Thursday, May 17, 2012


date: Thu, 20 Dec 2007 16:39:06 -0000
from: "Robert Matthews" <>
subject: Re: BBC Focus magazine
to: "Phil Jones" <>

Great - thanks ! I see the story has been picked up on CC-NET; perhaps you should post this
really handy rebuttal on there, before this story "gets legs" and is picked up by all the
usual suspects (It's the Christmas silly season, and the papers are desperate for


----- Original Message -----

From: [1]Phil Jones

To: [2]Robert Matthews

Sent: Thursday, December 20, 2007 4:32 PM

Subject: Re: BBC Focus magazine

This story has been doing the rounds for the last 18 months. There is nothing new
in it. It's been wrong every time it's been used. It comes from people who have
no comprehension of the climate system. I was surprised this time, as I thought
this writer ought to have know better.
Just look at the global temperature series that you now have the associated errors
for. There is a lot of variability on the annual timescale. A lot of this is just
natural variability of the climate system. The trend is upwards. Some of the
variability of the global temperatures is caused by El Nino/La Nina events. El Nino
tends to make the world warmer and La Nina cooler. A measure of El Ninos is
the Southern Oscillation Index (the difference in pressure between Darwin
and Tahiti). When you regress this against global temperature you can explain
quite a bit of the high-frequency variance due to the major El Ninos and La Ninas
that have occurred since the mid-19th century. I wrote about how to do this in
1990 (see the pdf).
The upshot of this is that 1998 is about 0.15 deg C too warm because of the
El Nino influence - this El Nino being the biggest of the 20th century.
So 1998 could be considered the problem, not the later years. There hasn't
been much of an influence either way for the last 6-7 years.
So if 1998 is reduced by 0.15, we would have all 7 warmest years as the warmest 7.
Another way of looking at this is that all 7 years (2001-7) have a global
anomaly above 0.4 deg C. The only year before this with a value above 0.4 is 1998
(with 0.52). The last 7 years contain the second through eight warmest years in
the series.
Finally, as a climatologist, I wouldn't look at a temperature trend over such a
period as 10 years. I know, as you now do, that the global temperature series
has error estimates. Given these errors, it would be impossible to get a statistically
trend for any 10 year set of global temperature data chosen from any period in the
global temperature record. It is likely that you will get a few periods that might be
but then you have to consider that you'd expect about 5% of samples significant.
So, knowing this, put the global T numbers for 1998 to 2007 into an excel
and calculate the trend. It isn't significant - it definitely isn't allowing for the
Despite this the trend is POSITIVE. So despite starting with the warmest year, a linear
trend fit through the 10 years from 1998 to 2007 gives a POSITIVE trend.
So the world is warming....
It will continue to, it just won't be a monotonic increase. It hasn't been like this
the past, and it won't be like that in the future. There is a case for the temps to
have risen in a series of steps --- well to my eye anyway.
The 1998 record will get broken - we just need the next reasonable sized El Nino.

At 16:07 20/12/2007, you wrote:

Hi Phil

Thanks again for your help with the global warming figures. As it happens, the New
Statesman has just published a piece about whether global warming is still continuing
(it's here: [3] ). I'd very much welcome your
views on it.
Best wishes

----- Original Message -----
From: [4]Phil Jones
To: [5]Robert Matthews
Sent: Tuesday, December 18, 2007 10:07 AM
Subject: Re: BBC Focus magazine
The attachment has the information you are after. This also has the full press
release - all the background information that journalists could have asked for on
Dec 13. This should have gone out from the World Met Organization from
Geneva on Dec 13 as well.
The error ranges are shown in 2 different ways.
1 . Figure 1 (95% confidence values - so 'true' value will be in the range
19 times out of 20) with the global T values ranked from highest to lowest.
2. Figure 2 (top panel for the Globe, but also with the NH and SH there as well).
These are the same values (and ranges) as in Figure 1 but plotted as a
time series (the usual way).
You'll see the errors are larger further back in time - especially in the 19th
century. This is because there are fewer obs and the coverage gets sparser
as regions drop out. 2007 has slightly wider error bars as we've estimated
There are plots for smaller regions - the tropics, extratropics (30-90degrees N or
arctic and antarctic sea ice areas, and some more local series for the UK.
At 18:24 17/12/2007, you wrote:

Dear Professor

I'm putting together a piece about the current rate of global warming, and was very
interested in the data presented at Bali as summarised by the BBC News website here:

As the values are point-estimates extracted from a large number of measurements,
presumably they should have some kind of standard deviation error bars associated
with them. I wondered if you either knew the approximate size of these error bars
(or even a graph showing them)?
Thanks so much for your help with this.
Best wishes

Robert Matthews
Science Consultant, BBC Focus Magazine
47 Victoria Road, Oxford, OX2 7QF UK
Email: [7]
Tel: (+44)(0)1865 514 004 / Mob: 0790-651 9126

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

__________ NOD32 2729 (20071218) Information __________
This message was checked by NOD32 antivirus system.

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

__________ NOD32 2738 (20071220) Information __________
This message was checked by NOD32 antivirus system.

No comments:

Post a Comment