To: "Phil Jones" <firstname.lastname@example.org>
Subject: Re: Fwd: CRU surface temperature dataset
Date: Thu, 25 Jun 2009 10:59:26 +0100
Thanks for this great detail. I am thinking that probably a raw radiosonde dataset may be better (I tried this before using the LKS dataset but station density was an issue and only ended up with around 20 station pairs) - it sounds as though things have improved dramatically in that area and will look at the sources you suggest. My hope is that at least I can find hundreds/thousands of stations near to my high elevation surface ones for comparison. If not I could interpolate spatially maybe between radiosondes to my surface sites since free-air climate (not meteorology) should be relatively smooth in space. I cannot interpolate between surface stations.
I agree that reanalyses can be a can of worms (esp NCEP/NCAR)!
As for the surface I'll also look at the site you suggest and get back if I have any Q/problems. I appreciate the time you have taken to answer some of my Q!
>>> Phil Jones <email@example.com> 24/06/2009 13:09 >>>
I don't want to put off, but there is an awful lot of things
wrong with NCEP/NCAR.
They are probably OK for month-to-month variability, but if you look at some
of the figures in Simmons et al (2004) you'll see that for trends they are
practically useless before 1979.
There is just so much wrong with the sondes which together with the
introduction of satellite data in 1978/9 makes reanalyses awful.
The Simmons paper is about how much better ERA-40 is than NCEP/NCAR.
It is also telling you that you shouldn't be using NCEP/NCAR for
trends - and ERA-40
is only OK in Europe and North America.
A group of us are hopeful of getting an EU project funded to go
Reanalysis input - surface and sonde. The aim is to put in all the
surface and sonde data, so giving reanalysis better data input -
and putting back all the
data that missed the real-time cut. I'm not sure you're aware that
no back data have
ever got into the reanalyses. If data doesn't make the cut in real
time, it can never get
in later. The reanalysis source input doesn't collect back data!
You'd be better off getting one of the newer sonde datasets.
HadAT2 although developed
in 2005 is beyond it's sell-by date. Have a look at the attached
and this web site
Ra-ob core version 1.4 is the latest.
The drop off in surface data isn't the fault of GHCNv2. The
folks in Asheville are doing all
they can to get additional datasets. Currently about 2000 sites are
exchanged in real time.
If the sites you want are not exchanged by Met Services in real
time we can't get access
to them except by asking each Met Service and/or waiting till the
next volumes of the 10-year
books (for 2001-2010) get released.
CRUTEM3 has some additional station data going in for Australia
and Canada, but apart
from this we will have nothing more than GHCNv2. We could get a
load more from the US
quite easily, but coverage is reasonable there compared to the rest
of the world.
GHCNv2 and ourselves have lots of historic series, but these
aren't updatable in real
time, without continuous effort. Lots of projects were funded in
the US and Europe in the
1980s and 1990s to get loads of data digitized, homogenized and accessible.
It is possible to do things with daily data (SYNOPS) but these are
only generally good enough for the good countries.
This site has what is available in real time - since 2001. This
site can be very annoying.
There is a link back to NCDC.
At 17:48 23/06/2009, you wrote:
>Many thanks for your reply. This is very helpful, esp the Simmons paper.
>I am aware there are issues with reanalyses although I do want to
>try and use data representative of free air (and not contaminated
>with surface obs)- hence NCEP/NCAR rather than ERA-40 maybe, and use
>of pressure level data rather than 2 m or surface reanalysis temps
>(which I think the Simmons paper is about). I don't want the
>reanalysis to respond to surface issues and want it to be
>independent (purely based on radiosonde and satellite coupled with modelling).
>Of course this doesn't make the points irrelevant and I am looking
>at these while deciding what to use.
>As regards surface data, I am interested in the Tmean you mention
>used for CRUTEM3. Is this available and for how many stations?
>GHCNv2 is not good after 1990 since many stations stop! It is
>particularly dire after 2005 as you may realise? Please let me know
>what you think?
>Best wishes and thanks for your help re this.
> >>> Phil Jones <firstname.lastname@example.org> 22/06/2009 10:38 >>>
> I was away when your earlier message can in March, and I must have
> forgotten it when I got back to Norwich.
> We generally only put the gridded data on the web site. The
>station data that
> goes into CRUTEM3 is only monthly mean temperature. It is only
>since the mid-1990s
> that countries have routinely exchanged monthly mean Tx and Tn
>data. Many countries
> don't use these data to calculate mean T, instead using their
>historical methods based
> on fixed hours.
> We do have an archive of historic Tx and Tn (monthly) but this
>is almost entirely
> based on GHCNv2 sources. We use these data in products like this paper
>Mitchell, T.D. and Jones, P.D., 2005: An improved method of
>constructing a database of monthly climate observations and
>associated high-resolution grids. Int. J. Climatol. 25, 693-712.
> When you compare with Reanalysis trends you want to consider
>looking at ERA-INTERIM
> available from 1989-2008. There are also longer reanalysis products
>developed by NOAA
> (Gil Compo) from surface station data only (i.e. no sondes and no
> consistent through time).
> Are you aware of this paper? Basically reanalyses will be wrong
>before 1979 - except possibly
> in Europe and North America. This paper has the reasons why
>reanalyses will be wrong.
>At 15:06 17/06/2009, you wrote:
> >Dear Prof. Jones
> >You maybe had forgotten that I e-mailed you a while ago (March)
> >asking about access to data for surface stations for work on
> >temperature trends in complex topography (original e-mail and
> details below).
> >Since then I have been awarded a Royal Society Travel Grant to do
> >some work on this in the U.S. and I will be examining the GHCNv2
> >dataset in detail (which I have). I would really like to be able to
> >include a CRU dataset as well, since I did this in my original
> >research and these datasets are highly regarded.
> >If you are not the correct person to ask, maybe you could guide me
> >to the right person!
> >Many thanks for your reply.
> >Best wishes
> >Nick Pepin
> > >>> Nick Pepin 09/03/2009 16:43 >>>
> >Dear Prof. Jones
> >You may remember that a few years ago (2005) I published a paper
> >with Dian Seidel looking at temperature trends at high elevation
> >surface stations and comparing them with reanalysis trends. I wish
> >to update this work as part of another project, and was looking on
> >the UEA website to see if any of the original stations have been
> >updated. It is important that they are homogeneity adjusted as much
> >as possible.
> >It appears that nearly all of the datasets available on the web are
> >gridded and therefore interpolated (which I don't want since
> >interpolation influences what I am examining). Are any of the 3000
> >approx original stations available (mean monthly maxima and minima
> >are good enough) which are used to create CRUTEM3 etc?
> >In my original analysis I combined data from the CRU station dataset
> >and GHCN (some stations were in both) and I would like to do the
> >same again if possible. This is part of work looking at the effect
> >of topography on temperature trend patterns on a global scale (it
> >will be more detailed than preliminary work on this in the attached paper).
> >Many thanks for your help
> >Best wishes
> >Nick Pepin
>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 email@example.com
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