section of the last IPCC report, is right in the middle of it all.

The Bishop Hill blog has the story of Briffa’s Yamal data set that has many of these elements.  I have been following this story for years at Steve McIntyre’s blog, but this is a very readable narrative.  It will really put a lot of what in in the CRU emails in context.  I highly recommend it.  Seriously.  In fact, I would go read some of John Holdren’s testimony in front of Congress on Climategate first, then read the Bishop Hill piece to get a sense for the whitewash.

Postscript: I have to laugh — when you see insiders in the alarmist community discussing the resistance to data sharing that really has no excuse, the excuse they use nonetheless among themselves to salve their conscience is the meme that “the FOIA’s were meant to be just harassing them and aimed at reducing the time they had to do real work” — ie they were (as the meme goes with skeptics) based on anti-science rather than any real desire to do science.

Here are a couple of bits from the Bishop Hill piece.

Meanwhile, however, McIntyre could begin to look at what Briffa had done elsewhere. It was not to be plain sailing. For a start, Briffa had archived data in an obsolete data format, last used in the era of punch-cards. This was inconvenient, and apparently deliberately so, but it was not an insurmountable problem — with a little work, McIntyre was able to move ahead with his analysis. Briffa had also thrown a rather larger spanner in the works though: while he had archived the tree ring measurements, he had not supplied any metadata to go with it — in other words there was no information about where the measurements had come from. All there was was a tree number and the measurements that went with it. However, McIntyre was well used to this kind of behaviour from climatologists and he had some techniques at hand for filling in some of the gaps….

Eventually, though, Briffa’s hand was forced, and in late September 2009, a reader pointed out to McIntyre that the remaining data was now available. It had been quietly posted to Briffa’s webpage, without announcement or the courtesy of an email to Mcintyre. It was nearly ten years since the initial publication of Yamal and three years since McIntyre had requested the measurement data from Briffa. Now at last some of the questions could be answered.

12 thoughts on “Example #2 Of Work That Needs To Be Replicated: Dendroclimatology”

  1. I’m glad to see you read Bishop Hill. His “Cosa nostra” post was interesting. You have Jones and Mann emailing back and forth about “our committee” and using some (committee?) resource to discredit RR McKitrick & PJ Michaels and prevent publication of their work and the need to keep it(the committee) secret.

    It’s a sad day when scientist for a committee to fight science.

  2. Not really germane to this post, but I am wondering more and more if the current State of the GW/CC science is akin to Lysenkoism in the early Soviet period.

    If your not familiar with his work, Trofim Lysenko was an agronomist in the Ukraine in the post-WWI period. He rose to prominence for two reasons: first, he claimed yields far in excess of other researchers, under conditions more extreme than they could (growing peas in frozen and otherwise snow-covered fields) and his theory of plant genetics fit the Marxist model of dialectical materialism almost perfectly. In short, his theory was that acquired characters could be passed down to the progeny of plants grown under unusual conditions, as both the thesis of the plant’s genetic stock and the antithesis of the environment would be combined in the synthesis of the plant’s offspring.

    Because of his popularity with the leaders of the early Soviet government, he was soon placed in charge of most, if not all of Soviet agricultural research, and his theories were made the curriculum of Soviet Agricultural universities. When early on it became apparent that his results could not be duplicated by other researchers, those who disagreed with his conclusions were politically denounced, stripped of their academic positions, and often arrested or exiled.

    Thus two full generations of Soviet scientists were created under a system that judged them not on how their work increased outputs, but how well their work reflected the theoretical basis of Lysenkoism. When the research did not result in improvements outside of the agricultural research stations, the fault was placed at the feet of the farmers, through their malfeasance or outright sabotage. Many were arrested and sent to the gulags on fatuous charges of conspiring against the state.

    After enough of these failures, however, the few scientists who had access to western scientific literature realized that not only were Lysenko’s theories bunk, but that his early work was most likely fraudulent. Lysenko was discredited along with many of the other Stalin-era officials, and the Soviet Union quickly and quietly purged him and his ideas from official policy.

  3. I think you’ll find that it was “Bishop Hill” who alerted Steve McIntyre to the fact that the Procedures of the Royal Society had a strict requirement for revelation of raw data (and when pressed, they upheld it).

  4. The big flaw in the “it would take too much time to respond to FOIA requests”, is this. If they posted all the data, metadata and software code on a public website at the time that they published in ‘peer reviewed’ journals, then all the information would be public. There would be no need for any major FOIA requests except possibly some small ones for missing details. Indeed the journals, in theory, require that the data be available. But they do not enforce it for these people.

    Look in Analytical Chemistry for an example of the way real scientists do this. All the data and the experimental setup are published in each paper. Other chemists can duplicate the work. No secrets and no FOIA required.


  5. All the proxies have the limitation that there’s only about 120 years or so of measured temperatures to compare with. Perhaps some of them rely though on simpler science than the physiology of trees and the ecology of stands of trees?

  6. Very bottom line of Climategate: If the skeptics previously had any doubts that their critiques were perhaps wrong or off base in some way, these doubts have been removed. Now they know they were right all along. The fraud has been clearly unearthed. The skeptics are now confident in their skepticism while those on the warming side who have even a scintilla of integrity, are now feeling a bit of a skeptical edge in their previously strongly held beliefs.

  7. Dendroclimatology…now there’s a mouthful! Assuming that temperature alone accounts for growth rate changes is more than a stretch…it’s pure fantasy. Attempting to ‘account’ for the multitude of other factors affecting growth rates must amount to, at best a well intended S.W.A.G, at worst self-deception, as these are literally Legion, and mostly completely un-trackable. These kinds of studies would require dozens, if not hundreds, of identical provenance samples to average out individual variance among the specimens…and even then there are still many other factors affecting growth rates that can’t really be accounted for…. This has to be considered at best a very speculative proxy for temperature, and worst only noise for any practical purpose. I don’t think it’s much of a stretch to say that any climate reconstruction based on tree-growth must be considered simply as a demonstration of a ‘Possible’ climate regime, but with absolutely no evidence of its probable accuracy at all.

  8. ADiff,
    I don’t blame the dendros for at least attempting to tie multiple proxies (from different tree species, at different locations and altitudes). Since I began following ClimateAudit 5 years ago, and reading what one must do in order to come up with some magical global climate signal, it has become apparent that thier attempt is nearly impossible. I certainly agree with you that it is almost statistical alchemy that people like Biffra practiced.

  9. I think the problems of dendrocloimatology run deeper than those mentioned here. The “divergence/hide-the-decline” data suggests to me that the correlation between tree ring late wood density and temperature is overstated at least in some species.

    If the post 1960 trees rings indicate declining temperatures but we have direct temperature measurements indicating a slight increase (viewed in the best light), then the reliability of the historical dendro temperature proxies seems to be in question.

    I suspect a truer reconstruction is provided by ice cores and lake sediments.

  10. Larry Sheldon: “Is it worth cutting down more really old treas for?”

    You don’t cut them down. You drill a hole with a hollow drill and get a 1/4″ thick sample of everything down to the center of the tree. It doesn’t hurt the tree noticeably.

    Forty-some years ago, I did a school science project on tree rings, with input from a Biology professor (my father). Graphing the samples vs. weather records basically gave a scatter plot with no apparent pattern at all. Conclusion: at least for certain species growing in town, any effect of weather was swamped by other influences on growth.

    Now, if I’d had computers and a half-baked knowledge of statistics, maybe I’d have run correlations until I found some trees that happened to show a correlation to temperature. And if I was a complete idiot, I might even have accepted that for some trees the rings were wider in warmer years, and in other trees they were narrower.

    And that is just part of what Mann and his colleagues did. They used computerized correlation functions to search through all the available records from extremely old trees (mainly bristlecone pines) until they found some that showed correlation to the ground temperature records – including some where the correlation was inverted. Then they used the correlation coefficient, positive or negative, to calculate temperature from the ring widths. Naturally, when they averaged these calculated temperatures, the average since 1850 graphed just like the temperature records. They’d picked the few trees that matched well…

    And then they extended the temperature calculations backwards, assuming the correlation coefficients remained constant. And the result was pretty much a straight line. Which is also what you get when you average a lot of random data.

    IOW, apply Mann’s methods to random data, and you will cherry pick data that just happens to fit the (massaged) temperatures for the last 150 years. That creates the blade of the hockey stick. Then averaging the data from 1000-1850 AD creates a straight handle, because it is random data, and now you’re out of the region where you cherry picked.

    And Mann is quite content that his awful mess of a program has been tested and demonstrated correct, because it produced the output he expected. (Idiot…)

    This doesn’t prove the hockey stick is wrong, just that it is meaningless. Historical records go pretty far to proving it is wrong, at least where we can find records of what crops grew where…

Comments are closed.