cc: gmcbean@eng.uwo.ca,walsh@atmos.uiuc.edu, grabhorn@earthlink.net,corell@dmv.com
date: Tue Jul 13 14:46:52 2004
from: Keith Briffa <k.briffa@uea.ac.uk>
subject: RE: Overpeck calibrated data
to: mmaccrac@comcast.net, Tim Osborn <t.osborn@uea.ac.uk>

   Dear Mike et al.
   Sorry for my tardy response, but it is important that you define the purpose of the
   Overpeck Figure as regards deciding on the "best" version of calibration.
   The question of what is the "correct" amplitude of variation in the reconstruction must
   take into account the rationale underlying the choice of method of calibration (in this
   case the way the data are scaled with respect to some modern mean). In our short piece for
   Science (perspective on Esper et al reconstruction), that addressed the uncertainty in
   regression of paleorecords on climate series, Tim and I showed the effect of scaling based
   on either annual mean or summer mean temperatures (see attached powerpoint slides) .
   Differences will also result from calibrating either inter-annual or decadally-smoothed
   data . If yearly data exist for both proxy and instrumental data , the resulting amplitude
   of longer time scale variation expressed in a reconstruction produced by regression
   involving smoothed instrumental and proxy data, will not match that of the appropriately
   smoothed reconstruction based on regression with annually resolved data. Similarly,
   differences will arise when the same predictor series is scaled against different
   area-average data (i.e. NH mean, NH north of 20 degrees North; Arctic temperatures). Hence
   a direct comparison of the Overpeck ; earlier Crowley; and original Bradley and Jones
   reconstructions (none of which is expressed with inter-annual resolution) with those of
   Mann et al.; Briffa et al , Jones et al etc.are confounded by the ambiguous definition of
   the appropriate regression model (in fact a couple of reconstructions do not use regression
   but only scale the proxy series by adjusting the overall mean and variance to match those
   of the instrumental series  -without considering any least squares fitting) , and
   definition of the seasonal and regional focus.
   These aspects need considerably more attention as regards simple statements as to which
   reconstructions agree or disagree with each other , or with variously-defined model
   outputs.
   The bottom line , is that it might be best to stick with published results for your
   purposes (ie the curve in Tim's piece (which he will send separately) - even though these
   relate to a larger region than the Arctic . If this is not what you want we would still be
   willing to discuss a more appropriate recalibration , taking into consideration the best
   area, season etc.
     At 13:45 05/07/2004, mmaccrac@comcast.net wrote:

     Dear Tim--Thanks very muchfor the offer. The idea of calibrating with the data for north
     of 60 N sounds very interesting, and I am sure everyone would like to see. I guess the
     issue with it would be that we would need to in this document carefully explain what was
     done as it is not published in that way. Or we could basically say that it was done
     similarly to your article, but simply for stations north of 60 N.
     Or if we want to stick with published info could use the best of the two you mention. It
     would be interesting to know how different the two records would be.
     Perhaps John Walsh and Gordon McBean (and others) can weigh in.
     Best regards, Mike
     > Dear Mike,
     >
     > Keith asked me to write a quick reply to your email about a calibrated
     > version of the Overpeck et al. Arctic temperature history.  Both Keith and
     > I are away for the rest of the week, but we can reply more fully when we're
     > back.
     >
     > The series, as published by Overpeck et al. was simply left in "normalised"
     > units -- i.e., standard deviations.  We calibrated it by simple linear
     > regression against the observed temperature averaged over all land north of
     > 20N -- trying both annual means and an extended summer (April-September)
     > temperature.  I can tell you next week which performed best, though I think
     > it was summer.  You are welcome to use either of these series -- I can send
     > them next week.
     >
     > But, for this particular record, a calibration against temperature from
     > north of 60N might be more appropriate for your purposes.  We have the
     > observed temperature series for this region, and we're happy to perform the
     > simple calculation of calibrating the Overpeck series against this if you'd
     > prefer.  Just let us know.
     >
     > Best regards
     >
     > Tim
     >
     > 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:      [1]http://www.cru.uea.ac.uk/~timo/
     > sunclock: [2]http://www.cru.uea.ac.uk/~timo/sunclock.htm
     >

   --
   Professor Keith Briffa,
   Climatic Research Unit
   University of East Anglia
   Norwich, NR4 7TJ, U.K.

   Phone: +44-1603-593909
   Fax: +44-1603-507784
   [3]http://www.cru.uea.ac.uk/cru/people/briffa/