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
Make that Scopes, not Stokes.
Sigh, ….
sturat (20:15:43),
I’ve read carrot eater’s denigration of skeptics, using insulting terms like “denialist” on alarmist blogs, so quit wasting your time apologizing for him.
Instead, let’s see those ‘robust’ analyses by tammy. No need to have secrets about climate data and algorithms; they’re not nuclear defense secrets.
The fact that tammy hides his methodologies indicates that he’s all bluster. All show and no go. All hat and no cattle. That dog won’t hunt.
That’s what we’ve come to expect from tammy’s closed mind.
Can anyone explain how temperatures are claimed to be accurate to +/- 0.2 degrees in the 1930’s, when electronic thermistors hadn’t been invented ?
Even later temperatures measured electronically to 0.1 degrees but recorded to only +/- .5 degrees
http://www.srh.noaa.gov/ohx/dad/coop/EQUIPMENT.pdf page 11
Lots of other errors on top of this, as has been much discussed.
Early temperatures measurements were a bit of a joke:
http://climate.umn.edu/doc/twin_cities/Ft%20snelling/1850sum.htm
Can anyone -ANYONE- say what the temperature anomaly would have been if CO2 had remained the same?
If you have to delete my other comment to hold to policy, I understand.
Looking at the graph that started this thread (not involving Tammyno) it looks like the warming since 1850 could be as much as 1C or as little as ~.2C? Am I really supposed to be quaking in my boots here? The dart board graphic Anthony uses has a whole new meaning, imho.
Malcolm (16:57:12) :
I don’t believe that for 1850, 1950, or even 2009. I wouldn’t even give anyone decimals.
I don’t believe anyone who thinks they can calculate the temperature of the entire planet using air temperature and sea temperature over the entire planet, using spatially non uniform measuring, introducing adjustments to the data using an algorithm, using adjustments that have dubious necessity, all the while retracting statements in the ‘most comprehensive’ scientific document supporting their position. So they lose me as soon as they say ‘point…’ I am just not buying it. My BS-dar goes off every time.
Open question:
Why even consider the minimum temperature if you are looking for a warming signal? Why not look at only maximum (since we are supposed to fry)? Or, better yet, why not look at max and min separately instead of smashing them together? Or, even better, why use averages? Temperature seems like one of those quantities where the better statistic to study would be median.
sturat (20:04:38) : But, it seems you still had time to post several rambling, little value added posts that did little to add to the discussion.
Your POV. I was responding to what folks asked. That you don’t like my answers is not very valuable.
I posted a clear link to the analysis that shows New Zealand has no warming trend without a specific station and a warming trend with it. A station that is dropped in GHCN half way through so it cools the past. I see no need to re-do already done work. It is one OF MANY existence proofs that the thermometer changes are toward warming bias of the data set.
I’ve also shown a comparison with a method for finding a (more or less) clear actual trend and showen that the dT/dt method results in an answer roughly in sync with the “non-biased constant station” set for New Zealand in the link (since that was not already up at my site).
Further, I’ve pointed out there there is a GIStemp benchmark up on the site (said link having been posted here too many times already and easily found under the GIStemp tab on my site. Hardly hiding anything nor putting anything where it can’t be found.
So you see, I’ve not only shown the data have a warming bias, but I’ve shown that the GIStemp technique does not correctly handle it while another technique can.
Further, I gave my evaluation of what Tamino did based on a cursory examination. That you don’t like my evaluation is of slightly less interest to me than was actually making the evaluation. I stated specific areas where I saw “issues” so other folks could explore them if they wished.
Now what is probably lost on you is the simple fact that Tamino might well have shown that some particular anomaly processing method could keep the trends of both data sets similar. Just like I showed that GIStemp fails to block all warming signal and another technique does pretty good.
NONE of those means the data were not biased by change. They only show that the process performed on that data has better or worse handling of that bias. To see the bias in the data does NOT require looking at Tamino’s stuff, it requires looking at the data. And in addition to the specific example with link given, there are dozens of others on my blog.
Thermometer change clearly biases the data toward a warming profile.
The only question is how well do various methods handle it.
Then you want to insult me by saying that my responses were not up to some hypothetical imaginary standard you have internalized? And folks wonder why I have no interest in playing these “You must prove FOO is wrong” games….
if we would only trust you.
I said nothing about trust. I said the existence proofs are up for inspection. Go inspect.
And, don’t bother you about asking for details since you are above question.
You would have me retype all the details that are already in postings you are too lazy to go read? Yeah, don’t ASK me for details that are already up in written form.
Please, give us all a break.
Happy too. Didn’t want this time sink to begin with. So your wish is granted. I’ll take a break from this complete thread hijack and pointless troll-fest until further notice.
All I’ve seen on your site and this one is trust me,
As I’ve pointed out above. I’ve never said “trust me”. I publish my code as soon as it’s QA’d and stable and every thing I do is described to the point where anyone can reproduce it (and many have). That you don’t see that speaks volumes.
I’m too busy, I’m going to publish it someday,
The first is true. The other is bent. I’m “looking into perhaps publishing” one feature which I’m in active discussions on. Sorry you don’t like that, but it’s fairly standard practice as I understand it. If it’s ‘not new’ it gets shut down. So tough.
But that portion is not needed to demonstrate there is warming bias introduced into the data by thermometer dropping. I’ve got 2 selected entry point examples for you AND several dozen other very detailed postings about the specific kinds of bias. Go read them. Start with the “GIStemp” tab at the top and pay particular attention to the “GHCN Global Analysis”.
What happens to that bias in processing in the various software packages for the various data series is up to debate. That the data is ‘warmed” is not.
other people stole my data, and …
Now you are just making up flat out lies.
I have never said anyone “stole my data”. First off, I don’t have any data. It’s all NCDC’s data. I just process it in interesting ways. Secondly, my code is put up under open software rules for folks to do with what they will. I have nothing proprietary for anyone to “steal”.
Sorry, but that puts you in flat out Troll land.
What I have asked for I believe is a reasonable request.
Sink a few days of my life into working out what someone else did wrong and showing how they can be proven wrong when they have chosen a broken method to begin with, and you expect me to use that method. Yeah, that’s real reasonable /sarcoff>
I’ll do it right after you prove that my zero point energy machine idea doesn’t work. It uses cold fusion and palladium, and you have to use them in your proof. (Do you see how impossible that is and how it can be an infinite time sink?)
Show a competing analysis on a comparable set of data that contradicts Tamino’s conclusion. I know Tamino took a couple of weeks to produce his results and that he continues to expand and refine the work.
So you “only” want to consume a “couple of weeks” of my life (or perhaps more since I’d have to reverse engineer whatever he dreamed up). And that’s “reasonable”…
It would be civil to say that you are “interested”
No. It would be a flat out lie. And I don’t do that. I have NO interest what so ever (as you would have known if you hard bothered to read my prior statements, but you seem to have decided they had nothing to say…)
where is the reproducible analysis that shows that Tamino is wrong.
Prove a negative. Uh Huh… Look. I stand by what I do. I don’t play negative proof games.
Oh, please don’t drag in the “court of law” meme. It really has not applicability here. Besides Mr. Stokes was convicted by a court of
I have no idea were this is coming from/ “Court of law”? I didn’t say anything about courts or law. And I’ve got no idea who this Mr. Stokes is that your dragging in.
At this point, you seem to have gone completely off the deep end.
So here is my suggestion. Look at the cases where I’ve demonstrated that the thermometer changes over time move station data from cold places to warmer places. Then consider that GIStemp compares ‘basket A in early times” to “basket B in later times” (as already described above) and that there is a benchmark run on NCDC data using GISS code that shows this bias comes through the code.
Then ponder the meaning of documented reproducible existence proof.
And yes, it’s already published on the site. And it won’t even take 2 weeks of your life for you to find it. And no, I’m not going to spoon feed it too you. I’m going to “give you a break”.
JGC & Ilya Goz ,
Wonderful effort.
This case of correcting MET Office to show they had less error in their grid cells helps show the balance of the blogosphere.
Also, your case show the profound capability of the blogosphere in real time.
John
Should I wear 3-D glasses to figure out what the graph shows ?
Well , at http://www.hidethedecline.eu a tiny tiny problem with the CRU Brohan 2006 data was reported..
http://hidethedecline.eu/pages/posts/temperature-corrections-of-the-northern-hemisphere-144.php
Its just the Northern Hemisphere area ..
Smokey (21:04:32) :
“The fact that tammy hides his methodologies indicates that he’s all bluster. All show and no go. All hat and no cattle. That dog won’t hunt. ”
Nothing is hidden. He’s described everything he did in pretty close detail. Anybody could program what he did, for themselves. Just as anybody could pretty well emulate what GISS does, for themselves, just by reading their papers. Which is what Tamino did, except he made a few changes, which he specifically mentions.
This is half the point of the exercise – you can do this stuff for yourself. Tamino did it for himself. It took him a month, presumably working on it in his spare time, but he did it. So it’s doable, even if it isn’t your day job. So if you make a claim, like those in the SPPI report, then you should also put the work in, and find out if the claim has any merit. Without doing the analysis, it’s a little empty.
E.M. Smith,
I’m curious about your use of existence proofs. Such a logical proof would seem to be applicable to the question “Can a warming bias exist if a reporting station is dropped from a data set” but not to the question of whether such a bias does, in actuality, exist. The question at hand is not a logicl one, but a statistical one.
He shows that two sets of over averaged data match in one period of time then asserts this means they must match in another period of time (when one of them is missing).
“Assertion” is, again, a term from predicate logic. The data presented by Tamino show that the trend up to 1990 without the missing stations is the same as the trend with those stations included. He also showed that NASA’s adjustments did not increase the warming trend. His methods may be flawed, and when/if he makes his results and calculations public they can be evaluated. But it’s wrong to say Tamino merely made an assertion.
I am not sure why you have such a tone of being put-upon, or why the word “demand” so frequently appears in inverted commas. No one has “demanded” anything. The only person who suggested you weigh in was Anthony Watts, and I feel sure that it was a request, not a demand. Perhaps the word “demand” appeared in your personal correspondence. It is nowhere in this thread,except in your comments.
Certainly, those that say Tamino’s analysis was flawed have a responsibility to show how he erred; those, like me and you, who have either not enough time or not enough expertise – or neither – can’t really make any judgment, can we?
Only by publicly posting his complete code & methodology, and anything else requested by skeptical scientists, can Tammy regain credibility.
That’s how the scientific method works. Wake me when s/he provides full and complete transparency, openness and cooperation.
E.M.Smith (22:30:42) : neither cast ye your pearls before trolls, lest they trample them under their feet, and turn again and rend you
Folks, one of the best ways for scientists, mathematicians, etc to demonstrate whether or not a hypothesis has legs (what I call substantiation research as opposed to original research) is to “do it a different way”, as well as repeat it. Both repetition and coming to the hypothesis from another angle are standard methods of developing hypotheses into theories.
Repetition: The standard form is to repeat exactly what was done but on a grander scale, IE with more subjects, soil plots, more sensors, improved sensors, etc. Any mathematician or scientist who has been in the trenches doing research understands this must be done. You can’t skip this test. And it is done best by a fresh set of eyes. To do this, all data and methods must be readily available. If you are not willing to do this, be prepared to have your hypothesis debunked in the second form of substantiation research:
Robustness (IE same hypothesis, different angle): The standard form is to study the problem from a different angle, using different experimental methods and analysis, and discover that the hypothesis is robust and powerful in other areas as well. For example if the CO2 theory not only increases temps, it should also increase water vapor absorption and re-radiation. So analyze water vapor absorption and/or OLR. If the CO2 hypothesis is both powerful and robust, it will affect other areas as well.
In my study, others had done the original research in several labs. But we wanted to find out if high frequency tone bursts were differentiated by the auditory brainstem in waves I as well as waves III and IV (these are labels for major synaptic junctions). We did indeed find this to be true. But just to be sure, the lab repeated the experiment with more subjects. Hypothesis confirmed. It was and is a stable and robust hypothesis. The auditory brainstem response is capable of differentiating high frequencies (stable) at its earliest synaptic junction (robust) as well as at later junctions.
So Tamino, put it all out there. Your research will be vastly improved, as will your standing in the blogosphere science community, if you are willing to put it out there.
Ian Jolliffe.
=====
His bias makes it seem as if he’s wearing blinders, but really, he just sees as through a glass darkly.
==================
Smokey (06:08:21) :
The complete methodology is posted there. When there was something left ambiguous about what he did, I asked and he answered. He says he’ll include his code in the supplementary info of whatever paper he writes, so we’ll see about that then.
But you’re gravely mistaken about the scientific method. If I describe what I did, then that is enough. You don’t need all my spreadsheets and code and scribbled notes to read, understand, verify or build upon my work. It’s generally better if you do all those things for yourself, as there’s then a higher chance of any errors being caught (or, demonstrating that the initial description wasn’t good enough).
As I said – Tamino started from scratch, and was able to build a program that roughly emulates (with some differences) what GISS does. That’s half the point – you can do this yourself, and you don’t need to copy somebody else’s code in order to do so.
The other point is this: The SPPI report 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.” This is a strong statement. Where is the analysis that shows this to be true? Is it in the SPPI report?
Zeke Hausfather (17:38:49) :
Exactly. Unless somebody at the WMO, NCDC or the collection of the national met services knows how to travel time, the original assertion seems more than a little weak.
“carrot eater (05:37:16) :
[…]
Just as anybody could pretty well emulate what GISS does, for themselves, just by reading their papers. ”
Now, maybe carrot eater has never written a program so he just can’t know, but let me assure the readership that it is impossible to replicate the exact behaviour of a program including all its bugs and side effects just from a description of the way it *should* work or the observation of a test run of the original program.
So, carrot eater is wrong here. 100% certified wrong.
I agree DirkH. Confirmation research is what it means. Same data set. Same code. Same analysis. But fresh pair of eyes. And if you are good, fresh interpretation and new and different angles to propose for further study into the robustness of the findings and/or technique. Skip this step at your peril. No drug company would EVER skip this step.
carrot eater (06:35:01) :
“The complete methodology is posted there… [Tammy] says he’ll include his code in the supplementary info of whatever paper he writes, so we’ll see about that then.”
Tammy – and you – can say anything. Either the complete methodology leading to his conclusions is posted, or it’s not.
In this case it’s not.
DirkH (06:37:53) :
Dirk, I can assure you I’ve written more than a couple programs. We’re not talking about replicating every behavior to the 10th digit. That isn’t what reproducibility is about. We’re talking about getting basically the same results. And this is exactly what Tamino has just demonstrated. He started from scratch, and even did a bunch of things differently from what GISS does, and still got the same basic results. If I were him, I’d go back and also write a version more faithful to the GISS methods, but that’s up to him.
This is just how things work in science. If somebody describes what calculations they did, I should be able to go home and try it for myself. Without copying his code. If I get basically the same results, all is well. If I can’t, then I ask a couple questions. Maybe that will uncover that one of us made some error in the programming, or math, or physics, or whatever.
Smokey (07:05:19) :
You can also just say anything. Can you back it up? What major description of his methodology is missing, from his posts over the last month?
If you sit down and implement what he did for yourself, you’ll have some minor differences. But all the important bits are described there. Unless one of you make a basic error, you should be able to get consistent results.
The processing that GISS does is really quite simple, people. You can make your own version for yourself, just as Tamino did (with a few modifications). It won’t be exactly the same, but that isn’t the point. The point is, can you start with the same sort of data, apply the same sorts of processing steps, and get consistent results?
carrot eater, Paul Daniel Ash, sturat E M Smith doesn’t need to do the work.
This Post http://wattsupwiththat.com/2010/02/26/a-new-paper-comparing-ncdc-rural-and-urban-us-surface-temperature-data/
completley contradicts Tamino’s work and I know who I would rather believe.
OK before you say it, it only covers 48 Sites, but the same algorithms are applied everywhere else by the NCDC.
Only by publicly posting his complete code & methodology, and anything else requested by skeptical scientists, can Tammy regain credibility.
That’s how the scientific method works.
Well, he’s said (http://tamino.wordpress.com/2010/02/25/show-and-tell/#comment-39877) he’ll make the code and results available. We’ll have to wait and see before concluding it’s an empty promise.
I don’t know where “anything else requested” fits into the scientific method; that sounds a bit more like the “demands” E.M. Smith finds so offensive.
The central point carrot eater made is a good one, though: there’s nothing stopping anyone who wants to from carrying out an analysis on their own. The question at hand is not Tamino per se, it is the validity of the temperature records. The onus is certainly on those who assert that there is a warming bias to show, not merely assert, that it exists. That is how the scientific method works.