From: Tim Osborn To: Stefan Rahmstorf , Keith Briffa Subject: Re: latest draft of 2000-year section text Date: Tue, 28 Feb 2006 17:10:45 +0000 Cc: jto@u.arizona.edu,eystein.jansen@geo.uib.no, Fortunat Joos ,drind@giss.nasa.gov Hi Stefan, our (Keith and mine) understanding of this issue is that Burger et al. (2006, Tellus, already published and therefore citable) already point out the von Storch et al. (2004) mistake in implementing the Mann et al. (1998) method. But we haven't stated this (or cited the Science in press comment) because Burger et al. also demonstrate that when they implement the method without the detrending step (i.e., following the Mann et al. approach more accurately than von Storch et al. did) then the bias is still there, though of smaller magnitude than von Storch et al. (2004) suggested. Given that we already say that the extent of any bias is uncertain, it does not seem necessary to go into the details any further by discussing the implementation by von Storch et al. of the Mann et al. method. Finally, I think (though here it is less clear from their paper and I am relying on my recollection of talking to Gerd Burger) that Burger et al. also show that the amount of noise von Storch et al. added to create the pseudo-proxies yields a pseudo-reconstruction that has much better verification skill than obtained by Mann et al. (1998) for their real reconstruction. If they increase the noise added (deteriorating the "skill" of the pseudo-proxies) until they get similar verification statistics as Mann et al. report, then the size of the bias gets bigger. In fact, the bias they obtain with the higher noise but "correct" no-detrending method is actually very similar to the bias von Storch et al. reported with lower noise but incorrect detrending method! So where does that leave us? I don't think there's room to put all this in. Of course the magnitude of the bias cannot be determined from any pseudo-proxy simulation anyway, and will be different for different models. We'd be interested to know if your (or others on the cc list) interpretation of Burger et al. (2006) is significantly different to this. Cheers Tim At 16:42 28/02/2006, Stefan Rahmstorf wrote: >Hi Keith and others, > >attached is the draft Keith sent on 21 Feb of the 2000-year section, >with comments and edits (grey) from me. > >I note that Von Storch et al. 2004 is cited without it being >mentioned that they did not implement the Mann et al. method >correctly - by detrending before calibration, the performance of the >method was greatly degraded in their model. I guess you left this >out because the comment to Science showing this is still in press? >Will it be added once this has been published? I think it is a major >point, as it was such a high-profile paper - Von Storch's contention >that the "hockey stick" is "nonsense" (cited in the US Senate) is >based on a mistake. > >Cheers, Stefan > >-- >To reach me directly please use: rahmstorf@ozean-klima.de >(My former addresses @pik-potsdam.de are read by my assistant Brigitta.) > >Stefan Rahmstorf >www.ozean-klima.de >www.realclimate.org > > > Dr Timothy J Osborn 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