RealClimate’s deceptive graphs

I sometimes wonder about setting up a “climate hypocrite of the month” award.
Previous winners might include:

  • Chris Rapley, who earlier this year published a report stressing the importance of establishing a dialogue with the public, yet has recently being sending audiences to sleep with a droning monologue at the Royal Court Theatre.
  • Stephan Lewandowsky, who accuses climate sceptics of being conspiracy theorists, yet wrote an article claiming that there is a “subterranean war on science“.
  • Naomi Klein, who has been flying around the world on a tour to sell copies of her book on the twin evils of fossil fuel emissions and capitalism.
  • ATTP, for criticising people for commenting on things they haven’t read, as well as for saying that people are only welcome at sceptic blogs if they agree with the blog owner (he has banned most people who don’t agree with him from his blog).
  • Paul Nurse, who wrote “We need to be aware of those who mix up science, based on evidence and rationality, with politics and ideology, where opinion, rhetoric and tradition hold more sway”. But he himself is a socialist and supporter of activism, and introduced the recent Royal Society report, full of opinion and rhetoric yet devoid of science.

Anyway, December’s hypocrite of the month is Stefan Rahmstorf, RealClimate’s master of the misleading graph, who in his latest post has criticised a graph available as part of a “widget” at WattsUpWithThat:

Rahmstorf claims the graph is ‘deceptive’ because it plots monthly data. Perhaps he thinks throwing data away is a good idea. Perhaps he prefers the decadally averaged graph that Thomas Stocker and the IPCC are so keen on, that hides the warming slowdown so effectively.

Rahmstorf also says “One needs to scale the CO2 data correctly for an honest comparison with temperature, so that it can actually be used to evaluate climate scientists’ predictions of the CO2 effect.” Shub points out on twitter that this is circular reasoning – the graph has to scaled “correctly” to make it fit with Rahmstorf’s predetermined conclusion.

Another deceptive claim by Rahmstorf is that the data shown was “from several km up in the troposphere”. In fact the TLT data is strongly weighted to the near-surface atmosphere, with a peak in the weight function at about 2 km, so “several km” is misleading.

But the hypocrisy is that Rahmstorf is himself an expert in making misleading graphs. A good example is this from his Oct 1st blogpost:

Notice how the blue line appears to continue upward while the data itself shows the well known pause in warming. Rahmstorf has been producing this sort of deceptive graph for years, since at least 2007, in papers and in a report for the Copenhagen conference in 2009, see Lucia’s “Fishy Odor” and “Source of fishy odor confirmed” posts from 2009. In that case, Rahmstorf extended the smoothing interval from that used in a previous paper, but did not change the description in the caption. What he had actually done had to be reverse engineered by the climate bloggers.
“Rahmsmoothing” has also been discussed several times at Climate Audit.
Another obvious deceptive aspect of Rahmstorf’s graph is the horizontal green line labelled “Preindustrial Level”, creating the impression that in “preindustrial” times the global average temperature was unchanging.

[ Historical note: I first started learning about climate science in about 2007/8, at which time I thought that climate science was fine but some stories in the press were over-hyped. A natural first port of call was the RealClimate blog run by Rahmstorf, Mann, Schmidt and colleagues (who at that time I had never heard of). I was quite shocked by the attitude displayed by these people in various ways, and I particularly recall the Rahmstorf smoothing. It makes no sense to continue a smoothed graph up to the end of the raw data, and even if you do that, you should explain how you did it. This was bad enough, but Rahmstorf went a step further, failing to explain even when asked. This kind of behaviour from Rahmstorf and his team was a significant formative factor in my views on climate science. This comment posted yesterday on the deceptive graph thread summarises the RealClimate problem well and shows that five years later they have learnt nothing from their earlier mistakes. More on that story later. ]

Updates 11 Dec:

There are now two three more responses to Rahmstorf, from Bob Tisdale at WUWT, Mike Haseler at Scottish Sceptic, and Lubos Motl
And there’s another nice comment from Steve Harris at RealClimate.
Shub has a blogpost on the Realclimate effect.


9 thoughts on “RealClimate’s deceptive graphs

  1. Thanks for highlighting that. I’m not sure if I spotted Bob Tisdale first ( but one of you inspired me to write an “unusual” blog post based on it:

    Unfortunately, I imagine that only me and Lewandowsky will find the subject interesting and only one of us will support what I’m saying.

    But on the widget, I was frankly disappointed by their effort. I could have done a better job without trying and Even if I were an alarmists I’d not bother putting it on my website.

  2. Dear Paul,

    In the caption he says that it’s a LOESS smooth. It’s easy to do this in R with a couple of lines of code and, yes, the smoothing goes right up to the end of the series. If, in addition, you calculate the uncertainties on the estimated LOESS trend they flare at each end. See also box 2 of chapter 2 of the IPCC AR5 WG1 report for an example with uncertainty ranges.

    The caption of the figure also says “Values are given relative to a preindustrial baseline, the exact definition of which may be debated but only adds a minor uncertainty – here it was chosen as the mean temperature of 1880-1900.” Which is, at least, specific. I guess you would quibble about “minor uncertainty”. The question really is, what do you think that uncertainty should be?


  3. John, thanks, yes he does say it is a LOESS smooth, which is a different method from the older examples. The LOESS method just uses increasingly one-sided averaging near the endpoints. But the LOESS smoothing method depends on a parameter that very significantly influences the result, as the R documentation illustrates. Rahmstorf does not say what parameter value he used and does not acknowledge that this changes the picture.

    I think the whole concept of a “pre-industrial baseline” is misleading, because it creates the impression that there was a pre-industrial temperature and then warming started.

  4. Yes, the “pre-industrial level” should be a band at 95% or so, going back to the last warm period. It’s grossly misleading to suggest that “pre-industrial” was a set temperature, without even so much as an “average” tacked onto it.

    Of course the effect would be to reduce alarm considerably, as we would barely be poking out of such a band.

  5. Paul,

    Yes, changing the parameters changes the wiggliness. But if you choose a parameter which gives more wiggles, you reduce the number of data points used in the trend fitting at any point and hence the uncertainty can go up. More smoothing implies more data points and lower uncertainty. However, if you look at the AR5 plot, which isn’t very different from Rahmstorf’s in the central estimates, the uncertainty at the end is such that it is consistent with a range of “underlying trends”.


  6. Thanks for your post and your remarks. I just have one idea that doesn’t want to leave me: there are so many graphs and analysis, but not enough understanding and real action. Many promises, many debates and forums, and, as you just have said, a lot of hypocrisy…..

  7. I don’t find it easy to determine the motive of those on both sides of the AGW divide who seek to cherry pick or adulterate data. As the saying goes “if you torture the data long enough it will tell you anything you want”. AGW and particularly CAGW is a religion and requires the same sort of belief system.

    I see you are a Richard Feynman fan Paul – well who isn’t or if their not, should be.

    Centred in this Caltech commencement address of his from 1974 is a section that tells you all need to know about why Climate Science is all wrong and ceased to be truly about science a long time ago.

    For anyone who hasn’t seen it before it underlines what science should be and Climate Science fails on accounts.

  8. The results are now in for the climate hypocrite of the month award for January.

    The clear winner, with a brilliant late entry, is Greg Laden.

    In this post at his own blog, Laden encourages people to sign a petition calling for Willie Soon to be fired from his post at the Harvard-Smithsonian Institute for Astrophysics.

    On the very same day, 29 Jan, he wrote a blog Bad faith criticism of science, where he complains about “anti-science strategy of personal attacks against individual scientists in an attempt to discredit valid scientific research one might find inconvenient”, adding later in the piece that “Some of this is not so much about science (or anti science) but just plain harassment”.

    Runners-up include ATTP, who it turns out ran a left-wing political blog before starting his climate blog where he accused climate sceptics of being politically motivated, and Chris Mooney, who wrote an article about the problem of the dysfunctional, polarised climate debate, describing James Inhofe as “one of the most notorious deniers”.

  9. Climate hypocrite of the month for March 2015 is Jim Bouldin. Jim is a former member of the Realclimate team, who left them after some kind of disagreement within the team.

    In a sneering tweet to Roger Pielke, he wrote
    “Junior accusing climate scientists of dishonesty is he?”
    (in respone to this comment). Pielke replied that he wasn’t. Bouldin made further vague sneers and then said
    “He’s accused scientists of fraud because of his own statistical stupidity on complex topics.”
    Now comes the hypocrisy:
    “Malice or otherwise, you don’t accuse people of those kinds of things without definite evidence.”
    As I replied on twitter, this is exactly what Bouldin himself was doing! He said that Pielke accused people of fraud, without providing any evidence. He replied
    “No, sorry but you’re wrong. He flat out, full bore, accused Rahmstorf of fraud.”
    We now know who he’s talking about, but still no evidence at all of Pielke having accused Rahmstorf of fraud (as far as I am aware, Pielke has only accused Rahmstorf of cherry-picking). Jim is doing what he himself said you don’t do.
    Bouldin continued with
    “The degree to which some of you “skeptics” have assumed the right of defamation is breathtaking”
    to which Howard Goodall replied
    “Indeed, Jim. To a level exceeded only by others’ capacity for mind-numbing hypocrisy”.
    As well as the hypocrisy of demanding evidence from others while not providing any yourself, there is the more general hypocrisy of reacting angrily when anyone dares to criticise some climate scientists, while remaining quiet about the smears and witch-hunts against others.

    A worthy runner-up is the Guardian, who run adverts for Shell alongside their campaigns to “keep it in the ground” and “divest from fossil fuels”.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s