Tuesday, May 22, 2012


date: Fri Jan 7 16:33:48 2005
from: Phil Jones <p.jonesatXYZxyz.ac.uk>
subject: Re: EA WG
to: "C G Kilsby" <C.G.KilsbyatXYZxyzcastle.ac.uk>

Tried ringing, so here are some thoughts about the Work Plan.
Before I forget chop the referees off the end of Craig's CV.
Task 1. Emphasize the essential need for the WG to be able to be
easily altered for a future climate. Could say here that this will be difficult
for WGs that use alternating sequences of wet and dry spells (this is what
Lars WG does for precip). Changes to these sequences are likely to occur
in the future but change from the model is most uncertain - at least cf
mean and SD changes. Need to word this wrt the way GNSRP works
with RCM output changes - like mean, sd, skewness.
Say that BETWIXT work has shown that changing precip is not enough
to change other variables. Need to have change fields for these other
variables as well - partic temp and vapour pressure. Change fields for the
mean for HadRM3H/P are very small for sunshine and wind so we will
likely omit these.
This whole aspect of being able to be modified for the future is the
key. A WG may be very good - if it can't be altered simply it is
useless for the future.
Our WG works on half months - get the annual cycle better.
Task 2 A lot of this can be got from Jones and Salmon (1995). This has
Strengths and Weaknesses. What are the EA requirements?
Task 3 This is the meet of the work. All you need to say is that our
WG will be fit to the daily gridded data for 1958-2002. We will need
all the intercorrelations between variables.
We then apply these to you generated precip series for each box.
I'm assuming that your GNSRP will maintain the basic stats and
the dry/wet spell lengths.
State upfront that our implicit assumption is that the relationships (as correlations)
between precip and the other variables will not change in the
future. We were going to investigate this in BETWIXT but never got around
to it. Could say we can test this ! The way to do this is to fit our WG
to the RCM data for a few boxes for the present climate and separately
for the future - then compare the correlations. We did this for SE England
in another project for HadCM3 and there was little difference. Result
should show that the RCM doesn't change these inter-relationships in the
future cf the present, therefore there is no need to consider this.
It doesn't matter if the model is wrong cf reality. If the model doesn't
change there is no way of altering reality.
Task 4
A lot of Figures and/or Tables of means, sds, skewness. We could
also do this for some extremes a la BETWIXT. For BEWTIXT this got out
of control as you had to have 100 simulations for each site to get
reliable stats for anything other than the mean. So need to be careful
we don't promise too much here. Remember there are 1000's of grid boxes.
Maybe a simple graphical output showing a few things. I'd prefer not
to go into extremes unless there is a robust one we can show for
rainfall - 5 day max total in a year ? Something that isn't too extreme
so we don't need to many simulations.
Task 5 is simple
Task 6 Should be simple
My home number is still 01953 605643. If we're out answerphone will be on.
Rushed through this a little, but this will give you something to go on. I have to get
through the whole of Ch 3 of IPCC report and submit by early next week, so will
spend all weekend on this - breaking just to hear that Rochdale beat a Charlton second XI
that Curbishley will put out !
Once you get this give me a ring or I'll try you in about an hour or so.
Attached CVs for me and Craig Wallace (the person we'll put forward here). Clare
is in Cuba - she gets back this weekend. I'll send her an email for a CV and to discuss
this briefly on Monday.
Our Admin person isn't here so I've made up some numbers. Doesn't seem to get
anywhere near 52K total so we need to discuss this.
I put most of Craig's time down for Task 3/4
I don't have anything about LARS WG. Maybe a sentence or two could go in
to say something like - ours is much easier to perturb?
50-50 OK I guess we will need to beef up the days accordingly but I reckon the 6
for Clare and me is fair - it's realistic. So add to Craig's.
One submission from you with a sub contract to UEA. I think this will make it easier
here as I don't need to get it signed off - a hope anyway.

At 11:36 07/01/2005, you wrote:

Find attached:
- rough draft of proposal
- rough/blank work sheet of costs etc
Urgent points to deal with:
#proposal: need input everywhere, especially
- need to flesh out methodology - have indicated where
- integration with other EA stuff - weak on this (DST etc)
- skills
- CVs (who do you have in mind)
- work plan (not started this yet - ideas welcome/needed!)
- am keen to put in comparison with LARS - do you have anything here to make us look
# costs - need your day rates and a stab at people, no of days in each task etc.
- guess we need to aim for �52-�55K or so. Will use the spreadsheet to "optimise" this,
but need some ideas from you first.
-I guess we split the costs 50:50? -
I see us doing:
- overall direction: rainfall model: (most of) software implementation, GIS/graphical
stuff ?
You doing:
- review of WGs, climate scenarios, WG met variables, PET
Sound OK?
# contracts etc: how do we do this - one submission, UNEW lead, but we have choice of:
- consortium or
- we contract, you sub-contract?
Haven't spoken with our finance people about this,the head honcho is away until Monday.
Will need to wrap up Tues am, courier Tues pm. Need finances sorting Monday though to
get signatures here for Tues.

Prof. Phil Jones
Climatic Research Unit Telephone +44 (0) 1603 592090
School of Environmental Sciences Fax +44 (0) 1603 507784
University of East Anglia
Norwich Email p.jonesatXYZxyz.ac.uk

No comments:

Post a Comment