cc: Jerry Meehl <meehl@ucar.edu>, Caspar Ammann <ammann@ucar.edu>
date: Fri, 16 May 2003 14:18:36 -0600
from: Tom Wigley <wigley@ucar.edu>
subject: Soon et al. paper
to: "Michael E. Mann" <mann@virginia.edu>, Phil Jones <p.jones@uea.ac.uk>, Keith Briffa <k.briffa@uea.ac.uk>

<x-flowed>
Dear folks,

I have just read the Soon et al. paper in E&E. Here are some comments, 
and a request.

Mike said in an email that he thought the paper contained possibly
'legally actionable' ad hominem attacks on him and others. I do not
agree that there are ad hominem attacks. There are numerous criticisms, 
usually justified (although not all the justifications are valid). I did 
not notice any intemperate language.

While many of the criticisms are invalid, and some are irrelevant, there 
are a number that seem to me to be quite valid. Probably, most of these 
can be rebutted, and perhaps some of these are already covered in the 
literature. In my view, however, there a small number of points that are 
valid criticisms.

[Off the record, the most telling criticisms apply to Tom Crowley's work 
-- which I do not hold in very high regard.]

The real issue that the press (to a limited extent) and the politicians 
(to a greater extent) have taken up is the conclusions of the paper's 
original research.

First, Soon et al. come down clearly in favor of the existence of a MWE 
and a LIA. I think many of us would agree that there was a global-scale 
cool period that can be identified with a LIA. The MWE is more 
equivocal. There are real problems in identifying both of these 'events' 
with certainty due to (1) data coverage, (2) uncertainty in transfer 
functions, and (3) the noise of internally generated variability on the 
  century time scale. [My paper on the latter point is continually 
ignored by the paleo community, but it is still valid.]

So, we would probably say: there was a LIA; but the case for *or 
against* a MWE is not proven. There is no strong diagreement with Soon 
et al. here.

The main disagreements are with the methods used by Soon et al. to draw 
their LIA/MWE conclusion, and their conclusion re the 
anomalousness/uniqueness of the 20th century (a conclusion that is based 
on the same methods).

So what is their method? I need to read the paper again carefully to 
check on this, but it seems that they say the MWE [LIA] was warm [cold] 
if at a particular site there is a 50+ year period that was warm, wet, 
dry [cold, dry, wet] somewhere in the interval 800-1300 [1300-1900], 
where warm/cold, wet, dry are defined relative to the 20th century.

The problems with this are .....
(1) Natural internally generated variability alone virtually guarantees 
that these criteria will be met at every site.
(2) As Nev Nicholls pointed out, almost any period would be identified 
as a MWE or LIA by these criteria -- and, as a corollary, their MWE 
period could equally well have been identified as a LIA (or vice versa)
(3) If the identified warm blips in their MWE were are different times 
for different locations (as they are) then there would be no global-mean 
signal.
(4) The reason for including precip 'data' at all (let alone both wet 
and dry periods in both the MWE and LIA) is never stated -- and cannot 
be justified. [I suspect that if they found a wet period in the MWE, for 
example, they would search for a dry period in the LIA -- allowing both 
in both the MWE and LIA seems too stupid to be true.]
(5) For the uniqueness of the 20th century, item (1) also applies.

So, their methods are silly. They seem also to have ignored the fact 
that what we are searching is a signal in global-mean temperature.

The issue now is what to do about this. I do not think it is enough to 
bury criticisms of this work in other papers. The people who have 
noticed the Soon et al paper, or have had it pointed out to them, will 
never see or become aware of such rebuttals/responses. Furthermore, I do 
not think that a direct response will give the work credibility. It is 
already 'credible' since it is in the peer reviewed literature (and E&E, 
by the way, is peer reviewed). A response that says this paper is a load 
of crap for the following reasons is *not* going to give the original 
work credibility -- just the opposite.

How then does one comprehensively and concisely demolish this work? 
There are two issues here. The first is the point by point response to 
their criticisms of the literature. To do this would be tedious, but 
straightforward. There will be at least some residual criticisms that 
must be accepted as valid, and this must be admitted. Cross-referencing 
to other review papers would be legitimate here.

The second is to demolish the method. I have done this qualitatively 
(following Nev mainly) above, but this is not enough. What is needed is 
a counter example that uses the method of reductio ad absurdem. This 
would be clear and would be appropriate since it avoids us having to 
point out in words that their methods are absurd. I have some ideas how 
to do this, but I will let you think about it more before going further.

You will see from this email that I am urging you to produce a response. 
I am happy to join you in this, and perhaps a few others could add their 
weight too. I am copying this to Jerry since he has to give some 
congressional testimony next week and questions about the Soon et al 
work are definitely going to be raised. I am also copying this to 
Caspar, since the last millenium runs that he is doing with paleo-CSM 
are relevant.

Best wishes,
Tom.






</x-flowed>
