cc: t.osborn@uea.ac.uk
date: Fri Sep  5 15:34:10 2003
from: Keith Briffa <k.briffa@uea.ac.uk>
subject: Re: An idea to pass by you
to: Edward Cook <drdendro@ldeo.columbia.edu>

   At 08:32 AM 9/3/03 -0400, Edward Cook wrote:

     Hi Keith,
     After the meeting in Norway, where I presented the Esper stuff as described in the
     extended abstract I sent you, and hearing Bradley's follow-up talk on how everybody but
     him has fucked up in reconstructing past NH temperatures over the past 1000 years

   so what more precisely was Bradley saying - we can discuss on phone

      (this is a bit of an overstatement on my part I must admit, but his air of papal
     infallibility is really quite nauseating at times), I have come up with an idea that I
     want you to be involved in.  Consider the tentative title:
     "Northern Hemisphere Temperatures Over The Past Millennium: Where Are The Greatest
     Uncertainties?"

   prefer something like "where is the consensus" - doesn't imply an academic analysis of
   statistical (space and time) confidence levels

     Authors:  Cook, Briffa, Esper, Osborn, D'Arrigo, Bradley(?), Jones (??), Mann
     (infinite?) - I am afraid the Mike and Phil are too personally invested in things now
     (i.e. the 2003 GRL paper that is probably the worst paper Phil has ever been involved in
     - Bradley hates it as well

   interesting to know why - I too share this feeling though , again we can discuss on phone ,
   but don't consider an

     ), but I am willing to offer to include them if they can contribute without just
     defending their past work - this is the key to having anyone involved. Be honest. Lay it
     all out on the table and don't start by assuming that ANY reconstruction is better than
     any other.
     Here are my ideas for the paper in a nutshell (please bear with me):
     1) Describe the past work (Mann, Briffa, Jones, Crowley, Esper, yada, yada, yada) and
     their data over-laps.

   fine - plus a detailed breakdown of how they are produced ( simple unweighted/weighted
   averages) , and , most important what went into each
   at different times - ie explicit how common data input increases back in time

     2) Use the Briffa&Osborn "Blowing Hot And Cold" annually-resolved recons (plus Crowley?)

   would only be latest Crowley (if annually resolved as I think it is)

     (boreholes not included)

   completely omit reference to Boreholes or it complicates all

     for comparison because they are all scaled identically to the same NH extra-tropics
     temperatures and the Mann version only includes that part of the NH (we could include
     Mann's full NH recon as well, but he would probably go ballistic, and also the new
     Mann&Jones mess?)
     3) Characterize the similarities between series using unrotated (maybe rotated as well)
     EOF analysis (correlation for pure similarity, covariance for differences in amplitude
     as well

   coincidence - just suggested this to Pavla Fenwick as suggestion for exploring similarity
   of chronology and core series in NZ (irrelevant comment other than it is unusual to see a
   simple case study where this is done)
   Also Tim has done some of this but I do not think it negates its inclusion here

     ) and filtering on the reconstructions - unfiltered, 20yr high-pass, 100-20 bandpass,
     100 lowpass - to find out where the reconstructions are most similar and different - use
     1st-EOF loadings as a guide, the comparisons of the power spectra could also be done I
     suppose


   yes of course is the crux of issue - but needs exploration of methods (eg using SSA ,
   provided choice of prediction error filter length does not bias results. Fine to use
   band-pass filters if can agree on bands - possibility also of calibrating against similarly
   filtered temperatures (but maybe out of scope, though it needs doing more systematically -
   though perhaps in Tim's paper). The power spectra (or coherency )
   do need to be compared.

     4) Do these EOF analyses on different time periods to see where they differ most, e.g.,
     running 100-year EOF windows on the unfiltered data, running 300-year for 20-lp data
     (something like that anyway), and plot the 1st-EOF loadings as a function of time

   agree

     5) Discuss where the biggest differences lie between reconstructions (this will almost
     certainly occur most in the 100 lowpass data), taking into account data overlaps

   yes - direction of discussion will have to wait on some results though

     6) Point out implications concerning the next IPCC assessment and EBM forcing
     experiments that are basically designed to fit the lower frequencies - if the greatest
     uncertainties are in the >100 year band, then that is where the greatest uncertainties
     will be in the forcing experiments

   yes , this is crucial issue regarding the significance of this agonising over what genuine
   independent confidence can be placed in variations at specific frequencies - seems to me
   that we could include a short mention of the work done so far (last Crowley , recent Hegerl
   paper etc.) and illustrate this and perhaps show is their results are consistent (which
   they are not). We tried to allude to this in the Hot and Cold piece , but they cut it.

     7) Publish, retire, and don't leave a forwarding address
     Without trying to prejudice this work, but also because of what I almost think I know to
     be the case, the results of this study will show that we can probably say a fair bit
     about <100 year extra-tropical NH temperature variability (at least as far as we believe
     the proxy estimates), but honestly know fuck-all about what the >100 year variability
     was like with any certainty (i.e. we know with certainty that we know fuck-all).
     Of course, none of what I have proposed has addressed the issue of seasonality of
     response. So what I am suggesting is strictly an empirical comparison of published 1000
     year NH reconstructions because many of the same tree-ring proxies get used in both
     seasonal and annual recons anyway. So all I care about is how the recons differ and
     where they differ most in frequency and time without any direct consideration of their
     TRUE association with observed temperatures.
     I think this is exactly the kind of study that needs to be done before the next IPCC
     assessment. But to give it credibility, it has to have a reasonably broad spectrum of
     authors to avoid looking like a biased attack paper, i.e. like Soon and Balliunas.
     If you don't want to do it, just say so and I will drop the whole idea like a hot
     potato. I honestly don't want to do it without your participation. If you want to be the
     lead on it, I am fine with that too.

   The idea is a good one and consistent with what Tim and I are thinking (Tim

     Cheers,
     Ed
     --
     ==================================
     Dr. Edward R. Cook
     Doherty Senior Scholar and
     Director, Tree-Ring Laboratory
     Lamont-Doherty Earth Observatory
     Palisades, New York 10964  USA
     Email:  drdendro@ldeo.columbia.edu
     Phone:  845-365-8618
     Fax:    845-365-8152
     ==================================

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

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