date: Fri Feb 4 12:14:26 2005
from: Phil Jones <REDACTED>
subject: Re: [Ch3-cas] [Fwd: Zero order draft of Chapter 3, AR4, IPCC]
to: "Thorne, Peter" <REDACTED>

We can discuss this in Chicago. I get in about 2pm on the 22cnd, so
give me a call when you arrive. We'll both be a bit knackered, so maybe an
earlyish meal (about 6pm - normal time for them yanks) and/or a couple
of expensive beers in the bar beforehand.
I have just got the CCSP report - in 4 emails last night. All on my stick.
Seems as though I'll be going through Chapter 5 and maybe 3. Just told
them 3 would be better than 5. Lots to read on flights to India (Feb 12-20)
and Madrid (Feb 6-10).
Nathan and I had a meeting with Kate yesterday. We talked about a date
for one when you came. I vaguely recall a date in March, but I didn't write it down
so check with Kate and Nathan. Just returned from 2 hours with U/Gs
and need a lunch break.
Kevin is the CLA for Section 3.4 and David and Brian Soden the LAs
who put most of it together. See what Kevin says in response. You have
summarised his views on the issue very succinctly in your take home
message from the section ! He is very pro Fu. As far as I know he
wasn't involved in CCSP and I've only yet looked at the Table of Contents
of it.
I'll send a summary of this to all in a few minutes.
At 10:57 04/02/2005, you wrote:

Kevin, Phil et al.,
my substantive comments on the upper-air portion only. Before I give
specific comments below I have some over-arching comments:
This draft and the CCSP report seem at best tangential - is this
desirable or sensible?
There is little effective communication in the main text of the
uncertainty that is inherent in these measures due to the poor quality
of the underlying data and metadata and to the choices made -
"structural uncertainty". It seems that a decision has been made that
RSS and the Fu et al. method are "right" or at least "most right" and
this is what we will put forward as gospel truth almost. Other datasets
are given a cursory once over almost. This completely ignores legitimate
concerns that "structural uncertainty" is large aloft - seemingly
reasonable choices made as to how you homogenise and then analyse the
data can have very large effects. This is not at all clearly
communicated in the current draft.
The essential distilled message that I think the analysis of UA
temperatures has left us since the TAR, and what this chapter should
say, is:
"Independent efforts to create climate records from satellite and
radiosonde records since the TAR have served to illuminate previously
unrecognised uncertainties in temperature evolution aloft (Seidel et
al., 2004, Thorne et al., 2005). Further, choices in post-processing
(e.g. Fu et al., 2004) may help to clarify satellite retrievals, but
legitimate concerns remain (Thorne and Tett, 2004, Spencer et al., 2005)
and other equally plausible approaches should be actively considered.
Our increased understanding of trend uncertainty aloft means that we can
no longer dismiss warming aloft of similar or greater magnitude than at
the surface over the satellite record. Nor can we discount a relative
cooling aloft. Uncertainties are largest in the tropics and Southern
Hemisphere high latitudes where radiosonde coverage is poorest.
Obviously, the climate has only evolved along a single pathway.
Therefore a major challenge to the climate community is to refine our
range of estimates."
This is what CCSP effectively says.
What, rightly or wrongly, I get out of the current draft on an initial
read is:
"We don't like UAH. We don't believe radiosondes over the satellite
period, but do over the longer period (paradox). We believe Fu et al. is
correct. There is no longer any problem whatsoever."
I don't think this simple message is actually remotely supported by the
science. Therefore at the very least efforts are required to balance the
text so that this is not the message communicated. I don't think we
should be scared of admitting that we just don't know, if indeed we just
don't know (which I believe is a fair reflection of the state of the
Specific comments:
p.23 lines 13-14 and 53-57 and p.24 lines 1-6. I disagree strongly with
these as written. I do not believe that Fu et al. weightings is some
panacea nor that the "cancellation" works on all space and timescales
(the statement needs to be *proved* it cannot be accepted as an article
of faith - that is not the way science works). I'd be amazed if it did.
The reservations raised in the peer reviewed literature need to be
better articulated here for the document to be fair and balanced. I
guess this whole area will evolve significantly over the next 12 months
or so though.
General concern: In the TAR we used 20N to 20S to define the tropics -
here (Table 3.4.1.b) you use 30N to 30S. I'd suggest 20-20 is physically
more logical and has backward compatablity and should be used. This is a
recommendation of the Exeter workshop report queued for review in BAMS.
Regardless, you need to alight on a single definition of these regions
here and elsewhere in the report and stick to it. If you look at zonal
mean profiles from any UA dataset then 20-20 shows marked trend changes
N and S of it (greater warming) so using 30-30 gives a chance of a fools
gold scenario arising.
In Table 3.4.1.b TLT is the acronym used in Christy et al. 2003 for T2LT
- this may very well cause confusion. Admittedly I was only scanning the
tables but I thought that this claimed there was a RSS 2LT channel
Page 26 lines 28-37. This is at significant odds with the CCSP report
conclusions as currently written. Much of this relates to the relative
weighting being given to the Fu et al. approach by the different author
teams. It will seem very odd to a policy maker to read two such
disparate threads. I particularly dislike the use on line 30 of "when
the stratospheric influence is properly taken into account (Fu et al.,
REDACTEDa)". How can we say it is properly taken into account that way?
There are a very large population of plausible approaches that could be
taken and to date we have two - a "physical" 2LT and a statistical
T850-300. That is grossly insufficient to make bold statements regarding
one of them properly taking the effect into account. Again, this needs
balance and caveats on the Fu et al. technique until we resolve
unanswered questions. Likewise, T2LT has not been proven to be untenable
in the peer reviewed press - so you cannot make this statement. My
feeling is that we are missing a significant opportunity here to outline
the considerable uncertainty in evolution aloft in favour of deciding
one subset of approaches is right and presenting this as gospel truth. I
am very uncomfortable with this. As I said it is at significant odds
with CCSP.
page 26. para starting on line 46. Seems almost an afterthought. For
HadAT (at least, but as they are so highly correlated, also highly
likely LKS) the long-term trend in the tropics is entirely an artifact
of the regime shift - if you split time periods then pre- and post-1979
have negative trends and the whole period has a strong positive trend.
So to state boldly that trends agree and therefore all is well is again
our living in a fools paradise. It is true, but it just shows that trend
metrics are very dangerous beasties and should be handled with care. The
Seidel and lanzante paper should also be quoted here.
page 68. Bullets on line 15, line 19, and line 31. Again, my concern
here is that these are far too narrow and you are effectively claiming
that one approach is right. Really refers back to my earlier points.
This is painting a light fuzzy grey as black when I don't believe the
science to date supports such an interpretation.
Page 110, line 55. Containing 676 stations (not CDRs).
Page 111, line 31 The Thorne et al. referenced is a paper under review
at BAMS that you don't have in your current reference list. Reference
is: Causes of differences in observed climate trends Peter W. Thorne,
David E. Parker, John R. Christy, Carl A. Mears

Common question 3.2. You'll be unsurprised to hear that I think this
paints too rosy a picture of our understanding the vertical structure of
temperature changes. Observations do not show rising temperatures
throughout the tropical troposphere unless you accept one single study
and approach and discount a wealth of others. This is just downright
dangerous. We need to communicate the uncertainty and be honest.
Phil, hopefully we can find time to discuss these further if necessary
either in Chicago or when I visit in March (has a date been decided
I'll be away from three weeks from today and unable to access this email
account. If we need to iterate further I can be reached (intermittently)
on REDACTED but will be fairly busy and then on
holiday in the middle week.
Peter Thorne Climate Research Scientist
Hadley Centre for climate prediction and research
Met Office, FitzRoy Road, Exeter, EX1 3PB

Prof. Phil Jones
Climatic Research Unit Telephone +44 (0)REDACTED
School of Environmental Sciences Fax +44 (0)REDACTED
University of East Anglia