Seen elsewhere
The calendar

Click to buy!

Support

 

Twitter
Buy

Click images for more details

Recent posts
Recent comments
Currently discussing
Links

A few sites I've stumbled across recently....

Powered by Squarespace
« Melting ice | Main | Wheat in India »
Monday
Jan302012

A Rose on winter

Over the weekend there was quite a lot of interest in David Rose's article in the Mail, which addressed new figures from the Met Office which appeared to confirm a lack of any warming in the last 15 years.

 

The supposed ‘consensus’ on man-made global warming is facing an inconvenient challenge after the release of new temperature data showing the planet has not warmed for the past 15 years.

The figures suggest that we could even be heading for a mini ice age to rival the 70-year temperature drop that saw frost fairs held on the Thames in the 17th Century.

Based on readings from more than 30,000 measuring stations, the data was issued last week without fanfare by the Met Office and the University of East Anglia Climatic Research Unit. It confirms that the rising trend in world temperatures ended in 1997.

Rose was also warning of further cold on the way, based on an assessment of Solar Cycle 24 and 25.

The Met Office have now responded, with a blog post that has a whiff of Bob Ward about it: "includes numerous errors in the reporting of published peer reviewed science". The argument seems to be that if you take decadal averages it is still possible to obscure the plateau in the temperatures.

Or words to that effect.

 

PrintView Printer Friendly Version

Reader Comments (65)

Jan 31, 2012 at 11:27 AM | Phillip Bratby

It would be interesting if Richard Betts would come on here and defend the Met Office statement. David Whitehouse is not very complementary of the Met Office trick to show the incline. Damn lies by the Met Office using statistics. The Met Office is being totally deceptive in trying to hide the lack of warming bu using ten year averages.

Hi Phillip

The Met Office blog post has been updated to include the graph of year-by-year temperatures up to 2011.

Cheers

Richard

Jan 31, 2012 at 1:05 PM | Unregistered CommenterRichard Betts

@ Richard Betts

Why are the Met Office not using their actual Decadal forecast/predictions?

Jan 31, 2012 at 1:11 PM | Unregistered CommenterGreen Sand

@richard

"...for Mr. Rose to suggest that the latest global temperatures available show no warming in the last 15 years is entirely misleading"

I really struggle to see how the response can in any way be described as refuting Rose's contention.

The rebuttal consists of decadal average graph that smooths the period under discussion into a single datapoint, and the yearly graph *supports* Rose's contention, albeit begrudgingly by reducing the period in question to a handful of pixels.

A response that says he's *right* - and he *is* right - in his claim, but which then shows why this contention shouldn't be considered significant would appear far more honest than what essentially looks like misdirection.

Jan 31, 2012 at 1:17 PM | Unregistered Commentermrsean2k

Jan 30, 2012 at 4:53 PM | mrsean2k

Two questions for Richard (or anyone)

Hi Sean,

Thanks for the excellent questions. I asked my colleague John Kennedy about these, and his answers are below:


(1) Why has the uncertainty increased for the 2000's when compared with all the averages as far back as 1940? Surely the bounds should contract and not expand?

"The coverage of data in the 2000s is actually poorer than it was in some earlier decades, so it's not inevitable that uncertainties will monotonically decrease. The land part of HadCRUT3 is updated using CLIMAT messages that only come in from a subset of stations. More than 2000 stations report CLIMAT messages each month, but only around 1200 of those have sufficient data in the climatology period for us to calculate anomalies. The uncertainty associated with poor coverage is the largest component of uncertainty and estimating it is tricky. We've never had observations over large areas of the polar regions so we have to estimate the uncertainties using reanalyses. The estimated coverage uncertainty actually hasn't changed a lot since 1850 because we never have observations from some regions:
http://www.metoffice.gov.uk/hadobs/hadcrut3/diagnostics/global/nh+sh/annual.png

The coverage uncertainty can also behave counter intuitively. If you add a small number of stations from a very variable region you don't necessarily reduce the uncertainty in the global average.

CRUTEM, the land component of HadCRUT, has recently been updated (just out in the JGR atmospheres in-press queue, but not yet incorporated in HadCRUT) and a lot of station records have been brought more up to date and some gaps have been plugged in the Arctic.

In addition, bias uncertainties (e.g. urbanisation) are constrained to average to zero over the climatology period, which inevitably means they have larger variance outside that period."

(2) What happens to that graph if it is plotted with a different central value for the 10 year period in question? We may choose to talk about averages of 1990 - 1999, 2000 - 2009 etc. but that's an arbitrary choice for rhetorical purposes. What happens to that diagram if the data are partitioned differently?

"The estimated uncertainties on individual years are so large that we can't unambiguously rank them. The uncertainties on individual decades are smaller relative to the differences between them so we can. This remains true for the past three 'decades' if we shift the start date for each decade around. This is a nice thing about the decadal plot, it's not overly sensitive to start and end points. It wiggles only a little if you change them."

John has also done bar charts of the decadal averages for different start and end points to illustate this, but I'm not sure how to attach them here. If anyone can tell me how to attach a bitmap image then I'll do that - or maybe I should just send it to BH and ask him to paste it in.

[BH adds: see here.

Thanks again for great questions,

Cheers

Richard

Jan 31, 2012 at 1:18 PM | Unregistered CommenterRichard Betts

@Richard

Thank you for responding. Why is the immediate reaction of the Met Office to mislead the public by misusing statistics? How can you defend such behaviour? (Although I note you did not defend it) Why is the misleading bar chart still presented?

Why can't the Met Office just give us honest science?

Jan 31, 2012 at 1:55 PM | Unregistered CommenterPhillip Bratby

mrsean2k @12:18 pm - I'm also baffled by the width of the bars on the decadal graph. I wondered if it was due to variance of the annual temps within the decade but it does not appear to be. If you look at both graphs (the decadal and new yearly one) on the metoffice blog post http://metofficenews.wordpress.com/2012/01/29/met-office-in-the-media-29-january-2012/, then in both cases loads of features are not labelled or explained. Maybe nobody at the Met Office could be bothered to generate graphs specifically for this purpose.

I have to say that I've seen far more misleading newpaper articles about climate than the one by David Rose. It is a bit misleading in the way it projects as a general message, in a rather breathless way, that everything is due to the sun and we'll all be skating on the Thames in July come 2050. But if you look beyond the superficial level, there are quite a few caveats and discussions of uncertainties. The Met Office have chosen to reply in a rather clumsy way in my opinion. They focus on this silly business of rising vs. non-rising temperatures in the last n years or m decades, and are rather economical with the statistical truth, as is well discussed by David Whitehouse in his post at the GWPF, http://thegwpf.org/the-observatory/4868-the-mail-on-sunday-the-met-office-and-the-temperature-standstill.html. They make a few comments about the solar issue, but don't really address David Rose's key point at all. This was that some people believe the sun has a much greater importance than is allowed for it in the climate models of the Met Office, and that this is consistent with the observed pattern in temperatures.

Jan 31, 2012 at 2:19 PM | Unregistered CommenterJeremy Harvey

Hi Richard its still not actual temps though is it...

It is an anomaly graph..

How many of the public will spot range of Y-axis is about 1.0 C,
It will come across across as Actual temperatures.

Not a 1C rise, in a century plus, and a very noisy signal..

Thanks for getting it changed though (twitter is good for something after all.)

It wil be interesting if the anomlalies go down for a few more years? ;-)

Jan 31, 2012 at 2:28 PM | Unregistered CommenterBarry Woods

@jeremy

I received this response to my query via the press office - full credit for that of course:

---------- START ----------

Here's a response from one of our scientists, I hope it answers your questions.

The uncertainty calculation was similar to that used in Brohan et al.
(2006). http://www.metoffice.gov.uk/hadobs/hadcrut3/

Instead of a binomial filter, we used a running 10-year average of the annual values. 'Decades' were 10-year averages starting in years ending with 0. e.g. the most recent decade was 2000-2009.

Two components of the uncertainty are larger in 2000-2009 than in some earlier decades: coverage uncertainty that arises because we don't have observations in all grid boxes, and bias uncertainty associated with
sensor exposure.

Coverage uncertainty is related to how many gridboxes we have observations for and where they are located. In 2000-2009 the coverage of land stations is lower than it is in some earlier decades, so it's not inevitable that uncertainties will monotonically decrease. The land part of HadCRUT3 is updated using CLIMAT messages that are only sent from a subset of stations. More than 2000 stations report CLIMAT messages each month, but only around 1200 of those have sufficient data in the climatology period for us to calculate anomalies.

Bias uncertainty is constrained to be zero in the climatology period (1961-1990). There may be non-zero biases in that period, but because we are dealing with anomalies rather than actual temperatures, uncertainties associated with biases in the climatology period will show up as a slight widening of the uncertainty range outside of the
climatology period.
---------- END ----------


Names snipped, but otherwise as it came in (any errors almost certainly my transcription)

I've done some overlays / eyeballing of the original chart, and decadal averages as calculated from source HADCRUT data, and for the earlier values, the average doesn't seem to be centred within the uncertainty bounds. Probably a misunderstanding on my part, but worth a bit of playing around with.

Jan 31, 2012 at 2:34 PM | Unregistered Commentermrsean2k

Richard - more questions for you.

The Met Office is funded by us taxpayers. Is it not therefore the duty of the Met Office to present the results of all the science that the funding enables in the most honest light?

When I was employed (ie before retirement), although I had done some statistics at university and had been sent on a few statistics training courses, whenever my work involved anything but simple statistics, I would go and consult the company statistician. Does the Met Office employ anybody who is a qualified statistician? If so, would he be consulted and agree to what we have seen on the Met Office blog? If not why not?

The more I see of the Met Office, the more sceptical I become of anything the Met Office produces. I am sure this sort of behaviour by the Met Office is what makes more and more people more and more sceptical. Just my thoughts on communicating science to the public.

Jan 31, 2012 at 2:39 PM | Unregistered CommenterPhillip Bratby

I think you might be refering to Doug Mcneall - Philip

https://twitter.com/#!/dougmcneall

Jan 31, 2012 at 3:13 PM | Unregistered CommenterBarry Woods

"The Met Office blog post has been updated to include the graph of year-by-year temperatures up to 2011."

I see they have. Updated the blog page, after comments, without indicating that it's an update.

The uncertainty range may come from the number of stations used, which increased to a peak in the 1970s and has since inexplicably declined. Using a 1/root(N) type argument would then give narrowest error bars in the 1970s.

Jan 31, 2012 at 3:20 PM | Unregistered CommenterPaul Matthews

It is about 'messaging'. The Metoffice is totally consumed by it as an organ of Establishment. David Rose's article is precisely the type of alternative interpretation the Metoffice should welcome, and even be authoring. As individuals, they appear less stifled by the chains of this oppressive and scientifically suffocating, politically correct messaging,

They should have learned by now this tactic has been spectacularly counterproductive. Not surprising, the public were never as gullible or as vulnerable to propaganda as they thought. Only with children may they have some success with climate brainwashing

It varies, but individual boldness breaks out from time to time. Richard Betts, certainly, although I suspect he is biting his lip a bit on this, and Paul Hudson, formerly Metoffice now at the equally stifling BBC, can be bold once in a while on his blog.

Jan 31, 2012 at 3:51 PM | Unregistered CommenterPharos

mrsean2k - what you write is entirely consistent with Paul M's conjecture. And it fits with my statement that the graph is an example of poor science communication - an 'error bar' is included without any hint as to what it actually means.

Just to belabour the point I made above once again - I really think that the Met Office don't do themselves any favours with this type of posting - as Phillip Bratby observes, all this massaging of yearly anomalies into decadal ones is fairly transparently an attempt to stay 'on message', and whenever people see that sort of thing, it makes them very wary. We expect that kind of thing from politicians, not from scientists. Also, as communication attempts go, not addressing the main point in David Rose's article is clumsy. Why did the Met Office not say something along the lines of "Well, our models do incorporate some degree of dependence on solar activity, and they predict that the effect of the sun should be quite low. That is what the Met Office research by Gareth Jones and Peter Stott was about. Some scientists, referred to in the article by David Rose, believe that there are other mechanisms whereby solar activity affects temperatures and that these may be playing a role in current temperature trends. At present, this is not a mainstream view and this is why it is not currently in our climate models." I think most of us here would accept that as an honest reply to David Rose's article (even if some of us would not agree it was correct).

Jan 31, 2012 at 5:17 PM | Unregistered CommenterJeremy Harvey

Guess what, ...

The whole world's head is asploding from David Rose and Tim Lambert who has a crush on Rose....has nothing on (!). Hope all's well int he house of Deltoid.

Feb 1, 2012 at 3:50 PM | Unregistered CommenterShub

@Shub

is that your favourite Captain Beefheart lyric or are you of the meds again :-)

Feb 2, 2012 at 11:47 PM | Unregistered Commenterdfhunter

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Post:
 
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>