Saturday, January 02, 2010

Questionable Code

My friend Luzr at Talk Polywell has analyzed a bit of the code found in the CRU data dump. He has come to the conclusion that

Given this bug, "anomaly .txt files" contain mostly random data....
That is pretty bad.

However, I think an analysis of the situation is in order. Dale Amon and the commenters on his post at Samizdata have done that. Let us start with a few words from Dale.
A second facet of 'Climategate' is the reported shortcomings in the model code base. Part of the document release included source code. In a discussion with Rand Simberg over breakfast in LA earlier this month I heard that some very knowledgeable open source programmers are having a go at it. If half of what he told me turns out to be true, the models used by IPCC are worse than useless.

John Costella recounts one of the e-mails between ClimateGate "scientists".
Phil Jones fowards it to Mike Mann:
I got this email from McIntyre a few days ago. As far as I’m concerned he has the data—sent ages ago. I’l tell him this, but that’s all—no computer program. If I can find the program, it is likely to be hundreds of lines of undocumented FORTRAN!
Any computer programmer would know that FORTRAN—a computer language so old that its name is spelt in uppercase, because computers did not have lowercase letters back then—is very efficient at performing mathematical calculations, but very obscure to understand if extensive documentation is not provided throughout the program, and very easy to make mistakes in if the program is not well-structured and well-documented.

So we now know that the Climatic Research Unit had no policies covering the checking of results, data archiving, or anything to control the writing and archiving of computer programs!
No wonder they are unwilling to release the code.

Back to Samizdata. An MMGW (Man Made Global Warming) proponent has this to say:
The reality is that nothing of any substance against the AGW theory has emerged from those emails. Even if you ignore all the CRU work, it changes nothing at all. Virtually every other piece of independent data and work supports the broad conclusions of CRU.
Another comment makes a similar point.
Incidentally the IPCC uses other data and models in addition to CRU, for example from the NASA Goddard Institute. The NASA data and source code is 100% open source and agrees well with the CRU results.
To which there was an excellent reply:
...if other founts of knowledge are consistent with the output of buggy code (as analysed here and elsewhere), what does that say about the quality of the output of those other founts of knowledge?

I'll give two views: (i) those other founts of knowledge are of similar quality of that from the deficient processing by CRU; (ii) we are deficient in knowledge of this 'consistency'.

And if we are deficient in knowledge of consistency, do we really know the relevance (if any, and I have strong doubts about it) of average temperature of the planet?
And there is more:
If you really want a specific example of scientific fraud, then I've already alluded to the Wang case. In order to show the urban heat island effect was trivial, the IPCC case relies on a paper by Jones, which relies on a paper by Wang, both of which rely on the assertion that a set of weather stations in China were selected on the basis that we knew they had "few, if any" station moves or other inhomogeneities. However, we know that this cannot be true, and that the researchers involved must have known it, because it is reported elsewhere that most of the stations have no metadata, and most of those that those that do were moved often very considerable distances.

There is no doubt that a claim was made and was relied upon that cannot be true. The person making the statement could not have seen evidence of it. So far, that might be a simple error. Where it becomes outright fraud comes after it was pointed out to them, when instead of simply responding with a retraction and correction, they first ignored the requests for information (because they were from "sceptics"), obstructed attempts to examine data, and denied there was any problem. A complaint of scientific malpractice was made and investigated, but in defiance of their own procedures the inquiry was held in secret, the accuser not permitted to be involved, or to see the report. There is no public explanation or justification. No defence has been presented. Nothing.

Until now, when we see the climategate climate scientists casting around for possible defences, and coming up with nothing. Kevin Trenberth offers "So my feeble suggestion is to indeed cast aspersions on their motives and throw in some counter rhetoric. Labeling them as lazy with nothng better to do seems like a good thing to do." Tom Wigley, Director of CRU at the time explicitly says that the accusation was correct. "Seems to me that Keenan has a valid point. The statements in the papers that he quotes seem to be incorrect statements, and that someone (WCW [Wang] at the very least) must have known at the time that they were incorrect."

That's a direct admission by a Director of CRU and respected climate scientist that the accusation was correct regarding a claim then under investigation as a matter of scientific fraud.
My understanding is that the Wang paper was subsequently withdrawn without explanation.

So what was done about the Urban Heat Island Effect (UHI) in the data? Climate Skeptic has a few words to say (from 2008).
However, many GISS adjustments for site location and urbanization are negative, meaning urbanization has been reduced at the location since 1900, certainly an odd proposition. In fact, if memory serves, the total net adjustment of all stations in the GISS system is negative for site location and urbanization. I know, from here, the net USHCN adjustment for combined site location and urbanization is negative, adding 0.15F to current temperatures as compared to those in 1900, implying that site location quality has improved over time.
So building a city (a heat source) around a measuring site causes temperatures to decline, thus requiring an addition to correct for the decline? It does tend to strain credulity.

All that is bad enough. But there is worse. The climate models fail on their own terms. The Fatal Missing Atmospheric Hotspot of CO2 Alarmist Theory
Dr. David Evans has written a very good summary of the fatal problem of the climate computer models that claim to provide a good match to the increasing temperatures of the late 20th Century based on atmospheric CO2 concentrations. What the modelers have done is taken a theory of greenhouse gases and added theories of sulfate aerosols, ozone, volcanic aerosols, and solar irradiance to plot the expected temperature changes as a function of altitude in the atmosphere and as a function of latitude over the Earth. In reality, they have tuned the many variables and theories to produce results which they claim are predictive of the land surface temperatures for the period from 1958 to 1999. If these theories are valid, then the predictions of temperatures at various altitudes and latitudes must also be correct.
A series of graphs is then presented showing the model results.
...if the factors modeled are the most important factors pertaining to climate change, such features as the very prominent hot spot near and above the equator at altitudes of 8 to 12 km with elevated temperatures of 1.0 to 1.2C should be measured by balloons equipped with transmitters to send back temperatures measured as the balloon ascends to high altitudes in the atmosphere. There are good records going back to the 1960s with hundreds of balloon flights having the ability to measure temperatures with an accuracy of better than 0.1C.
The graphs of the measurements are shown. Then he goes on to say:
The lower atmosphere, or the troposphere, is relatively uniformly warmed compared to the Santer model results used in the UN IPCC AR4 report of 2007. Note that the entire range of variation in the plot with altitude and latitude is much less than in the alarmist model plot. The cooling of the upper atmosphere, the stratosphere, is also less severe and is also more uniform. The hottest warm areas are mostly in the northern hemisphere and only reach up to about 0.3C warmings, not the 1.2C of the models touted as good matches to reality by the UN IPCC. Also, very notably, the major hot spot over the equator and latitudes nearby is missing. This area is largely warmed only about 0.1C and parts of this area are cooled by -0.1C! The warming is at least an order of magnitude less than in the alarmist model at 10 km altitude over the equator!

So, the UN IPCC model result, which is said by them to verify the theory of catastrophic anthropogenic global warming predicts much greater temperature changes due to their greenhouse gas theory in the atmosphere than is observed and the pattern of those changes is very different than that actually measured.
The normal conclusion when models don't match data is that there is something wrong with the models. Evidently the rules for climate science are different. Fortunately the Rulers of Climate Science appear to have taken a fall. And if temperatures keep going the way they have been going the last few weeks they may also be taking a very hard winter.


Silverfiddle said...

Yes, pay no attention to that man behind the curtain!

Everybody's talking about the e-mails, but to cadge from Shakespeare, the code's the thing.

I don't mean to blogwhore, but I put together a collection of links to this specific issue:

The piece by Charlie Martin is especially illuminating and entertaining

M. Simon said...

Charlie Martin is a friend and a code geek. Thanks.

And if you are going to blogwhore make the links clickable. It shows professionalism.

Western Hero - fudge

Pastorius said...

Great summary, M. Simon. Thanks.

M. Simon said...


Greatly honored to be of service.