To: Phil Jones <firstname.lastname@example.org>
Subject: Re: Visit to Met Office
Date: Tue, 10 Feb 2009 09:54:16 +0000
Cc: David Parker <email@example.com>
as David says I'll be away in Oklahoma first week in March. Antarctic
data first piqued my interest with the Science paper on raobs trends
which was clearly non-physical but hard to nail down how wrong it was. I
did some minor digging into READER and found that in the UA domain it
was qc'ed but not homogenised. I've made a rather rash assumption that
this would also be the case for the surface data but am happy to be
Its clear to me that Antarctica is a uniquely difficult environment to
collect long-term homogeneous data in. So I have substantial doubts that
all the manned station pegs in Steig et al. are adequate. Does this
really matter? I'm not sure.
What Steig et al., satellites, and potentially reanalyses does do is
allow us, in principle, at least to get around the no-neighbours issue
in assessing homogeneity away from the peninsula.
For example we could use a bootstrapping of the Steig et al approach by
creating say 50 realisations of each station series using randomly
seeded combinations of manned station pegs as the S et al. RegEM
constraint (excluding the candidate station) to make a neighbour
composite ensemble. We could then add in the available reanalysis field
estimates and satellite estimates and make a reasonable punt about the
existence and magnitude of any breaks based upon multiple lines of
evidence (of course, we lose some of these before 1979 ...). We could
use this information to assess in a more rigorous way than has been done
to date the homogeneity of these sparse stations. Then cleaned up data
could be fed back through Steig et al. afterwards to see how it impacts
that analysis making for a nice clean self-contained study.
My understanding from the blog discussion of Steig et al. is that the
analysis step is fairly trivial so such an ensemble realisation approach
should be plausible with a humble PC so long as it has the coding
Of course, this doesn't resolve any fundamental methodological concerns
about the S et al. approach that may exist but it does give us a
reasonable chance of creating a much more homogeneous READER manned
station dataset for next IPCC AR and our future products.
My suspicion is that actually changing the manned station data in this
way may make S et al. more different to the straight average of the
READER data as used (effectively) in AR5 and point to the importance of
the long-term homogeneity of the data pegs in RegEM ... this may, of
course, be felt to be a can of worms too far ...
On Mon, 2009-02-09 at 16:53 +0000, Phil Jones wrote:
> I think I misinterpreted your email when in Switzerland. I think I thought
> you wanted a talk and a possible project. Now I read it and it is just a
> possible project.
> I've done a lot with the Antarctic temperature data - I also have an
> archive of MSLP data for most sites (for some it is station level pressure).
> With regards homogeneity it is difficult to do much beyond the Peninsula
> (and be confident about anything) as the stations are too far apart. There is
> an issue I could ask Adrian - whether ERA-INTERIM is good enough since
> 1988? This could also assess the AVHRR, but this may be circular.
> I've read Steig et al now, and I can see all the comments on the CA and
> RC sites about some of the data. It seems that BAS have made some mistakes
> with some of the AWS sites. The only AWS site used in CRUTEM3 is the one
> at Byrd, as this is at one of the manned sites. The issue with the AWS's is
> getting reasonable data in real time. Whilst I was away the checked monthly
> data arrived for 2002! I will add Byrd's data in. The problem is
> that some sites
> get buried, but still seem to transmit.
> What Steig et al have done is a paleo-type reconstruction of the
> full field
> from the AVHRR for a recent period and extended it back to 1957. If the
> data are OK, all you're assuming is that covariance structure
> remains the same.
> I did this paper (attached) ages ago, but it doesn't seem all
> that relevant.
> Anyway - I do need to come down to see Ian. Possibilities would be coming
> mid week, say Feb 25/26 or March 4/5. How do these dates suit? I'd need to
> spend the night - maybe that Travel-lodge near you, it is only one night!
> At 16:04 30/01/2009, David Parker wrote:
> >Thanks. I hope the GCOS meeting goes well: Roger Saunders will be there.
> >We look forward to your thoughts on the Antarctic data, and to your
> >visit whenever that may be convenient for you,
> >On Fri, 2009-01-30 at 15:56 +0000, P.Jones@uea.ac.uk wrote:
> > > David,
> > > The Swiss extremes workshop has afternoons off for skiing.
> > > As I don't, I've been on 60 or 90 mins walks along snow covered
> > > trails. Snow is 1m deep off the trails.
> > > Anyway back now. So looking at emails. As the sun drops,
> > > the temperature plummets. I'm at the GCOS Imp Plan meeting
> > > next week in Geneva. Back in CRU on Feb 6.
> > > I've been reading the Steig et al paper. I've looked
> > > at homogeneity issues with the Antarctic data in the past.
> > > Difficult to do much except in the Peninsula. Anyway,
> > > I'll give your proposal some thought. Will talk to others
> > > like Kevin T next week as well about the paper.
> > > Glad to hear Ian is settling. It would be a good idea
> > > to do two things on the visit. I'm sure we can think of more!
> > > Glad also you're helping out Brian. I just couldn't
> > > rearrange my UEA teaching again - already done this so I can
> > > be here now and Geneva next week.
> > >
> > > Have a good weekend - if a little cold!
> > >
> > > Cheers
> > > Phil
> > >
> > > > Phil
> > > >
> > > > Peter Thorne and others have suggested that you visit us in the near
> > > > future to set up a project in which CRU would homogenise the "Reader"
> > > > surface temperature data for Antarctica. This subject arose in
> > > > connection with Steig et al.'s paper on Antarctic temperatures in last
> > > > week's NATURE, and is also relevant to the possibility that we may
> > > > include interpolations over the Arctic Ocean and Antarctica in our
> > > > analyses for IPCC AR5. Peter challenges the results of Steig et al. on
> > > > the grounds that the in situ surface temperatures may not be
> > > > homogeneous. Maybe you could even give a seminar on e.g. Antarctic
> > > > observations.
> > > >
> > > > Please let me know when a visit would be convenient for you. You could,
> > > > of course, combine it with a review of Ian's progress. Ian is now well-
> > > > settled into using our computing systems, and has started to calculate
> > > > r-bar from the daily precipitation fields for the UK regions, with a
> > > > view to estimating uncertainties in the regionally-averaged daily
> > > > values. As a cross-check, and to gain a deeper appreciation of this
> > > > myself, I have independently written some software to calculate r-bar.
> > > > This is leading to some ideas which I will send to you when I have had
> > > > more time to think them through.
> > > >
> > > > I understand you're busy as I am expecting to attend the Malaria meeting
> > > > at Imperial on 12-13 Feb when you aren't available.
> > > >
> > > > Hope you've had good meetings in Geneva
> > > >
> > > > David
> > > >
> > > > --
> > > > David Parker Met Office Hadley Centre FitzRoy Road EXETER EX1 3PB UK
> > > > E-mail: firstname.lastname@example.org
> > > > Tel: +44-1392-886649 Fax: +44-1392-885681 http:www.metoffice.gov.uk
> > > >
> > >
> > >
> >David Parker Met Office Hadley Centre FitzRoy Road EXETER EX1 3PB UK
> >E-mail: email@example.com
> >Tel: +44-1392-886649 Fax: +44-1392-885681 http:www.metoffice.gov.uk
> 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 firstname.lastname@example.org
> NR4 7TJ
Peter Thorne Climate Research Scientist
Met Office Hadley Centre, FitzRoy Road, Exeter, EX1 3PB
tel. +44 1392 886552 fax +44 1392 885681