cc: Tom Crowley <tcrowley@duke.edu>
date: Thu, 6 Jun 2002 14:59:26 -0400
from: Gabi Hegerl <hegerl@duke.edu>
subject: Tom and my reconstruction
to: Phil Jones <p.jones@uea.ac.uk>, t.osborn@uea.ac.uk

   Hi Phil and Tim,

   Here comes the information on Tom and my reconstruction. I can easily redo

   it for seasonal or growing season means, I'd actually be happy to and check

   if that is better.

   I count on your discretion since I have not yet fully published this anywhere,

   but I am working at it!

   proxycompclg1.2.as is the CLH reconstruction as ascii timeseries (1: year, 2:
   reconstruction 3: NH 30-90N mean annual instrumental for comparison).

   reconstructionsnewphil.as is an ascii file that explains how this was done and gives for
   example the weights and stations used (it has a longer list, but those I wound up using are
   marked by **, Mongolia treerings got kicked out despite ok correlations because of the low-

   resolution and Tom had some other worries about them) and some correlations

   recordclh.psc is a ps file showing the timeseries compared to instrumental and Tom's old
   reconstruction.

   recordsused.ps is a plot of the timeseries used.

   I'll also attach two files, cl6.as with the proxy timeseries as used, and cl6.readme with
   the headers (matlab cannot easily digest a mix of ascii and numbers). On this, I count also
   on your discretion since I am not sure exactly

   what Tom has as agreements with the people, we can talk about all this in Thera!

   Hope this helps!

   See you in Thera, Phil! Greetings

   Gabi

      Gabi,

         Can you send the series to me and Tim Osborn (t.osborn@uea.ac.uk)? No rush as we're
      both away for much of the next 3-5 weeks. Also can you send the weights and the
     original
      series as well.  Maybe also send the series you didn't use when correlations were
     small.
      I agree that combining a simple way maybe better than MLR but we need to compare a
      range of methods.
         If there isn't time we can discuss all this on Thera.
      Cheers
      Phil
     At 13:01 16/05/02 -0400, you wrote:

     The revised timeseries I made with Tom (and I promised to send, should I send
     it to you or Tim O?) uses weights determined by the correlation of a timeseries with the
     nh mean (so it is like a multiregression but does not account for
     correlation between timeseries which is what tends to give you the overfitting
     and strong weights of crazy signs, in theory its the second best thing to do
     but in practice I think it is safer than multiregression).
     So it is a positive weight highest for
     timeseries going well with NH mean, and sometimes negative for timeseries that
     tend to be anticorrelated, but I think there is only 1 case.
     If you want me to, we can also make one for NH summer. Tom is keen on annual,
     but I agree with you that summer may be more meaningful.
     I pre-screened the timeseries by knocking out those with very poor correlations
     with local temperature.
     Tom uses that one now, the correlation of it with instrumental is nearly the same as
     with his old one though, but there are some reasons why he likes the new one better. See
     you in Thera!
     Gabi

      Gabi,
          Adding this to 1.a.2 is fine. Remove the 'an' before improved on the 5th line.
      I will be on Thera so we should take the opportunity to discuss these aspects whilst
     there.
      I will be there the whole time though arriving a little late (Monday if I recall
     correctly).
         Doing what I propose in 1.a.2 will need to involve Mike as we need to get all his
     data,
      and a lot more news stuff that has been produced in the last few years that none of us
     have.
      What I want to get at most is the weights of each series going into the reconstruction,
     so
      we know better which series are really doing the work. These are the ones we want to
      encourage possible extensions. At present we don't know which contribute most, although
      what Tom and I do by giving them equal weight may be just as good.
         See you on Thera.
      Cheers
      Phil
     At 10:14 16/05/02 -0400, you wrote:

     Hi Tom and Phil,
     How about this on 1.a.2 , this has "integrate reconstructions" to say we want to
     combine information. Phil, does that cover what you suggest?
     Greetings
     Gabi
     2. Assess climate variability on decadal and centennial timescales, from the
     instrumental and paleo record and from new records, attempt new reconstructions and
     compare and integrate reconstructions for hemispheric scale climate variability with the
     goal of obtaining an improved estimates of natural internal and natural forced climate
     variability (Crowley / Jones).
     --
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
     Gabriele Hegerl - NOTE CHANGE IN ADDRESS FORMAT
     Department of Earth and Ocean Sciences,
     Nicholas School for the Environment,
     Box 90227
     Duke University, Durham NC 27708
     Ph: 919 684 6167, fax 684 5833
     email: hegerl@duke.edu

     Prof. Phil Jones
     Climatic Research Unit        Telephone +44 (0) 1603 592090
     School of Environmental Sciences    Fax +44 (0) 1603 507784
     University of East Anglia
     Norwich                          Email    p.jones@uea.ac.uk
     NR4 7TJ
     UK
     ----------------------------------------------------------------------------

     --

     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
     Gabriele Hegerl - NOTE CHANGE IN ADDRESS FORMAT
     Department of Earth and Ocean Sciences,
     Nicholas School for the Environment,
     Box 90227
     Duke University, Durham NC 27708
     Ph: 919 684 6167, fax 684 5833
     email: hegerl@duke.edu

     Prof. Phil Jones
     Climatic Research Unit        Telephone +44 (0) 1603 592090
     School of Environmental Sciences    Fax +44 (0) 1603 507784
     University of East Anglia
     Norwich                          Email    p.jones@uea.ac.uk
     NR4 7TJ
     UK
     ----------------------------------------------------------------------------


--

   ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

   Gabriele Hegerl - NOTE CHANGE IN ADDRESS FORMAT
   Department of Earth and Ocean Sciences,
   Nicholas School for the Environment,

   Box 90227
   Duke University, Durham NC 27708
   Ph: 919 684 6167, fax 684 5833

   email: hegerl@duke.edu, http://www.env.duke.edu/faculty/bios/hegerl.html

   Attachment Converted: "C:\EUDORA\Attach\recordsused.ps" Attachment Converted:
   "C:\EUDORA\Attach\recordclh.psc" Attachment Converted:
   "C:\EUDORA\Attach\reconstructionsnewphil.as" Attachment Converted:
   "C:\EUDORA\Attach\proxycompclg1.2.as" Attachment Converted: "C:\EUDORA\Attach\cl6.readme"
   Attachment Converted: "C:\EUDORA\Attach\cl6.as"
