date: Thu, 3 Feb 2005 16:40:36 -0500
from: Henry Pollack <hpollackatXYZxyzch.edu>
subject: Re: cry for help!
to: Keith Briffa <k.briffaatXYZxyz.ac.uk>
I am now somewhat dug out of the deluge that awaited me upon my return from
travels in mid-January.
I never heard back from you about my comments sent on January 13 and copied
below. However, I don't want to let the discussion grow cold. I know that my
own contribution needs shaping and refining, but before I do more I need to
know your reaction to my critique. To enable things to continue on a slow
simmer, I would be grateful to see the latest 'zero-order' draft, along with a
time-table for advancing to a 'first-order' draft. I will have more discussion
(already I have a few more topics that I did not include in my rushed critique
of January 13) to take up with you. Perhaps when Tim Osborne visits later this
month he and I can discuss some of these.
What do you make of the resignation of Chris Landsea from the IPCC tropical
storms working group?
Thanks for the latest version, which in the borehole discussion differs notably
from the earlier version you sent, presumably due to comments from others.
Because I am traveling again (leave Friday, back on Tuesday), I will send only a
few general comments today, in no particular order.
1. Figure 1b, as discussed on pp.1,2 . This figure shows borehole results, and
'pseudo' borehole results, but the borehole curves are never mentioned
explicitly in the text. The pseudo borehole results are the so-called 'optimal'
results of Mann. Mike has never analyzed borehole temperatures; he simply
manipulated reconstructions that I sent him, in a manner that dealt with
spatial noise in a totally inadequate way. I think it is inappropriate to show
the so-called 'optimal' reconstruction because it is flawed (even after the
correction of gridding errors); the gridded curve should be the one shown in
Pollack & Smerdon 2004, not the Briffa and Osborn 2002. Tim has told me of how
that was constructed, and the P&S 04 version should be substituted.
2. Your discussion of uncertainty on pages 4,5 is good. The paragraph following
acknowledges that there is generally greater centennial-scale variability in the
ensemble of reconstructions than is displayed in the hockey stick. I think the
hockey stick is effectively an end-member reconstruction, rather than a
'centrist' whose uncertainty bounds encompass all other reconstructions.
3. The statistical efforts (yours, others) to retain more long period
variability in the dendro series have led to greater variability in the
reconstructions, all toward a cooler past in the 16th-17th centuries. There is
no a priori reason that the improved dendro series should have led to a cooler
past; retention of more long period variability might conceivably have led to a
warmer past, but it did not. That makes me think that the 16th-17th centuries
were indeed cooler than the hockey stick portrays.
4. page 7. The Von Storch, Zorita model results suggest that the regression
reconstruction techniques do not extract signals well in the presence of noise.
There is a new paper soon to be submitted by Tom Crowley, Gabi Hegerl et al that
reaches a similar conclusion. I believe that this is an important issue, one
that is at the heart of my criticism of how Mike Mann re-analyzed the borehole
results. May I suggest that in the FOD (first order draft) that you include a
discussion of field reconstruction methodologies, particularly as to how they
fare in the presence of noise. This would compare and contrast the regression
techniques, the von Storch approach, the Crowley approach, the revised
Rutherford Mann (2005)methodology, and perhaps the wavelet approach that Anders
Moberg has proposed.
I do not think that the problem that you cite in the last paragraph on p.7 is
strictly with the particular simulation of Von Storch. The same general
conclusions are reached in the new Crowley, Hegerl paper, which use different
5. It is not clear what you are saying on p.8, 3rd paragraph when you say that
the GST histories imply a very much higher level of 'recent' warmth when
compared to the general picture of proxy reconstructions. If 'recent' means the
period of the instrumental record, the GST reconstructions compare very well,
and if anything are a little muted. Moreover, both the von Storch and the new
Crowley results show a very good overlap in the 17th century of borehole
reconstructions with proxy reconstructions that use non-regression
6.GCM simulations are new players in the reconstruction arena. The work of Von
Storch, Simon Tett, Caspar Ammann are examples. I think a discussion of how
these are being integrated into the historical debate would be useful. You make
the point that there are few new proxy data since the TAR. But what is new are
the methods of reconstructing the climate field , and the role of
the GCMs in testing certain methodologies with pseudo-proxies. The field has
come quite a way from what is in the TAR, but your summary seems to convey the
feeling that little has changed.
7. I am not offering any comments on the rest of the material (hydrological,
etc.), as I have not had time to digest it yet.
Thanks for the opportunity to comment! The comments are offered in the most
constructive of spirit. You have undertaken a task of which I have not the
slightest of envy. I am confident that the final product will be fair and
Please let me know the time frame of when this document will undergo further
shaping and refinement. I am grateful to have had the opportunity to
___ ___ Henry N. Pollack
[ \ / ] Professor of Geophysics
| \/ | Department of Geological Sciences
|MICHIGAN| University of Michigan
[___]\/[___] Ann Arbor, Michigan 48109-1063, U.S.A.
Phone: 734-763-0084 FAX: 734-763-4690