2011-03-13 23:38:32Hockey Stick arguments...help needed
Klaus Flemløse


Hockey stick discussion

1) I have been reading  the ”Surface temperature reconstructions for the last 2000 years” from National Research Council.


On page 90 a discussion of  ”Spurious Principal Components” takes place referring to a simulation experiment  done by McIntyre and McKitrick using  the R-program. So far so good.

I have not been able to show that this example is not relevant and that there is something in the Hockey Stick procedures which comes around this problem.

I am more or less familiar with the theory behind principal component regression techniques (I.T. Jolifer (2002)) and I have looked through Realclimate, but have not been able to find arguments that disproves this example.   

I would be pleased if someone could guide me to a relevant reference.   

2) I have read somewhere that there is a program written in R-code which can replicate the Hockey Stick from the data used by Mann et al.

I would be pleased if someone could give me at link.


The Hockey Stick debate is coming up again and again in the Danish debate on www.Klimadebat.dk. So far I have not been good enough to refute the skeptical arguments against the temperature reconstructions. Therefore I need help.


2011-03-14 02:04:43
Daniel Bailey
Daniel Bailey

D Kelly O'Day has many such R-Scripts here:



Clear Climate Code has more here:



Tamino did a guest post on Real Climate on PCA vs non-PCA here:


2011-03-14 04:01:15comment
Robert Way


Take a look at Wahl and Ammann (2007).

MM05 in their criticism of the hockey stick said that the recentering process applied in mbh98 was inappropriate and showed that when they retained the same number of PCs then with the correct recentering process that the results were significantly different and showed a pronounced 1400s peak. The problem is that the rule for selecting the number of PCs to be retained is dependent upon the recentering process so in actuality MM05 should have retained more PCs than the number retained by MBH98 and when this is done there is very little difference in the reconstruction.

There are of course issues with Mann's proxy selection but when MM05's criticisms are implemented on the data in the *correct* fashion, you get very similar results (Wahl and Amman, 2007). I'm not a Mann fan, I think some of his work can come close to crossing lines he shouldn't, but I think that Mcintyre is not just in search of answers... the PCA example I gave above is pretty good evidence for that.

2011-03-15 06:43:01Thank's
Klaus Flemløse


Thanks you links/remarks:

My conclusions from reading some of the links are:

1) MBH made a breakthrough in respect of reconstruction of global/local temperatures

2) MBH’s climate field reconstructions are robust

3) MM has at point in respect of centering.

4) Several authors has pointed out that the amplitude of MBH’s reconstruction is on the low side

Much more can be written about MBH/Hockey Stick and it will take  me some time to do...


Klaus Flemløse

2011-03-19 16:14:33only slightly off topic
Peter Miesler

This seems like such ancient stuff but they do cling to it. 

I'm currently getting my noise rubbed in the discrepency between the WMO1999 graph, and Briffa's actual graph. 

It does seem to have a taller tail (business end of that damned ol' hockey stick) than what the Briffa1999 SQR averaged graph seems to justify.  {Seems to me either this proves the global conspiracy or ;-)  there is a simpler explanation I'm unaware of.}

~ ~ ~ ~ ~ ~ ~



2011-12-01 04:20:37
Klaus Flemløse


I will be pleased, if someone will give at link to a R-program, which generates the Hockey Stick.

I have found at couple of versions, however the links does not work any longer.  I whan to dig into the details of the methods and data.