date: Sat Feb  2 19:09:40 2002
from: Mike Hulme <m.hulme@uea.ac.uk>
subject: Fwd: RE: UKCIP confidences
to: j.turnpenny@uea.ac.uk

     Date: Tue, 29 Jan 2002 13:54:02 +0000
     From: "Mitchell, John FB" <john.f.mitchell@metoffice.com>
     Subject: RE: UKCIP confidences
     To: 'Mike Hulme' <m.hulme@uea.ac.uk>
     Cc: "Senior, Cath" <casenior@metoffice.com>,
      "Wood, Richard" <rwood@metoffice.com>
     X-Mailer: Internet Mail Service (5.5.2653.19)
     Hi Mike
     I've put some comments in the text. Having said repeatably for good reason
     that regional climate change is very uncertain, I think we have to be very
     careful not to get carried away in UKCIP.
     With best wishes
     John
     John F B Mitchell, Head of Modelling Climate Change
     Met.Office, Hadley Centre for Climate Prediction and Research
     London Road,  Bracknell, RG12 2SY UK
     Tel +44 (0)1344 856613/6656 Fax+44 (0)1344 856912
     E-mail john.f.mitchell@metoffice.com
     [1]http://www.metoffice.com/research/hadleycentre
     > -----Original Message-----
     > From: Mike Hulme [SMTP:m.hulme@uea.ac.uk]
     > Sent: Monday, January 28, 2002 9:32 AM
     > To:   Mitchell, John FB
     > Cc:   Senior, Cath; Jenkins, Geoff; Jones, Richard; j.turnpenny@uea.ac.uk
     > Subject:      Re: UKCIP confidences
     >
     > John (and others),
     >
     > Thanks for the comments on the confidence levels for the UKCIP02 report.
     > I have been through these and am happy to go with most of your suggested
     > changes.  But I have the following five statements where I would propose a
     > slightly different resolution, plus one additional one re. THC.
     >
     >
     > Temperature
     > 1. "SE>NW gradient":  could we retain the 'H' confidence if we restricted
     > this just to summer?  There is certainly better model agreement for this
     > gradient in summer, and the physical basis should be better
     > continentality, the winter role of the ice-albedo f/b at higher latitudes
     > counter-acting this gradient is absent, the weakening of the THC leads to
     > reduced NW warming
     > [Mitchell, John FB]  My reason for suggesting medium was that ist does
     > depend to a large extent on the land surface and cloud parameterizations-
     > both of which we have limited physical understanding. having said that,
     > summer drying in the southern half of Europe is I think a featrure of most
     > , if not all recent models, and the continentality and THC/deep mixing in
     > the ocean do provide some physical explanantion.
     > 2. "Night>day in winter, Day>night in summer":  I had 'M' and you suggest
     > 'L'.  I would have thought the physical basis for this is reasonable
     > greater cloud in winter but higher humidity, and less cloud in summer
     > allowing greater insolation.  Not sure about other models, but I would
     > have thought 'M' is reasonable. [Mitchell, John FB]   Humidity changes
     > alone I suspect are not enought to explain this, and I don not think that
     > one can put anything more than L on something that depends on cloud- why
     > dose the cloud increase?  Again why odes summer cloud decrease- becasue
     > the surface dries- but the surface dries because of reductions in cloud
     > cover.  Cloud cover:
     > 3. "Reduction in summer cloud, esp. in south":  you had only 'L', I had
     > 'M'.   I would have thought a consistent picture is being told here
     > reduced cloud, reduced precip. (which we give 'M' for), reduced soil
     > moisture (which we give 'M' for), and increased in summer DTR (see above).
     > Hard to argue that with these other changes we would actually get
     > increased summer cloud.
     > [Mitchell, John FB]  See comments above about cloud generally-
     > precipitation is faily consistent across models , but has anyone looked at
     > cloud- also it is possible have precip decrease but more boundary layer
     > stratus because of a more statically stable atnosphere- think of SE
     > England with easterlies.
     >
     >  Wind
     > 4. "Average windspeeds increase more over south than north"  I had 'L',
     > but you would rather omit.  If we exclude this one then we will have no
     > headline statements about wind.  I agree that model consistency will be
     > low, but the physical reasoning I would have thought is due to the
     > circulation changes and storm tracks pushing further south  if Had3AMH
     > model is worth anything then isn't it giving us greater confidence about
     > storm tracks.  I would prefer to retain an 'L' here.
     > [Mitchell, John FB]  Have you looked at the winds over land? They are
     > awful. I have no physical explanantion for th changes, I doubt if ti is
     > statistically significant, and HadCM2 had different changes. I appreciate
     > we have no headline statements about wind- but I don't think we should be
     > putting something as flaky as this in a headline statement.
               Soil moisture:
     > 5. "Decreases in summer, esp. in SE": you had 'M' and I had 'H'.  This is
     > part of the story above, reduced cloud, reduced precip, increased DTR, and
     > therefore reduced soil moisture.  Even though precip. change in summer may
     > only be 'M', surely there is greater physical basis to think that soil
     > moisture levels will decrease with large summer warming?  I would prefer
     > to retain 'H'.
     > [Mitchell, John FB]  As I noted above, we are not sure what triggers the
     > drying. I suspect it is mainly a combination of reduced precipitation and
     > increased evaporation. The reduced preciptitation does seem to be a large
     > scale response in the SRES runs ( ie not merely a local feedback with the
     > land surface) but I would still feel slightly uncomfortable with H.  M or
     > H,  I would definitely  confine the bullet to the SE
     > One statement that seemed to get lost was this one.  It should be accorded
     > a 'H' status.
     >
     > 6. "Although the strength of the Gulf Stream may weaken in future, it is
     > very unlikely that this would lead to a cooling of UK climate over the
     > next century".
     > [Mitchell, John FB]  I agree
     >  Or if you don't like this one, some other equivalent that makes some
     > clear statement about THC and that we are not going to 'freeze' in the
     > foreseeable future!!
     >
     >
     > Thanks  I'd be interested in your views.  Would be happy to discuss these
     > on the phone sometime if you think we need to.
     > [Mitchell, John FB]  I am around most of this week- Linda (01344 856656)
     > usually can find me if I am wandering round the building,- next week I am
     > running a mmeting Mon- Thurs AM
     >
             John
     >
     >
     >
     > At 16:47 23/01/02 +0000, Mitchell, John FB wrote:
     >
     >
     >       Dear Mike
     >
     >       Sorry about the incomplete email - finger trouble. It stopped at a
     >       particularly inappropriate point! I also discovered I had mssed some
     > of the
     >       Table  I assume the Table is a high level summary and hence should
     > contain
     >       that which is useful and in which we have reasonable confidence
     >
     >       I attach comments on the revised table of confidence levels- I have
     > tried to
     >       assess the levels using the three criteria- physical understanding,
     >       consistency amongst models and statistical significance ( in the
     > Hadx3
     >       models). In many cases this information is not available, so I have
     > had to
     >       guess. Richard agrees there is no time to carry out statistical
     > tests on all
     >       the variables mentioned (ideally what we should have done), and in
     > most
     >       cases, we do not have information from other models.I talked with
     > Richard
     >       and Cath about the confidence levels.  Richard,  I understand,  will
     > see you
     >       tomorrow (Thurs) .
     >
     >       I have denoted by a **** where we disagree with the draft
     > assessment, (in
     >       one case, I think the table is wrong- temp seasonality increases-
     > this
     >       should have at most the same confidence that summer/autumn warm more
     > than
     >       winter/spring (in my view low)) and we have no confidence in the low
     > level
     >       winds, especially over land. Some of the quantities didn't seem to
     > useful to
     >       me, and could be left out (especially if they are low confidence) We
     > would
     >       leave lightning and fog out as they are based on empirical indices
     > which are
     >       not well validated.I n general we are not sure how useful statements
     > on
     >       seasonality are, and the seasonal variation in the UK is not
     > currently well
     >       defned in many regions anyway. I don't think we have any confidence
     > in
     >       winter temperatures being more variable ( in most regions in most
     > models the
     >       opposite is true as albedo feedback reduces temperature gradients) .
     >
     >       I don't think we should put the reasons for the assessment in the
     > report,
     >       but we should keep a record of how we have assigned the confidences
     > in the
     >       final verison of the report
     >
     >       WIth best wishes
     >       John
     >
     >
     >
     >
     >
     >       John F B Mitchell, Head of Modelling Climate Change
     >       Met.Office, Hadley Centre for Climate Prediction and Research
     >       London Road,  Bracknell, RG12 2SY UK
     >       Tel +44 (0)1344 856613/6656 Fax+44 (0)1344 856912
     >       E-mail john.f.mitchell@metoffice.com
     >       <[2]http://www.metoffice.com/research/hadleycentre>
     >
     >
     >

