Sunday, June 17, 2012


cc: Byron Gleason <>,,, Russell Vose <>, Aiguo Dai <>, Lisa Alexander <>, David Parker <>
date: Thu Jul 14 16:48:04 2005
from: Phil Jones <>
subject: Re: Some more thoughts on DTR map for 1979-2004
to: Kevin Trenberth <>

I've asked Lisa to produce a plot for 1979-2003 so we might get that tomorrow. I was
going to suggest we look at the CRU data but Aiguo has used that to some extent.
As Spain was so different in the two maps, I've sent an email there for the definitive
answer. I'm involved in a paper on this work, so asked for the trends over the 2 periods.
I'll study them all - including Jay's tonight.
Aiguo's grid looks as though it could degraded to Russ's. Not sure if Lisa's could
be made available in a grid form - maybe at the HC model grid. If we got these
we could then look at spatial correlations.
Need to check what the threshold for years acceptable is first. Russ's should be the 18
that Byron used for the 1979-2004 mean temp maps. The other way to check
Russ's is to average max and min and then compare with the mean trend map
for the year for 1979-2004 - and similarly for the others. We have some more confidence
in that field.

At 15:57 14/07/2005, Kevin Trenberth wrote:

Hi all
I am attaching some figures from Aiguo Dai. These include the trends in DTR, clouds and
precip for 1050-2004 and 1979-2004. He has also produced a number of other figures,
such as correlations among these fields: DTR mostlt negatively correlated with cloud
(exception Africa), cloud strong positive correlation with cloud everywhere.
For convenience I am attaching the other ones as well.
Notes from Dai: The precip is from Chen et al. (2002 and updates). The merged DTR and
cloud cover were
derived from CRU_TS_2.02 and my synoptic obs. based on which data set had better
(i.e., larger number of reports or stations for the CRU case) for each individual
monthly/season at each
grid box and a common/overlapping period was used to adjust the two data sets to have
the same mean.
This is a preliminary analysis that will be explored more as part of my NOAA proposal.
ASOS cloud data
since 1994 for N.A. were not used and U.S. military cloud data were included.
Firstly for 1950-2004 there seems to be a very good agreement with Lisa's map thru 2003
except Aiguo has Australia, Africa, S America and other areas also filled in.
However for 1979-2004 there are major differences with the Vose one from NCDC. South
America still has negative trends in the Dai work vs positive for Vose. The US,
Australia and Europe have distinctly more negative trends in Dai: the sign changes in
many points. In Spain it goes the other way? There seems to be some agreement at some
islands where presumably the data are limited and more likely similar: Hawaii, Bermuda,
near Madagascar,?? The lack of reproducibility here is very disturbing. Lisa can
weigh in on which, if any is closest to hers.
Comments? Suggestions?
Kevin Trenberth wrote:

Hi all
I am sufficiently concerned about this that I have asked Aiguo Dai to use his data to
compute DTR trends for 1979 to 2004. Aiguo has all the synops and so values are based
on a much more comprehensive dataset with wider coverage, including oceans, but perhaps
with problems and inhomogeneities such as ASOS. Still it will be a useful sanity check.
Phil Jones wrote:

Kevin, Byron,
Agree that we will have to wait till Russ gets back to check this out. Added Russ
on this reply. Russ sent Fig 3.2.2 for the ZOD, which we still have in the FOD figure
at the moment. We will need this updating. Presumably the new one will be similar
to one from the ZOD. That figure had a trend of -0.08 C/decade for 1950-2003. From
a guesstimate of the trend over 1979-2003 in this figure, the trend is about -0.2 C in
total. This sort of trend is in the first light blue category, just less than -0.1
I reckon that the average of the trends in the 79-04 map would average to about this
So, it seems OK. I am though a little alarmed at how variable the trends are
between adjacent boxes. From knowledge of the likely variations in spatial density,
most of these seem to occur where station numbers are likely low. Where density
is high (US, Canada, northern/central Europe, northern Russia and China/Japan
and eastern Australia) changes between adjacent boxes are a lot smoother.
I also know that DTR amplifies any problem in station homogeneity (as opposed
to the mean which reduces it). Problems are usually of the opposite sign for
max and min.
So, can you see if a similar plot for 1950-2004 would be better behaved. This period
would be 55 years as opposed to 26. It could be that we might need to go to more than
18 years to get reliable trends for 1979-2004 - if you're using the same threshold as
mean temp maps.
I have plans with Russ to look at these data for 1958-2001 and 1979-2001
with the ERA-40 Reanalyses (along with Adrian Simmons). ERA-40 should
be good enough for the shorter period to assess reliability, but this will beyond
the AR4.
At 21:39 12/07/2005, Byron Gleason wrote:

Russ Vose did the analysis and currently he is on vacation in Pennsylvania, and will
return Monday.
- Byron
Kevin Trenberth wrote:

Hi Byron
Thanks for the figure.
I have to express surprise over the figure. It shows a lot more red than I expected
based on other material. Over the US Dai has produced DTR trends for a slightly
different period: 1976-2004 and it is much bluer all round. I am also a bit alarmed at
some spots where trends are opposite: 2 large positive spots near others of opposite
sign. Is this consistent with Alexander et al 2005? Is this consistent with Fig 3.2.2.
Byron Gleason wrote:

Fig. 3.2.11 (global map 79-04 of dtr trends) now available on
This map was a little tricky style wise because it is temperature related so I wanted to
use the same color scheme as the other temperature maps, however there isn't global
coverage like the other temperature maps, thus the oceans end up being mostly grey.
Nevertheless, I think it looks better than if I used the precip. color scheme (it would
be a little atypical to depict temperature using green and tan).
- Byron
Phil Jones wrote:

Dear All,
I've picked up a file from Kevin where he has put all the new figures. Also
them all myself, and have ben looking at them over a coffee for the last 10-15 minutes.
Comments, thoughts under each variable.
Temperature Maps
The layout and grey/white are fine on these. The aim with the 1979-2004 ones
is to have all 5 for this period with the same scale, which they do.
As we are likely to get comments on these in this review round, I reckon the
ones we currently have are fine. The annual plot for 79-04 will look a bit
washed out cf 1901-2004, but it wouldn't if it was the DJF plot for the same
So, I'm happy to go with these.
Precipitation Maps
These now have the better coverage (compared to the ZOD), so you've solved
that issue. The problem with these plots is the two greys. There is a lot more
little trend areas on these maps as opposed to the temperature ones. To be
consistent with the temperature ones, I think we will have to go with the
light grey for missing and white for little trend. There should only be a largish
areas of white on the 79-04 annual map. There should be a lot less on the
seasonal ones. Apologies for getting this wrong the other day.
Can you make these changes to these 6?
The only other plots needed are the one for DTR (map will be 3.2.11, annual
for 79-04, or whatever final year Russ is happy with) and a max/min/dtr annual
series for the globe.
Also there is the time series plots for precip, which Dave said all that was missing
was the longer CRU series back to 1901.
Thanks again for all your help with these figures.
Thanks Jay for the details of how these have been produced. I'll use this to
send Kevin some captions, if I get a chance today.
At 00:23 12/07/2005, Kevin Trenberth wrote:

Thanks Byron
I think I got them all. But have not looked at them: Phil
I'll try to redo the Fig file tonite.
Byron Gleason wrote:

Figs (with new naming scheme) can now be found in:
Note: everything has been centered on 0.0 degrees and the tmean maps have been redone to
account for a mistake where I was still saying "Grey Intervals" and it should have said
"White Intervals".
I will probably get a new ftp directory in the future, but for now this should suffice,
and more maps to come tomorrow.
- Byron
Kevin Trenberth wrote:

Hi all
I have put the new figures into my file. I do note that 180 is in the center. For a
lot of fields this is desirable because of the large effects of El Nino. But for land
stuff it seems desirable to have the break elsewhere, perhaps with 0 meridian centered.
I think this will be the case for precip, and so it probably should be for T too? It
will mean we have a mixture in the chapter but that may be better that strict adherence
to something that makes no sense in some cases.
I am opening this up for your views.
Byron Gleason wrote:

Kevin E. Trenberth e-mail:
Climate Analysis Section, NCAR
P. O. Box 3000, (303) 497 1318
Boulder, CO 80307 (303) 497 1333 (fax)
Street address: 1850 Table Mesa Drive, Boulder, CO 80303

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
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

Kevin E. Trenberth e-mail:
Climate Analysis Section, NCAR [4]
P. O. Box 3000, (303) 497 1318
Boulder, CO 80307 (303) 497 1333 (fax)
Street address: 1850 Table Mesa Drive, Boulder, CO 80303

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

No comments:

Post a Comment