Thursday, September 13, 2007

Climate Cuttings #9

His ecclesiastical eminence has posted up the ninth in his series of Climate Cuttings. There's lot sof interesting stuff there, including the saga of the utter discrediting of NASA climate scientist, James Hansen.
While the (allegedly) real scientists were engaging in ad-hominems, the amateurs at Climate Audit followed up with further revelations of faulty work from Hansen. The latest batch of errors were found when the site started to raise questions about the way that Hansen combines different versions of the temperature record for a particular station. This appeared peculiar because Hansen was combining records and ending up with an average lower than any of the individual temperatures in the series. Because Hansen has not adhered to the basic scientific standards and released his code, it was necessary to derive what he had done by trial and error - guessing the procedure from the limited explanation in his publications. Eventually it was suggested by a commenter that the solution lay in understanding what Hansen did where the temperature for a particular date was missing from one of the versions. If you and I had this problem we would take the temperature from the other version. It was thought, however, that Hansen was "estimating" it somehow. This obviously represents a corruption of the temperature record, but this is climate science where pretty much anything goes.

The code has now been released and a project is now underway to get the code understood and working—made more complicated by the fact that it is written in the somewhat superannuated Fortran language*.
With the code in place a full summary of the way Hansen's methodology works (at least as far as it is currently understood) was posted by John Goetz, the CA commenter who discovered the importance of the missing records. This makes it clear that, while the effect on the trend for the station could be up or down, it appears that more often than not the effect is to lower earlier temperatures - ie to make the warming trend look artificially high.

The latest headline about the integrity (or lack of it) of Hansen's work is the revelation today that, unannounced, he has made large changes to the temperature records for the US. This has happened in the last few weeks - since the Y2K errors were revealed last month. From the outside this might be mistaken for an attempt to get the temperature of recent decades up again.

Either way, it's pretty clear that Hansen's credibility is shot. Can NASA really tolerate this sort of junk science from one of its leading officials any longer?

We shall see.

As for the rest, it is the usual smorgasbord of exciting revelations about how climate scientists are lying like the fucking bastards that they are.
Well, Anthony Watts has now surveyed 33% of the US surface stations and has released preliminary results. Only 13% (yes, you read that correctly) of the network is of a standard suitable for climate monitoring according to the standards set out by CRN - the new high standard network currently being developed.

So, the land temperature record is actually heavily skewed—towards the high end.
A new paper in the Journal of Remote Sensing claims that there is an order of magnitude uncertainty in forecasts of temperature due to our lack of knowledge of clouds.

No surprises there: some of us have been banging on about water vapour being by far the largest contributor to the greenhouse effect.
The BBC cancelled a proposed global warming day, claiming, apparently in all seriousness, that it didn't have a "line" on the issue. Nature Climate Feedback reported that the BBC had commented that the alleged consensus on global warming is "increasingly strong (but not overwhelming)" - a massive downgrading of their previous position of "We're all going to fry!!".

However, as Bishop Hill pointed out a couple of days ago, the Beeb are still not exactly unbiased on the whole topic.
On the BBC's climate change portal at the moment, the main stories include
  • Calls to strengthen the EU emissions trading scheme for airlines

  • Calls to encourage homes to go green

  • A report that the British are addicted to cheap flights

  • A report that the risk of flooding due to climate change has been underestimated

  • A conference to discuss tackling climate change

  • Increases in forest fires due to climate change

  • A report that APEC has muddied the climate change waters

  • A way to track your carbon emissions through your phone

  • A report that winter sports threaten mountain ecosystems

and lastly, and surely with tongue firmly in cheek, an entry from the Editors blog in which Head of TV news, Peter Horrocks says that the BBC has no line on climate change.

As the good Bishop says, you couldn't make it up.
Meanwhile all those reporting the disappearance of the Arctic ice and the opening of the North East passage managed somehow to overlook that Antarctic sea ice has reached record latitudes, a fact which was reported here, here and here.

Do go and read the rest: there are two things that all of this emphasises. The first is that we really don't know and so we shouldn't be committing billions of dollars to this AGW shit, and the second is that we are being lied to by both scientists and politicians, as well as a good number of vested interests.

Your humble Devil used to accept that the world was warming, although he was sceptical about the level to which man was contributing: he is now coming around to the idea that we are contributing nothing (or so little as to be effectively nothing) and I am unsure that the world is even fucking warming, frankly.

So the Tories, for instance, can join Greenpeace in sucking my big, red, pointy cock.

* UPDATE: my tame scientist and programmer has sent me the following, in response to my query about Fortran (a language that he knows).
Fortran has been around in various guises since 1966, with an updated version being released every 11 years or so since. In its original form it was used on cards with holes punched in them, which was fed into the computer. Of course, then the keyboard/monitor came along so this form of input wasn't necessary anymore.

It's not as popular as it used to be, but is certainly not obsolete. Its very much a "does what it says" language, i.e. its not particularly abstract, which means its relatively straightforward for scientists in general to use, not just computer programmers. It's in this field that it's still popular, i.e. science and engineering, "number-crunching"-style calculations. If you were developing a word processor application you wouldn't use it though.

Reading between the lines of your question, the fact that the climate model is written in Fortran would not be a useful line of attack. I would imagine pretty much every climate model in current use is written in Fortran, and rightly so, its the best language for that kind of job.

There: haven't you learnt something today?


Mark Wadsworth said...

I am unsure that the world is even fucking warming, frankly

It certainly looks a bit dubious. But my problem is, global cooling scares me a lot more than global warming.

Neal Asher said...

I was thinking about that video you posted of the hoody ranting about how he'll grow up in an age when there'll be no fish in the ocean. I then connected that to the huge quantity (560?) basking sharks spotted off the Cornish coast, there feeding on a massive bloom of plankton. Then there was the recent stuff about livestock farmers being happy this year since their stock were able to graze throughout the Winter. But of course we're all going to die... Yawn.

Roger Thornhill said...

I know FORTRAN and I indeed began by using it on punched cards (but that fact is entirely irrelevant to what FORTRAN is).

FORTRAN can be a pain to migrate depending on how it is written, as it permits GOTOs and other things that can make migration difficult if the original is not well designed.

God hope they do not migrate it to C or C++, which is likely as anything to introduce MORE errors into the model!

Anonymous said...

Also if I remember correctly from my Bachelor studies of Mech Eng, FORTRAN is the only language which handles imaginary numbers (square root of -1). This is used fairly commonly in engineering, hence why FORTRAN is used.....

Ian said...
This comment has been removed by the author.
Ian said...

Python can handle imaginary numbers, as can Perl using the Math::Complex library.

Anonymous said...

The leading software for statistical analysis particularly in large data sets is SAS. I find it hard to believe that Fortran is widely used anymore.

Paul said...

Alright, 2 years after the event but I can't let it lie. Fortran *is* widely used in certain fields, in particular high performance computing, i.e. supercomputing, but not for statistical analysis. It is used most commonly by scientists and engineers for *modelling* something, i.e. capturing physical processes in a computer model so that experiments can be performed that you couldn't realistically or practically do in real life, e.g. star formation, molecular processes, or climate modelling. The statistical analysis comes *after* the modelling stage e.g. you run the model many times with different starting conditions, different parameters etc, then use SAS or whatever to analyse the results. Gah.