Thursday, June 7, 2012

4947.txt

date: Thu May 1 16:11:50 2003
from: Tim Osborn <t.osbornatXYZxyz.ac.uk>
subject: Re: Writing up the "Long simulations"
to: Simon Tett <simon.tettatXYZxyzoffice.com>, Keith Briffa <k.briffaatXYZxyz.ac.uk>

Hi Simon,
At 10:18 17/04/2003, Simon Tett wrote:

Dear Keith & Tim,
I've been thinking how I want to write up the long simulations. My current thought
on the title is
"Testing simulated climate change of the last 500 years".

Nice and snappy, but perhaps too ambiguous? Also see point below about single vs. multiple
papers...

What I would like to do is compare the simulations with:
1) Tree-ring timeseries.
2) Bore-hole data.
4) Instrumental & early instrumental data.

Was there a 3, or just 1, 2 and 4?
When we get into it there is a lot that can/should be done on each comparison. For
example, for tree-rings there is comparisons along the lines of Collins et al. (st dev,
spectra, EOF), but then the whole question about decline and shortwave signal - which would
necessitate backing it up with some in depth consideration of known tree-shortwave
relationships or at least some additional analysis to increase confidence. Then the
borehole stuff should, at some point, consider soil-air temperature differences and the
influence of land cover and snow cover on these differences, plus the comparison itself.
Then instrumental & early instrumental data (and some documentary records) should also
include seasonal-differences (winter/summer are quite different in the
observations/documentary) as a test of model performance under forcings. Then, of course,
there is the question of what we go just for HadCM3 and what we do just for ECHO-G and what
we do for both models.
But, are you suggesting that we do a first paper, just for HadCM3, that takes a quick look
at each of the 3 comparisons that you list above, that is sufficient to make it an
eye-catching (==well-cited) paper, but that leaves some more in-depth and comprehensive
work for later papers. In which case we could fit it all in to one paper, though we may
risk coming to incorrect conclusions if we haven't done the in-depth analysis yet.

I have just been reading Mann's latest JGR paper on boreholes -- I am unclear what he is
doing..... but I think the assumption that underlys it is flawed. To my my wind he
assumes that bias in the 20th century in the borehole data is a coherent pattern over
the previous 400 years. What do you think?

I read a draft version of this, but haven't yet got round to reading the final published
version. I assume they're similar, though he had great trouble in getting it accepted so
perhaps they're not. Anyway, from the draft version, I agree that he is assuming a 20th
century bias has the same pattern as bias in the earlier part. Perhaps this can be
explained/supported by arguments about the bias being related to an incorrect determination
of the background (i.e., steady state, if climate had been constant) heat flux out of the
ground, though I'm not fully convinced. What I found more worrying however was that I
didn't really feel they could claim to have found the true, unbiased 20th century pattern.
The statistics just didn't seem strong enough and one then wonders whether the authors were
only happy with the result because it matched Mann et al. Anyway, I really ought to read
the published version before saying anything more.
Cheers
Tim

No comments:

Post a Comment