cc: Scott Rutherford <srutherford@rwu.edu>, Caspar Ammann <ammann@ucar.edu>, "Wahl, Eugene R" <wahle@alfred.edu>
date: Tue, 10 Oct 2006 12:33:48 -0400
from: "Michael E. Mann" <mann@meteo.psu.edu>
subject: GKSS results
to: Keith Briffa <k.briffa@uea.ac.uk>, Tim Osborn <t.osborn@uea.ac.uk>

<x-flowed>
Dear Tim/Keith,

I hope all is well with both of you.

We've been doing a number of sensitivity tests w/ RegEM using both the 
CSM simulation, and now more recently the GKSS simulation data we got 
from you. There are some methodological developments we'll describe 
soon, related to what is the most reliable regularization method in 
RegEM, ridge regression and truncated total least squares. We are now 
leaning towards the latter because of potential non-convergence problems 
in some cases w/ the former. More on that soon.

More relevant, however, are the results. As you can see from the 
attached plot, RegEM works quite well w/ GKSS, using a short calibration 
period (1900-1980, corresponding to years 900-980 in the attached plot) 
and both white and red pseudoproxy noise (we used rho=0.5 in the 
attached, but similar result for other values).

The most interesting result is that while RegEM reconstructs the full NH 
series well throughout, in the case of the CSM simulation, it does 
modestly underestimate the warmth of the earliest centuries in  the GKSS 
Erik simulation (it fits everything else, including the LIA cooling, 
very well). We feel that this is likely due to problem of correctly 
identifying the 'drift' pattern using CFR methods.

The long and short of this is that we would like to be able to show this 
result in a (very short!) J. Climate response we need to finalize, to a 
comment on Mann et al (2005) J. Clim by Zorita and Von Storch. We would 
show you this response for comment of course, and would add you as 
co-authors. We have cleared with Andrew Weaver that this would be an 
acceptable course of action.  We are hoping you are in agreement with this?

please let us know ASAP, we have to finalize our response within days.

thanks,

mike

-- 
Michael E. Mann
Associate Professor
Director, Earth System Science Center (ESSC)

Department of Meteorology              Phone: (814) 863-4075
503 Walker Building                    FAX:   (814) 865-3663
The Pennsylvania State University      email:  mann@psu.edu
University Park, PA 16802-5013

http://www.met.psu.edu/dept/faculty/mann.htm



</x-flowed>

Attachment Converted: "c:\eudora\attach\whiteredcompare.jpg"
