|
From:
Steve McIntyre
Sent: January 22, 2004 6:09 PM To: Ziemelis, Karl Cc: Ross McKitrick Subject: MBH98 Dear Karl We
hope that your Christmas vacation was pleasant. We have not heard from
you recently and would appreciate an update on the status of the
inquiry into our materials complaint concerning MBH98. As you perhaps
know, we have submitted an article to Nature dealing primarily with
the MBH98 principal component algorithm. If you have not seen it, we
would be pleased to forward a copy to you. AGU’s policies regarding the archiving and citation of data (applying to Geophysical Research Letters among other publications) are clearly stated at http://www.agu.org/pubs/data_policy.html. While, to our knowledge, Nature does not appear to have explicitly adopted a comparable policy, we presume that Nature would not intentionally apply a lower standard. As articulated, the AGU policy is an excellent one and would deal with many of the data disclosure issues which we encountered in connection with Mann et al (1998) (although its application by AGU editors appears to be sporadic at best.) The AGU policy requires that data sets cited in an article must come from a public archive, discouraging use of “gray” archives, even if operated by academic institutions. Their policy also requires identification of the data center, together with series identification code or number. Some of the series used in MBH98 (the ITRDB tree ring series for NOAMER, SOAMER and AUSTRAL) include ID codes, which are sufficient to identify the sites (although even this detail is inadequate to deal with problems arising from updated series at WDCP with the same site code.) More seriously, many of the series used in MBH98 come from “gray” sources and are data that have been passed around within a small community without ever being adequately archived. Out of the 81 non-principal component series, we have only been able to identify 27 series, for which the MBH98 version matches the WDCP version either exactly or through only an annualizing transformation. The remaining series are incorrectly attributed, have never been publicly archived or have some other issue preventing an exact match. Proper citation and attribution will require considerable tidying, since the entire situation is rather a mess and should include the following:
This would be a vastly more stable approach than merely coopering up an unsatisfactory and poorly documented gray archive. On a different topic, we have tested the following statement made by Mann et al. in connection with MBH98 (http://www.ngdc.noaa.gov/paleo/ei/ei_nodendro.html ): “In selecting data from the International Tree-Ring Data Bank (ITRDB), MBH98 set criteria designed to minimize these problems. Starting with the full data bank available in 1997, they identified 1589 site chronologies, each representing a unique combination of species, variable (e.g. ring width or maximum latewood density) and location. Only the 251 chronologies that met the following criteria were retained: · Reliable information on the methods used to remove biological trend was available; · The median length of the individual segments used to build the chronology was greater than 150 years; · The mean correlation of these individual segments with the site chronology was greater than 0.5; · The first year of the chronology was before AD 1626, and it contained at least 8 segments by 1680; · The last year was after 1970, and there were still 8 segments after 1960.
We calculated mean correlations between individual segments at WDCP and the site chronology and inspected the start year and number of 1680 segments. We were unable to reconcile the figure of “251 sites” as we identified over 381 tree ring sites used in MBH98 in the ITRDB, of which 320 sites were identifiable in ITRDB records at WDCP. At least 39 sites had not started by AD1626; at least 22 sites had fewer than 8 trees available in 1680 and at least 171 sites had mean correlation of the ring width series to the site chronology below 0.5. One site (proxy #60 – Churchill Composite) had a mean correlation of only 0.045 between the individual segments and the site chronology. We wrote the originating author, Rosanne d’Arrigo, pointing this out and also pointing out that the site chronology extended to 1982, while the latest ring width measurement was in 1978. She advised us that she would request that this site chronology be withdrawn from WDCP. This obviously indicates that the quality control stated by Mann et al. to have been applied to tree ring sites in the preparation of MBH98 did not take place in the form stated.
Again, we appreciate your co-operation in this matter.
Yours truly,
Stephen McIntyre and Ross McKitrick
|