cc: <K.briffa@uea.ac.uk>
date: Tue, 14 Jun 2005 09:50:50 +0100
from: "Rob Wilson" <rjwilson_dendro@blueyonder.co.uk>
subject: Re: IPCC AR4
to: "Tim Osborn" <t.osborn@uea.ac.uk>

   Hi Tim [and Keith],

   Hopefully I will soon have reached a compromise with Philip Brohan with regards to the
   coral reconstruction paper - he has issues with the use of regression and we have been
   playing around with a pseudo-proxy data-set to test the method.



   I have comments from Keith and colleagues at GKSS, but will not start editing the paper
   until I have received something from you.

   I would like to do this as soon as possible as I move house in 2 weeks and would like to
   have the paper submitted before then.



   another a related issue.

   From both my TR NH recon and the coral recon, I noted that RE values seem to be grossly
   inflated.

   The attached figure shows a pseudo-proxy example - in this situation, calibration was made
   over the 1940-1981 period and verification over the 1897-1939 period.

   Clearly, the predicted values (Lrec) are too high.

   RE = 0.63, while CE = -0.74.



   In this example, a sole reliance on RE would be wrong and could result in a highly bias
   reconstruction. CE is telling the correct story - i.e. the predicted values should not be
   trusted.



   I was wondering what you and Keith have experienced over the years.



   There has been criticism by Macintyre of Mann's sole reliance on RE, and I am now starting
   to believe the accusations.

   Although RE is stating that it is still better to use the predictive values than the mean
   of the calibration period, that is essentially meaningless - surely?



   sorry for the pressure - but friendly pressure I hope

   regards

   Rob





   PS. Will get comments to you about the coral work very soon!

   Attachment Converted: "c:\eudora\attach\negative CE.pdf"
