Eli Rabett (Josh Halpern) meent een climatescepticsgate te hebben ontdekt. Hij is er tien jaar na oprichting eindelijk achter dat er een yahoo groep bestaat die sceptisch over het klimaat debatteert. De vraag is waarom komt hij er nu pas achter? zie Lucia Liljegren’s The Blackboard voor een levendige discussie over deze maillijst.
Op 26 januari 2003 om half elf ’s avonds schreef ik dit op de climatesceptics mail list:

Dear Forum,

Mann et al did the following: They compared North american treelines and the Principal component #1 of North American Treerings for the period 1400-1980. This was matched. Then they correlated North american tree rings with Northern hemisphere temperatures 1860-1980.

How well does Northern hemisphere temperature correlate with North american temperature ? Actually it doesn’t. Eg for the 48 USA states the 1930′s were comparable to the 1990′s

The logical conclusion is that Northamerican trees respond better to global average temperatures than to local temperatures.

My Big question Why did MBH1999 not use North american temperatures to calibrate north american tree ring data??

Then the following jump was made: the ITRDB Principal component#1 doesn’t show a medieaval warm period therefore the average northern hemisphere (average globe) must not show a MWP. And the hockestick was born.

Hans E


Dit leidde tot een reactie van Steve McIntyre

Re: MBH1999 – a review (1)

Hans – Just to note that I think the dissection of MBH1999 (also
1998)is very important. I’ve spent a fair bit of time on this and
intend to reply to your posts, but plan to spend some attention on radiative physics for a little while.

But briefly, I find Mann’s underlying statistical methods a bit
bewildering and they deserve some thought. For others, what (I think) Mann does is the following: (1) he collected (a) 112 tree ring, O18 data, etc. and a couple of long instrumental series. He may have reduced this number by taking principal components of the tree ring series, but I can’t tell; (b) Jones gridded 20th century data; (2) calculated 5 principal components of the Jones data; (3) on a series-by-series basis, he regressed the proxy series against the Jones principal components for the period 1902-1980. There is no report on R^2, serial correlation of residuals or other such mundane information which would be useful to auditors (as far as I can tell); (4) for earlier periods, he reverses the process to go from the proxy series to the Jones PCAs and thus back to temperature. Mann’s papers are all conveniently accessible on the Internet through his web page – google mann climate virginia.

Mann observed that he could not obtain any results through canonical analysis, but was able with the above methods. Mann uses phrases like “trainee network” and “skilful”, which I find a bit ingratiating and offputting, but perhaps statistical language has changed since I went to university. I am not able to comment at present on his methodology, but my sense is that there are weaknesses to it, which deserve careful auditing.

One other point which you might note. Somewhere Briffa has observed that tree-ring proxies in the late 20th century do NOT follow the late 20th century warming. Briffa posits some unknown anthropogenic cause to this and as I recall excludes post-1960 data in his transfer function modelling. It’s all quite curious and hasn’t received enough analysis to date.

Regards, Steve McIntyre

De rest is geschiedenis, en de “Briffa trick” is nog steeds actueel

Hier in een recensie van de Hockeystick illusion komt mijn naam ook voorbij:

It began with a question, by Dutch scientist Hans Erren, on a blog. A retired Canadian statistics expert, Steven McIntyre, undertook to examine the statistical underpinnings of the claim: Mann’s report was based not on any straightforward report of field data but on abstruse, mysterious statistical manipulations of secret data.