Judith Curry Opens Mouth, Inserts Foot
NOTE: See the UPDATE at the end of this post.
I didn’t expect Judith Curry to embarrass herself more than she did with her fawning over Murry Salby’s folly. But she’s topped (perhaps I should say “bottomed”) herself by a huge margin.
Anthony Watts was so excited he actually suspended his blog hiatus to report the story. He quotes the GWPF that “BEST Confirms Global Temperature Standstill,” and cites a story in the Daily Mail reporting that Judith Curry (a member of the Berkeley team) has roundly criticized Richard Muller (leader of the Berkeley team), on much the same basis. He quotes Curry herself, from the article:
As for the graph disseminated to the media, she said: ‘This is “hide the decline” stuff. Our data show the pause, just as the other sets of data do. Muller is hiding the decline.
With this statement, Curry reveals how little she understands the data created by the team of which she is a member, let alone the “other sets of data.” Which we might have anticipated, given that from what we’ve seen, she made little or no contribution to the actual Berkeley analysis.
Judging by her comments, it’s a good thing for the Berkeley team that she didn’t. The Daily Mail quotes her thus:
‘There is no scientific basis for saying that warming hasn’t stopped,’ she said. ‘To say that there is detracts from the credibility of the data, which is very unfortunate.’
That’s a pretty strong statement, Judith. Do you have a scientific basis to back it up? Have you actually analyzed the data from your own team? Do you even know?
Now that the Daily Mail and Anthony Watts have not only let the cat out of the bag but put it on display to make a “scandal” out of it, Curry seems to want to distance herself from her criticism. On her own blog she says:
I told Rose that I was puzzled my Muller’s statements, particularly about “end of skepticism” and also “We see no evidence of global warming slowing down.”When asked specifically about the graph that apparently uses a 10 year running mean and ends in 2006, we discussed “hide the decline,” but I honestly can’t recall if Rose or I said it first. I agree that the way the data is presented in the graph “hides the decline.” There is NO comparison of this situation to Climategate. Muller et al. have been very transparent in their methods and in making their data publicly available, which is highly commendable.My most important statement IMO is this: ‘To say that there is detracts from the credibility of the data, which is very unfortunate.’ My main point was that this is a very good data set, the best we currently have available for land surface temperatures. To me, this should have been the big story: a new comprehensive data set, put together by a team of physicists and statisticians with private funds. Showing preliminary results is of course fine, but overselling them at this point was a mistake IMO.
It seems to me that there are only two possibilities. Either Judith Curry hasn’t bothered to analyze the data from her own team — or she’s not competent to.
Just looking at the data isn’t enough. It’s all too easy to view some graph and get the wrong impression — in fact that what groups like GWPF exploit. Here, for instance, is the graph featured by GWPF:
The impression it gives, that global warming has stopped, is wrong. It’s the kind of easy misinterpretation based on the absence of both context and analysis which is constantly exploited by fake skeptics. I expect GWPF to disdain both context and analysis. But Judith Curry is supposed to be an actual scientist. I’m skeptical.
Since Judith Curry seems unwilling or unable to do the analysis, we’ll just have to do it for her. Let’s start with the data used by GWPF. Here it is with a trend line:
Please note that it starts in January 2001 and ends in May 2010, so it covers a tiny bit less than a decade! That’s a great big “red flag.”
The slope of the trend line (from ordinary least squares) is only 0.03 deg.C/decade. It’s fair to call that “flat.” But the standard error from that calculation — even if we use a white-noise model — is 0.13 deg.C/decade. So the real value could be as high as 0.29 deg.C/decade — which, curiously enough, is a tiny bit higher than the trend rate in the Berkeley data from 1975 to the present — a period for which you can only deny warming if you’re both blind and insane. And yes, it could easily be an even higher slope since we’ve used a white-noise model, which underestimates the uncertainty. Hey Judith: there’s a scientific basis for you.
But that’s not all! The Berkeley data also include uncertainty levels for each monthly value. Let’s graph the uncertainties:
Whoa!!! What’s that huge spike at the end? It rises faster than a “hockey stick”!
The huge spike is the huge uncertainty level of the final two data points — April and May 2010. While the other data have uncertainty levels around 0.1 deg.C, those two months have uncertainty levels of 2.8 and 2.9 deg.C. Which makes them, plainly, unreliable.
And notice that one of those too-uncertain-to-be-useful data points is the deep “dip” at the end, the April 2010 value which is about 2 deg.C below the trend line. It’s called an “outlier,” in fact it’s an outlier of the first magnitude, a really big super-giant “this ain’t right” staring you in the face. And since it’s a super-giant dip it’s gonna skew the trend estimate low. The simplest correct thing to do is to omit those two data points.
So let’s do it again, but end at March 2010 rather than include the uber-fuzzy April and May 2010 data. Here’s the trend line from 2001:
Now it’s noticeably upward. The estimated slope is 0.14 deg.C/decade — more than four times as large, just from removing two errant data points. Its standard error is 0.11 degC/decade, so the real trend rate could be as high as 0.36 deg.C/decade, quite a bit larger than the average rate since 1975. And yes, it could easily be an even higher slope since we’ve used a white-noise model, which underestimates the uncertainty. Hey Judith: there’s another scientific basis for you.
To show just how sensitive the results are to such a short time span — and how misleading it can be to make pronouncements on that basis — let’s consider a time span just one year longer, starting in 2000 rather than 2001. That gives this:
Now the trend rate is up to 0.27 deg.C/decade. That’s just about the same as the trend rate from 1975! Hey Judith: there’s another scientific basis for you.
As a last bit of analysis, let’s omit the outlier data, then compute the trend for each starting year to the present, for all start years from 1975 to 2005. Let’s also estimate the confidence intervals using a more realistic error model. What would that show? This:
The red dashed line shows the trend rate from 1975. Note that not one single start year gives an estimate which contradicts that rate. That is evidence — damn strong evidence in fact — that the underlying trend rate has not changed since 1975. Hey Judith: there’s a REAL scientific basis for you.
Here’s the kicker: the uncertainty in those trend rates is probably higher, perhaps by a substantial amount, because that graph is based on an AR(1) model for the noise. Using an ARMA(1,1) model instead gives this:
That shows just how mistaken, how foolish, how downright boneheaded it is to say that “There is no scientific basis for saying that warming hasn’t stopped.”
Judith Curry’s statement is exactly the kind of ill-thought-out or not-at-all-thought-out rambling which is an embarrassment to her, and an embarrassment to science itself. To spew this kind of absolute nonsense is shameful. Judith Curry, you should be ashamed of yourself.
Richard Muller, the target of Curry’s ire, also managed to embarrass himself:
However, he admitted it was true that the BEST data suggested that world temperatures have not risen for about 13 years. But in his view, this might not be ‘statistically significant,’ although, he added, it was equally possible that it was – a statement which left other scientists mystified.
Of course we’re mystified. Didn’t you do the analysis? Don’t you know? Honestly, what kind of insane world do we live in when we get this level of discussion about one of the most crucial issues of our time, from people who are supposed to be scientists?
In my opinion, the WORST aspect of this is that Judith Curry states unequivocally that “Our data show the pause,” which utterly ignores the extreme level of uncertainty in temperature trends over periods as short as 10 years or less, after she has so often indulged in self-righteous posturing on her own blog about the “uncertainty monster.” It looks like the “uncertainty monster” decided to turn around and bite her on the ass.
Regarding Curry’s involvement with the press release, and the submission of papers, the Daily Mail reports that “Prof. Muller failed to consult her before deciding to put them on the internet earlier this month, when the peer review process had barely started, and to issue a detailed press release at the same time.” She stated, “It would have been smart to consult me.” On her blog she says:
My continued collaboration on this project will be discussed this week with Muller and Rohde.
It seems to me that what would really be smart is to find someone else to be on the Berkeley team.
UPDATE: Judith Curry protests that she was misrepresented by the article in the Daily Mail, and several readers have mentioned that David Rose, the author of the article, is just the man to do such a thing. It’s easy to believe that she was indeed the victim of his malfeasance.
But even after reading this post, she still hasn’t disavowed the statement “There is no scientific basis for saying that warming hasn’t stopped.” In fact she commented on her own blog saying, “There has been a lag/slowdown/whatever you want to call it in the rate of temperature increase since 1998.” Question for Curry: What’s your scientific basis for this claim?
No comments:
Post a Comment