cc: "Birgitte von Christierson" <b.v.christierson@hrwallingford.co.uk>
date: Tue Jul 28 12:05:07 2009
from: Phil Jones <p.jones@uea.ac.uk>
subject: Re: Sampling UKCP09?!
to: "Steven Wade" <s.wade@hrwallingford.co.uk>

    Steven,
       We'll discuss this with Geoff next week.
    A few things
    In your 3) the way MOHC produced the probabilistic projections GCMs and RCMs
    (with their perturbed physics) were given greater weight if they simulated the
    20th century better. This is a part validation, so better models and parameter sets
    get preferred.
    In 4) not clear what 'more homogeneity than the 11 RCMs' means?
    Your key bullet lower down is the third one. You do need to sample the full range -
   somehow.
    This is difficult to do in the joint probability case.
     In 5) totally agree that you can get very different results depending on how you sample
    UKCP09. This should be clear from reading the reports, but it is difficult to get across
    when people use UKCP09 as a tool - without complete understanding.
    The one thing you don't mention relates back to the work we did for the EA
    project in 2005/6. This is that it is still very likely that historic droughts will
    be more extreme than future ones. This is because the GCMs and RCMs and
    the WG don't do low-frequency variability very well. The WG will have reasonable
    extremes, but is unlikely they will bunch together into drier and wetter decades
    and bi-decades. This is discussed in the large report under blocking, and also
    in the WG report. The WG may not do 2-3 year droughts that well - but neither
    will the RCMs.
    Cheers
    Phil
   At 10:19 27/07/2009, Steven Wade wrote:

     Phil,

     I wondered whether you might have a view on how to make UKCP09 a little more practical
     for water companies?

     I attach a note on our attempt to capture the range of UKCP09 by using Optimal Latin
     Hypercube Sampling that makes sure we can sets of factors that cover a reasonable range
     of UKCP09 and maintain characteristics of the Met Office emulator outputs (combined
     probabilities in P and T for each season and between seasons)

     We are suggesting that using 10-20 samples is a good first step and use of say 100
     Weather Generator runs would come later for sites where there are significant risks.

     You must of tackled some of the same issues trying to make your WG outputs consistent
     with UKCP09 (?) so I would welcome your opinion on this.

     Regards,

     Steven

     Steven Wade - Group Manager
     Water Management
     HR Wallingford Ltd
     Howbery Park, Wallingford, Oxfordshire OX10 8BA, United Kingdom
     e: s.wade@hrwallingford.co.uk
     t: +44 (0) 1491 822214 (direct), +44 (0) 1491 835381 (switchboard)
     f: +44 (0) 1491 825916 (direct), +44 (0) 1491 832233 (general)
     [1]www.hrwallingford.co.uk

     **********************************************************************
     HR Wallingford uses Faxes and Emails for confidential and
     legally privileged business communications. They do not of
     themselves create legal commitments. Disclosure to parties
     other than addressees requires our specific consent. We are
     not liable for unauthorised disclosures nor reliance upon them.
     If you have received this message in error please advise us
     immediately and destroy all copies of it.
     HR Wallingford Limited
     Howbery Park, Wallingford, Oxon, OX10 8BA, UK
     Registered in England No. 02562099
     **********************************************************************

   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.jones@uea.ac.uk
   NR4 7TJ
   UK
   ----------------------------------------------------------------------------

