There’s an amusing little incident with the deleted “original” data set that was posted up for a few minutes at Mann’s website – you know, the data set that was first demonstrably referenced by a CA reader in the early morning of Sep 5. (I’ll reserve comment for now on issues relating to the timestamp of this data set and the Gavin Schmidt hyperlink to it, presently pointing to a data version that did not exist at the time that the hyperlink was supposedly created.)
Within a day, on the afternoon of Sep 5, the data set was deleted and replaced with another data set, again without notice, in a bewildering concatenation of replacements that is reminiscent of our experience with the Hansen’s GISS data almost a year to the day ago. However, both myself and others took the precaution of downloading the Sep 4 version as soon as we saw it – just in case it disappeared. Not an imprudent precaution, given its almost immediate deletion.
I’ve now had an opportunity to forage through the deleted version. The deleted data had 1357 series, from which 148 series were deleted to yield the 1209 series that now appear in the “original” data. But surely the 1357 series is “more” original than the 1209 series? What criteria were used to winnow out the 148 removed series? Inquiring minds want to know. There’s not a whisper on this topic in the paper or in the SI and, of course, all traces of the 148 series were ruthlessly scrubbed from Mann’s website.
Surprisingly, on the list of deleted series was a series entitled “Yamal 2002″. Now Yamal is very familiar to CA readers, as Briffa replaced the Polar Urals update (with a high MWP) with his Yamal version (with a HS) and this one substitution affects a number of reconstructions (discussed on many occasions:
- The Euro Hockey Team and Yamal
- Briffa 2000 and Yamal
- Polar Urals: Briffa versus Esper
- Polar Urals and Yamal Mean Ring Widths
- Polar Urals Spaghetti Graph
The graphic below illustrates the difference between the Polar Urals update (black) and Yamal (red) (See above links for the impact of the Yamal substitution on the Briffa 2000 reconstruction, a spaghetti graph)
Figure 1. Comparison of Polar Urals update to the Briffa Yamal version.
So why was this series excluded from Mann et al 2008? And if it was no good for Mann et al 2008, why is it any good in the “other” studies?
Just for fun, I plotted up the Mann 2008 version of Yamal (shown in the graphic below):
Quite obviously this data set doesn’t have a HS and doesn’t look anything like the Briffa version of Yamal. Where did it come from? This data version is the Yamal version from the original authors (available at WDCP here ). This version matches the graphic in Hantemirov 2002. (In passing, I’ve observed previously that Juckes et al and other Team articles have cited Hantemirov, when they actually used a Briffa version that is not located at WDCP.)
Where did the Briffa Yamal version come from and why is it different than Hantemirov’s? Presumably they use different standardization methods, but then one would have to see the measurement data, which Briffa thus far has steadfastly refused to disclose (for 8 years and multiple articles.)
I still don’t know why the Hantemirov version wasn’t used in Mann et al 2008. Would it “matter”? Yes and no. As a single series, it wouldn’t necessarily “matter”, but it would be nice to know why these series were excluded. Curiously, Mann used Briffa’s Tornetrask version (not Grudd’s), so it’s not like he refused to use grey Briffa versions. It’s odd that the Hantemirov version got into the mix in the first place.
Now there’s a nice little ending to this story. The “wrong” data set – the one deleted from Mann’s website – is the one that Mann sent to WDCP, an archive of record, where Mann can’t change “inconvenient” data sets without leaving a trace. So WDCP has one “original” data set and Mann’s website has another.
Will the real Slim Shady please stand up? (Re-Mix).