date: Fri, 16 Apr 1999 16:38:11 +0100
from: "Jean Palutikof" <j.palutikofatXYZxyz.ac.uk>
subject: FW: TSUNAMI - Extreme Weather in Northern Europe
to: "Tim Osborn" <t.osbornatXYZxyz.ac.uk>, "Trevor Davies" <t.d.daviesatXYZxyz.ac.uk>
I thought you might both be interested to see this.
James' final comments are interesting - more analyses and more variables for
Trevor - although not involved, you'll be interested to know that next
Friday's meeting has been cancelled. UCL have failed to come up with
anything - they're now promising for next Tuesday. James was all for
meeting anyway although I tried to head him off, but fortunately Goodman has
said he wants to restrict the whole thing to just one meeting at which we
can table a new proposal and close the deal.
I'll circulate UCL's efforts when they arrive.
From: James Orr [mailto:email@example.com]
Sent: 15 April 1999 18:04
Subject: Fwd: TSUNAMI - Extreme Weather in Northern Europe
Futher to our discussion this afternoon, I attach my original "reviewers'
comments" note to Richard Chandler.
I have also spoken to Richard and he agrees that the meeting should be
based on the original proposals, rather than the "under development"
proposals. He will also give a presentation on Friday.
P.S. Richard mentioned that he shared your concern about the amount of
work and number of meetings that the "merger" process was creating. This
is just to confirm that I recognise this and am grateful for your support
Received: from bsfiles.nerc-bas.ac.uk
by pcmail.nerc-bas.ac.uk; Thu, 15 Apr 1999 17:57:13 +0100
Received: from egon.stats.ucl.ac.uk by bsfiles.nerc-bas.ac.uk
(8.8.8/NERC-1.5(Solaris 2.x, SMTP))
id RAA20497; Thu, 15 Apr 1999 17:57:13 +0100 (BST)
Received: (qmail 2462 invoked from network); 15 Apr 1999 16:57:10 -0000
Received: from orpheus.stats.ucl.ac.uk (184.108.40.206)
by egon.stats.ucl.ac.uk with SMTP; 15 Apr 1999 16:57:10 -0000
X-Mailer: Windows Eudora Light Version 3.0.1 (32)
Date: Thu, 15 Apr 1999 18:01:51 +0100
From: "Caroline Connell" <cconatXYZxyzail.nerc-bas.ac.uk> (by way of "Richard E.
Subject: TSUNAMI - Extreme Weather in Northern Europe
Content-Type: text/plain; charset="us-ascii"
I have now read the three sets of reviews that gave us feedback on the UCL
and UEA proposals. The selection of remarks is mine (hence it is
subjective) and I have tried to arrive at the essence of their comments.
Both proposals were viewed as well-written, by competent and experienced
research teams, but were seen as sitting comfortably within the bounds of
their existing fields of research.
1. Rejection of physical model approach is disappointing.
2. Extreme Value Methods should be included, given the subject matter of
the Announcement of Opportunity (AO).
3. Would benefit from applying time-series analysis, to deal with
non-linearity (reference to techniques developed by Harvey) and the dynamic
(as in time-varying) nature of weather.
4. Project plan is vague on content and gives only a brief outline of the
5. The approach shows promise to give new insights, but needs to be more
6. Did not show a great appreciation of risk. The rejection of Extreme
Value Methods was seen as evidence of this.
1. Competent, but not innovative. Needs to question existing paradigms
(sorry their word, not mine!) about use of physical models.
2. NCEP a sensible choice of data and shows evidence of appreciating its
3. The choice of insurance-sensitive variables is appropriate.
4. ENSO and NAO are sensible choices for input variables, but perhaps
limited in their number.
5. Reliance on GCM model data implies rejection of stochastic approach,
which is disappointing.
6. Would benefit greatly from a time-series approach that incorporated
allowance for time-variance, non-linearity and multi-variate analysis.
My overall impression of the reviewers' comments is that they see benefits
in the UEA proposal being opened up to include consideration of stochastic
modelling, hence UCL's involvement.
Also, the range of statistical methods used should be broadened, to
include time-series (multi-variate, time-varying and non-linear) and
extreme value methods.
Finally, the number of explanatory variables that is studied needs to be
I hope that this is of use to you. I would like to copy this note to Jean
Palutikof, but will wait for your permission before doing so.