cc: Eugene R Wahl <wahle@alfred.edu>, Francis Zwiers <francis.zwiers@ec.gc.ca>
date: Wed, 24 Oct 2007 09:31:41 -0600
from: Caspar Ammann <ammann@ucar.edu>
subject: Wahl/Ammann vs Zwiers
to: Phil Jones <p.jones@uea.ac.uk>

   Hi Phil,

   a short note regarding what you perceive as differences in the reconstruction results by
   Lee et al (Zwiers) and ours. Actually, I don't think that there is anything that is
   inconsistent. There are a number of differences in the approach and so a one-to-one
   assessment is not possible. However, all general results agree very well with our
   observations and exercises, so I have absolutely no issues with Francis' results and paper
   (other than small bickering and suggestions for what is a really nice and clean paper):

   - more proxies: better performance (daahhh)

   - smoothing: better performance (! Key !) This is not systematically done in tradiational
   MBH, but I bet that this is actually the largest component for the improvement... you can
   do the acrobatics of xyz, in the end, if you do the fit on smoothed data, what is most
   representative of the CLIMATIC evolution in the series, then you get the best fits and
   amplitude preservation.

   - dependence on models: none

   - dependence on calibration period coverage: YES. Francis does not assess this, but we show
   that if you calibrate over period where a good chunk of the warmest and a good chunk of the
   coldest years are present, then your potential for amplitude loss is reduced

   - detrending: simply doesn't make sense, statistically as well as geophysically (removing
   the pattern that is associated with the global average and then try to recapture that in a
   truncated space??? stupid...)

   Caspar

   --
   Caspar M. Ammann
   National Center for Atmospheric Research
   Climate and Global Dynamics Division - Paleoclimatology
   1850 Table Mesa Drive
   Boulder, CO 80307-3000
   email: [1]ammann@ucar.edu    tel: 303-497-1705     fax: 303-497-1348

