Gotland by Esper

One of the main Hockey Team studies is Esper et al 2002, which is published in Science and, naturally, no data is archived. Esper has not deigned to reply to any emails by me requesting data. Esper and coauthors have just published an article in QSR leading with a discussion of the Mann controversy. I’m told that Esper could not even bring himself to mention my name in connection with this and therefore cited Regalado [2005] (the Wall Street Journal article) – no doubt surprising the reporter. Actually it’s rather ironic that ES&T trashed the Wall Street Journal for this article – can you think of any other newspaper article in recent memory which has been cited in a serious academic journal and by Esper no less.

Esper does not do archiving; so it’s hard to get much traction on Esper et al [2002]. However, I recently read an article at his website discussing RCS on the Gotland site (mentioned in Esper et al 2002). So I’ve got a bit of a foothold.

Esper has posted up Esper J, Cook ER, Krusic PJ, Peters K, Schweingruber FH (2003) Tests of the RCS method for preserving low-frequency variability in long tree-ring chronologies. Tree-Ring Research 59, 81-98. This contains a detailed discussion of the RCS method on the Gotland site. Esper himself doesn’t do archiving of data, but, in this case, the site can be shown to be swed022, which was archived at WDCP.

The graph below shows (top): ring width and bottom – maximum density. A mere civilian looking at the top panel would not be able to deduce that 20th century ring widths are exceptionally wide. But the Hockey Team has ways of extracting unique 20th Century warmth even from a tough-looking data set like this. We’ll see how they do it (though not today)

Figure 1. From Esper et al 2003.

For now, I’ve contented myself with proving that swed022w and swed022x were the same data sets as Esper used (since he wouldn’t do anything as banal as giving a data citation.) The figure below shows that the data sets are the same.

Figure 2. Replication of Esper figure using swed022w and swed022x.

Just for fun, I’ve included a couple of diagrams that I’ve used from time to time (you’ll remember the one for Tasmania). Here’s a grass plot for Gotland.

Here’s the same data by age, rather than year.

Now that I’ve got a bit of a foothold, I’ll look at how Esper handles the Gotland site. There’s much of interest in the article.

Update- Sept. 13: I’ve deleted a sarcastic honorific.


  1. TCO
    Posted Sep 11, 2005 at 11:21 AM | Permalink

    Love the grass plots.

    How the heck do these guys get so much attention in the big journals like Science and Nature. Just goes to show how full of puff stuff is in science nowadays. I actually trust the specialty journals more.

    But what did these guys do to get so annointed that all their papers go off in Science?

  2. John A
    Posted Sep 11, 2005 at 2:26 PM | Permalink

    Since archiving is optional, can I just make up the underlying data and publish my results in a peer reviewed journal?

    Does Esper get his research money from the NSF as well?

  3. TCO
    Posted Sep 12, 2005 at 5:41 AM | Permalink

    Yes, you can. But you’ll have much better plausible deniability if there is some data (that you skewed) and if you use the process to cover skewing. that way, when it comes out, it is just a misunderstanding. Not fraud.

  4. per
    Posted Sep 12, 2005 at 7:08 AM | Permalink

    what is QSR, and can you please provide a citation ?

  5. TCO
    Posted Sep 12, 2005 at 7:11 AM | Permalink

    Quaternary Science Reviews (by googel investigation)

  6. Paul Gosling
    Posted Sep 12, 2005 at 7:35 AM | Permalink

    As a ‘civilian’ it looks like something odd has happened since about 1900. Where has all the variability gone, especially in ring width? This seems to suggest the trees are under extreame stress and not growing much at all.

  7. Dave Eaton
    Posted Sep 12, 2005 at 7:36 AM | Permalink

    Fascinating stuff. I need to see what archiving requirements there are in physical science journals I’m familiar with. Our data tended to come from samples, which we would provide, even to people we didn’t like much, and the data not subject to too much manipulation. I just assumed these were the rules, but it might be the folkways of the scientific area I worked in when my stuff could still appear in the open literature.

    TCO- not to pry, but when I see TCO, I read ‘transparent conducting oxide’. Is this the origin of the initials? Again, feel free to ignore this inquiry- I do, however, respect your commentary here, and I just gather from the way you say things that our backgrounds might be somewhat parallel.


  8. Steve McIntyre
    Posted Sep 12, 2005 at 7:49 AM | Permalink

    Paul, the modern trees are much older than the average sample throughout the history – so there is a fundamental non-homogeneity in nearly EVERY tree ring chronology – called by Melvin “modern sample bias”. Most species have narower ring widths as they get older. So the dendro guys adjust for aging. Then, all of a sudden, the 20th century instead of having narrow ring widths, ends up showing “unprecedented” warmth. Age adjustment becomes an important issues.

    There are a lot of effects going on. One that interests me (and this is a deleted preamble to another post where the preamble was inapplicable): germination seems to require even warmer temperatures than staying alive. So if you have changing climate in a temperature-stressed site, the germination will be more heavily biased to warm periods. Thus, on average, as trees get older, they will be experiencing a colder climate. Thus, some portion of the age-adjustment curve will be a result of biased getting colder: presumably the slope of the actual age-adjustment curve would not be as steep as the estimate resulting from biased germination.

  9. TCO
    Posted Sep 12, 2005 at 7:56 AM | Permalink

    Dave, you’re a shmarty…first beer is on me. 😉

    I agree with Paul about the 1900s, looking like something is different.

  10. Paul Gosling
    Posted Sep 12, 2005 at 8:55 AM | Permalink


    Couple of questions.

    1) Why are modern trees older? I seem to remember a post on this but cannot find it.


  11. Paul Gosling
    Posted Sep 12, 2005 at 9:01 AM | Permalink


    A question and a point.

    1) Why are modern trees older? I seem to remember a post on this but cannot find it.

    2) I doubt that the effect you are referring to is a germination effect. Woodland floors tend to have a much less variable temperature than above the canopy and germination is pretty robust. It is the first few years after germination (in trees) which are the most precarious. Young trees are more vulnerable to frost damage and to browsing by large herbivores. If conditions are unfavourable they will be in this vulnerable stage longer. There may be an issue with seed availability. If the adult trees are stressed they will probably not be producing much seed.

  12. Steve McIntyre
    Posted Sep 12, 2005 at 12:04 PM | Permalink

    Paul, the moden trees tend to be sampled with a minimum of 5 cm trunk at 1 m height. These tend to be either younger fast growing trees or older slow growing trees. So if you look at the “grass plot” for Polar Urals, you’ll see that there are no trees with cumulative ring width

  13. Max
    Posted Sep 13, 2005 at 8:05 AM | Permalink

    I have got a question. You are speaking of archiving, as if it were something totally unheard of in climate science. You have to archive all your data and results, to make a reasonable study and to show evidence for your findings.
    If you don’t do it, how would other scientists know that it was rightfully done in a scientific way, or just patched together well-wishing.

    Or did I understand something wrong?

  14. TCO
    Posted Sep 13, 2005 at 8:13 AM | Permalink

    In the physical sciences, data is often not archived. That’s why Wilson urges actually PUBLISHING data. It becomes part of the record that way and allows people to see if previous work was flawed and even to correct after the fact to still get some value out of flawed work. Of course, this becomes a logistical issue. With modern computers and such, however, we may be moving more towards a world of data being “published”. Of SI-like journals even. It is an interesting issue…

  15. Steve McIntyre
    Posted Sep 13, 2005 at 8:23 AM | Permalink

    Re #13: Max, you understand wrong. In climate science, you don’t have to archive data and results. There is a great deal of data that is archived, but it is patchy and compliance in the major multiproxy studies is abysmal. The authors have genrally refused to provide the data upon specific request as well.

    As to your neatly expressed query about “patched together well-wishing”, that summarizes my entire line of questioning.

  16. Paul Penrose
    Posted Sep 13, 2005 at 5:15 PM | Permalink

    The problem seems to be that the researchers are offended that anybody would even question their methods or data. We should just trust them – they are the experts after all, and incapable of making mistakes, let alone fabricating data.

%d bloggers like this: