This is a repost of two articles from John Graham-Cumming’s blog. I watched with interest earlier this month where he and a colleague identified what they thought to be a math error related to error calculation when applied to grid cells. It appears now through a journalistic backchannel that the Met Office is taking the issue seriously.
What I found most interesting is that while the error he found may lead to slightly less uncertainty, the magnitude of the the uncertainty (especially in homogenization) is quite large in the context of the AGW signal being sought. John asks in his post: “If you see an error in our working please let us know!” I’m sure WUWT readers can weigh in. – Anthony
The station errors in CRUTEM3 and HadCRUT3 are incorrect
I’m told by a BBC journalist that the Met Office has said through their press office that the errors that were pointed out by Ilya Goz and I have been confirmed. The station errors are being incorrectly calculated (almost certainly because of a bug in the software) and that the Met Office is rechecking all the error data.
I haven’t heard directly from the Met Office yet; apparently the Met Office is waiting to write to me when they have rechecked their entire dataset.
The outcome is likely to be a small reduction in the error bars surrounding the temperature trend. The trend itself should stay the same, but the uncertainty about the trend will be slightly less.
===============================================
Something odd in the CRUTEM3 station errors
Out of the blue I got a comment on my blog about CRUTEM3 station errors. The commenter wanted to know if I’d tried to verify them: I said I hadn’t since not all the underlying data for CRUTEM3 had been released. The commenter (who I now know to be someone called Ilya Goz) correctly pointed out that although a subset had been released, for some years and some locations on the globe that subset was in fact the entire set of data and so the errors could be checked.
Ilya went on to say that he was having a hard time reproducing the Met Office’s numbers. I encouraged him to write a blog post with an example. He did that (and it looks like he had to create a blog to do it). Sitting in the departures lounge at SFO I read through his blog post and Brohan et al.. Ilya’s reasoning seemed sound, his example was clear and I checked his underlying data against that given by the Met Office.
The trouble was Ilya’s numbers didn’t match the Met Office’s. And his numbers weren’t off by a constant factor or constant difference. They followed a similar pattern to the Met Office’s, but they were not correct. At first I assumed Ilya was wrong and so I checked and double checked has calculations. His calculations looked right; the Met Office numbers looked wrong.
Then I wrote out the mathematics from the Brohan et al. paper and looked for where the error could be. And I found the source. I quickly emailed Ilya and boarded the plane to dream of CRUTEM and HadCRUT as I tried to sleep upright.
Read the details at JGC’s blog: Something odd in the CRUTEM3 station errors
DirkH (08:12:42) :
First off, Tamino will release his code when he publishes.
Second, it’s all a red herring. The ccc guys released their code, to the same end. Now what, Dirk? You’ve got the code.
Then, this is indeed how science works. I publish a paper, saying what I did. I should give enough of a description for somebody to recreate and get consistent results, if they were curious. I’ll cite other previous papers as appropriate.
So researcher B is interested, and tries the same thing. If it’s experimental, he’ll do his own experiments. If it’s mathematical processing like this, he’ll write his own code. If he gets consistent results, great. Maybe he’ll go on to improve on my method. If not, then he looks to see if he made a mistake. Then re-reads my paper to make sure he didn’t miss something. Maybe it becomes clear that I made a mistake. If he still can’t figure out the reason for the divergence, then he might send me an email. Maybe it gets to the point where we compare code. It’ll get sorted out in the end.
But the point you’re missing is that what Tamino did is quite simple, assuming you have a little bit of background information. If you think he maybe messed up, then try it for yourself. If you don’t have the background information, then read 3-4 papers and the documentation for GHCN, and you’ll be set. You didn’t have to wait for Tamino to do this; you could have done it yourself before he even started it.
And the critical point goes back to the original SPPI document. It says:
““Calculating the average temperatures this way would ensure that the mean global surface temperature for each month and year would show a false-positive temperature anomaly – a bogus warming.””
What is the basis for this statement? All I see is a bunch of absolute temperatures averaged together. (Please note that I’m not asking for the code anybody used to do that; I can average numbers together for myself). But a bunch of absolute temperatures just averaged together is absolutely meaningless. What you need is something like what Tamino, or the ccc guys, did. Where is it?
carrot eater (08:40:24) :
Oh dear, wrong again “The ccc guys released their code”, they may have but it is no longer available as they are conveniently rewriting it in Python. The links to the Code lead nowhere.
Plus of course they are using the totally corrupted GISS data, or haven’t you been reading the work done by other countries using their original data to show that GISS representing their temperatures are completely wrong?
See the work by Finland, Russia, Australia and New Zealand, plus even the UK is having to revew their own data and calling for a complete re-appraisal of Temperature data.
A C Osborn (10:48:44) :
What are you talking about. The ccc code for splitting between dropped and not-dropped stations is right there. The current release of ccc-gistemp is up there. The link leads right to what you want, if you want to test this on ccc-gistemp. It wouldn’t be my choice, but it’s all right there.
Tamino used the GHCN raw. No adjustments. Nothing ‘corrupted’. Just saying things doesn’t make them so. The only adjustment GISS does is a crude one for UHI, and that has nothing to do with the station drop issue.
Again, you don’t have to copy Tamino’s code to do this. Did Tamino have to copy anybody’s code? Try writing something for yourself. If you doubt Tamino’s or the ccc’s result, that’s what you should do.
carrot eater (11:37:02) : please provide a link to the Code, where it is visible as the link that they provide does take you to the code http://code.google.com/p/ccc-gistemp/.
CCC used GISS, I didn’t say Tamino used it.
A C Osborn (12:32:33) :
GISS starts with GHCN raw data for everywhere except the US lower 48. The only adjustment GISS makes is the UHI one. But in any case, what’s being tested here is whether the stations that were dropped in 1990 were somehow different, such that dropping them gave a warming bias.
Did you not follow the link above? The description is here,
http://clearclimatecode.org/the-1990s-station-dropout-does-not-have-a-warming-effect/
the code to split the input is here
http://code.google.com/p/ccc-gistemp/source/browse/trunk/tool/v2split.py
and the latest version of ccc-gistemp is here
http://code.google.com/p/ccc-gistemp/downloads/list
Even the page you linked has links to the download pages. I don’t use ccc, so any questions specific to ccc, direct to them, not me.
@Mr. Eater…. of Carrots
“What equations do possibly you want?”
I’ve been thinking a lot and hard, mind on this maybe simple question. To me it isn’t really a simple question, since I, like I imagine most teachers want, actually want to se how the simpleton, er pupil, solved the problem.
But by all means, believe what ever people tell you to believe. After all you are, I hope, free enough to believe what ever you want to believe in. Personally I rather want to know the details of what’s what.
With that said, can you produce the equations the little flute used? That you can’t produce the actual implementation is obvious, but maybe you do know all the equations used? How about which equation he used in the selection process? Do you actually know which stations he used and which he disregarded?
“Everything else he did, you can figure out for yourself.”
For all the special person I am, I’m not supposed to need to figure everything out for myself. If some fool claims he owns the moon he better be able to f*cking prove it.
1DandyTroll (17:53:25) :
Sit and wait for his code then. In the meantime, you can read the ccc’s code, which of course uses the existing ccc-gistemp code. Or Zeke Hausfather’s; he’s done his own version now, with the code available. I don’t care. The conclusion on this one isn’t going to change with minor changes in implementation.
Forget Tamino. Suppose you had just read the SPPI report. Did that prove its claim? How would you approach the question of seeing whether the station drop claim had any merit?
carrot eater (18:38:45):
“Forget Tamino. Suppose you had just read the SPPI report. Did that prove its claim?”
Earth to carrot eater: Yes, we would like to forget that tamino, who hides his hidden methods, is exempt from scrutiny… if you ever stopped being his publicity agent.
And keep in mind that skeptics have nothing to prove. The promoters of any new hypothesis have the burden of showing that it explains reality better than the long held theory of natural climate variability. So far, CO2=CAGW fails; natural variability prevails.
carrot eater (14:02:30) :
The whole point is that by using GISS the damage is already done.
Raw values must be used otherwise the “FIX” is already in, lowering old readings, raising new ones, adding in missing data points or deleting data points to suit the required “Trend”.
You obviously do not believe all the evidence that other Scientists around the world have been submitting or you would not gloss over starting the analysis process with garbage.
Have you read the Finnish and Russion Reports?
Do you believe them?
carrot eater (14:02:30) :
Thank you for finding the CCC Code for me.
So to summarize:
GISS adds 2 +2, gets 5.
Tamino adds 2 + 2, gets 5.
carrot eater says Tamino used a different method, got the same result, therefore 2 + 2 equals 5 is confirmed. And since it was the same result as GISS, Tamino’s work is confirmed. You don’t have to know exactly how Tamino got that result, take a month yourself and see if you too can get 2 + 2 to equal 5. For further confirmation, the Clear Climate Code people followed the method used for GISS and their code also gets 2 + 2 to equal 5.
There is a consensus, the science is settled, the debate is over. 2+2=5 and that’s that.
Smokey (19:50:22) :
Like I said, if you really want code, there are now two groups who’ve put theirs out, for this exact problem. or you could wait for tamino’s. Or you could try to work it out for yourself. I don’t care, but just complaining about not having one person’s code is going to get you anywhere.
“And keep in mind that skeptics have nothing to prove. ”
Really? They can just say anything they want, without having anything to back it up? A specific claim was made. They don’t have to prove it? Interesting.
kadaka (06:07:30) :
You think everybody’s getting a wrong result, just because you don’t like the result?
The question here is, were the stations that were dropped circa 1990 somehow different, such that simply dropping would cause a spurious warming?
You are convinced that this statement is true? Why? What have you seen, to make you think that?
A C Osborn (05:18:55) :
GISS starts with raw data for everywhere except the US. Do you realise that? You can then study the UHI adjustment they make.
carrot eater (06:39:23) :
You think everybody’s getting a wrong result, just because you don’t like the result?
Not at all.
Repeating an error is not confirmation there is no error.
Besides, “everybody” is not getting the same result anyway.
And, if the original data is bad, then the result and the methods used to get the result does not matter anyway.
That’s that. Have a nice day!
Curious.
Yet earlier in the thread:
So, is carrot eater saying that Zeke Hausfather has now done something, as of the 27th of February, that Zeke himself said he could not do back on the 25th?
Wow, that Zeke fellow sure can learn and code in a hurry!
Anyone got a link to Mr. Hausfather’s original code?
Interesting.
Hansen releases Method A (Hansen and Lebedeff 1987) with details.
GISS uses Method A, releases “GISTemp” code.
Clear Climate Code uses Method A, releases ccc-gistemp code (not yet finished).
Tamino uses Method B, does not release Method B with details, does not release code.
Proposed: It does not get you anywhere to complain about not having code that uses Method B (nor about not even knowing what Method B is), as you already have two sources of Method A code, and you could always make up your own Method C and your own code for that.
Meanwhile, just sit back and accept that Method B yields valid results according to its code.
O-kay… Moving along…
The more this goes on, the more it seems the Hockey team simply wasn’t very good at the science, AND that they knew it enough about their inadequacies to be insecure about it.
Back in the 1960s there was a book out named The Peter Principle. The Peter Principle stated as its observation that people are promoted to their level of incompetence.
I don’t feel sorry for these losers, because they were such blow hards, and who can feel sorry for jaggoffs? It is such a good thing that – inch by inch – the adults are beginning to have their say in this.
What will come of it all?
The IPCC will be eventually disbanded, due to the numbers underlying all their arguments are, piece by piece, being shown to be just flat wrong. The scientists were so in love with being the center of attention (remember that before 1988 climatology was just a scientific backwater) that they fudged – consciously or unconsciously (and does it matter which?) – to keep the limelight on themselves and the free trips and treatment as royalty.
In their insecurity the rats were not about to be cornered, they ad hominem attacked anyone who threatened to put them in that corner.
GOD BLESS WHOEVER LEAKED THE EMAILS, for the rats are being shown for what they are – not very good at statistics and programming.
WHY they didn’t have a statistician on staff is beyond me. No one was there to correct them? How does that happen?
And why not true programmers? Some need to keep it all close to the vest? That is exactly what people do who are insecure about their work and their status – keep everyone else from seeing them for the frauds that they feel they are.
@A C Osborn: I am genuinely concerned. What could we, Clear Climate Code, do to make finding the code easier?
Various people have asserted that ccc-gistemp is not finished. This is true in the same sense that, say, Microsoft Windows, or Firefox, is not finished. Since our goal is code clarity, and there are ways in which our code is not perfectly clear, the project is not finished. However, we have published releases which consist entirely of our code, which reimplement the GISTEMP algorithm entirely, and which match the GISTEMP results.
The most recent release is 0.3.0. Clarification work since 0.3.0 can be seen here. This ongoing work includes changes to eliminate all intermediate rounding and truncation of data, and to expose and document all numerical parameters in a separate module. There will be a release 0.4.0 at some point in the next couple of weeks, to package up these changes.
Nick Barnes (04:35:40) :
Various people have asserted that ccc-gistemp is not finished. This is true in the same sense that, say, Microsoft Windows, or Firefox, is not finished. (…)
Both of which use release numbers starting with an integer greater than zero. Thus it has not been “officially” signaled that ccc-gistemp is a finished product, which has been done for both Windows and Firefox. They have entered into the normal cycle of ongoing revisions and modifications as needed, there are versions you can point to and say they are finished products, such as Win 95. This has not yet happened with ccc-gistemp. Thus ccc-gistemp is not finished in the same sense that Windows or Firefox is not finished.
Re: kadaka (09:22:23) :
Should be “Thus ccc-gistemp is not not finished in the same sense that Windows or Firefox is not finished.”
But, heck with it, time to stop abusing the English language. ccc-gistemp does not have a finished version, Windows and Firefox have finished versions. ‘Nuff said.
kadaka: Both of which use release numbers starting with an integer greater than zero. Thus it has not been “officially” signaled that ccc-gistemp is a finished product, which has been done for both Windows and Firefox.
Don’t pay too much attention to release numbers. Years ago I worked on a project which used the names of fish for releases (Whitebait, Halibut, and so on), which avoided all this reading of release number tea-leaves. I run the project, and I can tell you “officially” that release 0.3.0 is more finished than a lot of software ever gets. It is being continually improved – 0.4.0 will be quite a bit better than 0.3.0; I have a notion of what release 1.0.0 will be like, and a vaguer notion of when it might be produced – but as it stands it is already pretty good, and it is certainly “finished” (in the sense that it is a complete implementation of the GISTEMP algorithm in freshly-written Python).
If you really want to understand the GISTEMP algorithm, I recommend that you download ccc-gistemp 0.3.0 (or the current sources, or wait for 0.4.0). If you look at that code and find it is unclear, that is a bug, and we would welcome your bug report.
You do really want to understand the GISTEMP algorithm, right?
Smokey (19:50:22) :
And keep in mind that skeptics have nothing to prove. The promoters of any new hypothesis have the burden
Uh oh, Anthony, one of your readers is saying you have the burden of proving that “instrumental temperature data for the pre-satellite era (1850-1980) have been so widely, systematically, and unidirectionally tampered with that it cannot be credibly asserted there has been any significant “global warming” in the 20th century.”
Smokey says you better get busy on proving that.
kadaka (08:42:28) :
Anyone got a link to Mr. Hausfather’s original code?
http://drop.io/0yhqyon
Wow, that Zeke fellow sure can learn and code in a hurry!
He said it was “a bit” beyond his programming ability, but he also said it wasn’t “that difficult.” It’s an example of what you can do if you have some skill and a desire to learn.
More about his method and his results here:
http://rankexploits.com/musings/2010/a-simple-model-for-spatially-weighted-temp-analysis/