"As to the Mann data,
it has not "always been available". Some information was available in 1998. Proxies and other
info for MBH99 were available in 1999. In April 2003,the proxy data was not in the public section
of Mann's FTP site (which itself started only in July 2002). When I inquired in April 2003, Rutherford
said that there was no public location of the data and after a few weeks provided a URL in the /sdr/ directory,
pointing to a file which had been archived in August 2002. There turned out to be problems with this dataset
and after publication of our first article in Oct 2003, Mann placed the directory now labelled MBH98
in the public area of his FTP site in Nov 2003 and deleted the dataset in the /sdr/ directory previously
referenced. The data as used in this new directory was inconsistent with the original SI in some respects
and Nature required a Corrigendum which was published in July 2004. Unusually, Nature also required a complete
new SI.
One of the reasons was that the number of
PC series used could not be decoded from the Nov 2003 directory and was provided for the first time in the
Corrigendum SI in July 2004 (which also provided the underlying temperature dataset no longer available
from CRU). You remember the 159 series which Mann said were used in Nov 2003 - this could not be
deduced from the Nov 2003 directories. Of course, this figure was also incorrect. It looks like the July
2004 dataset relates to the one originally used, although the source code posted up in connection with the
Barton inquiry calls for datafiles that do not exist in any of the presently archived datafiles. While it
is possible to approximately replicate MBH98 results (as both we and Wahl and Ammann have done -
and our current emulations are almost identical), neither of us has been able to replicate MBH98 results
exactly. This is aside from the inaccurate description of methodology in MBH98 - most notably the inaccurate
description of principal components methodology which we have discussed elsewhere. But there are other oddities
which one would never have guessed - such as arbitrary extensions of some series and truncations ofother
series, and the use of obsolete and/or grey versions of other series. Residual series used for calculation
of verification statistics remain unavailable to this day. So don't say that this has "always"
been available.
As to other multiproxy studies: Crowley
has "misplaced" his original data and was only able to locate a smoothed and transformed version
of his data. He also could not recall where he got the digital versions of certain series, so this is problematic.
Briffa
has never identified even the sites used in Briffa et al
[2001]. Some of the data used in Esper et al. [2002]
may be archived, but much isn't and it is impossible to verify
versions. Most of Moberg's data is located in public
archives, but some isn't and he has been unable or unwilling
to provide the missing data. Some of the data used in Mann
and Jones [2003] is not archived. Phil Jones says that
he doesn't know the weights used in this study.
Posted by Steve McIntyre at November 20,
2005 07:15 PM" in Comments: IPCC Hockey Stick Matters
(Seite nicht mehr aufrufbar)
|