cc: Bette Otto-Bliesner <ottobli@ucar.edu>, Keith Briffa <k.briffa@uea.ac.uk>,Jonathan Overpeck <jto@u.arizona.edu>
date: Tue, 04 Jul 2006 10:21:33 +0100
from: Tim Osborn <t.osborn@uea.ac.uk>
subject: Re: hockey stick
to: Stefan Rahmstorf <rahmstorf@ozean-klima.de>

<x-flowed>
At 18:46 03/07/2006, you wrote:
>To the hockey game experts - you might have seen my exchange with 
>von Storch in Science this week, and be interested in my comments to 
>their Response available here: http://www.pik-potsdam.de/~stefan/vscomment.html
>
>Cheers, Stefan

Thanks for the email, Stefan.  You make some interesting points and 
it's another useful opportunity to highlight the error in the von 
Storch et al. implementation of the MBH method.  I think, however, 
that in places you are confusing (or at least not properly 
distinguishing) the issues of bias and of random error.  They really 
need to be considered independently.  The main thrust of your letter 
appears to be that the size of the bias indicated by von Storch et 
al. for the HadCM3 simulation is no larger than the random 
uncertainty error published by MBH.  But this bias (if it exists at 
all) is present *as well as* the random uncertainty error.  So, 
regardless of whether any bias is larger or smaller than the 
published random error, if the bias exists it will affect both the 
reconstruction central value and the reconstruction uncertainty range.

Please don't take this email to mean that I agree with all of von 
Storch et al.'s reply, and disagree with all of your points, because 
I don't.  I just think that the issue of whether different error 
components are smaller or larger than each other is a red herring... 
ideally we should not ignore any error components.  In practise we 
can ignore any components that are shown to be always 
negligible.  The calibration bias has been shown to be negligible in 
some situations and for some methods; we need to do more work to find 
those methods for which the bias can be shown to be negligible in all 
reasonable situations, or if this is not easily possible, then we 
would have to accept that bias might exist and then quantify its 
likely magnitude.

Best wishes

Tim


Dr Timothy J Osborn, Academic Fellow
Climatic Research Unit
School of Environmental Sciences, University of East Anglia
Norwich  NR4 7TJ, UK

e-mail:   t.osborn@uea.ac.uk
phone:    +44 1603 592089
fax:      +44 1603 507784
web:      http://www.cru.uea.ac.uk/~timo/
sunclock: http://www.cru.uea.ac.uk/~timo/sunclock.htm

**Norwich -- City for Science:
**Hosting the BA Festival 2-9 September 2006

</x-flowed>
