Wednesday, November 25, 2009

Too Jive CRU

jive –noun
1. swing music or early jazz.
2. the jargon associated with swing music and early jazz.
3. Slang. deceptive, exaggerated, or meaningless talk: Don't give me any of that jive!

Too Jive CRU...2 Live Crew...see what I did there? Yeah, I know...lame. I doubt anyone even remembers today what 2 Live Crew was. Anyhoo...

When I first heard that someone had hacked into the University of East Anglia (UK) Climate Research Unit (CRU) and posted something north of 60MB worth of files and e-mails, I thought it was no big deal...there'd probably be a bunch of indecipherable climate data files and incomprehensible e-mails talking about the indecipherable climate data. But the e-mails between Phil Jones discussing "tricks" for cooking the books and - worse - conspiring to conceal their methods and data from prying eyes and FOI requests were in plain enough English.

Ace has a great post up today summarizing the chronology to date, but his addendum at the end about climate models, to me, is one of the most damning things I've seen yet on the charlatanism that is climate research. It seems that the models used to predict future climate trends - yes, those models which politicians world wide are citing to tax the living shit out of everyone and reduce us to a hunter-gatherer society - can't even properly predict the past, which, as Ace quips:
...is unfortunately quite knowable, and so we can check their "predictions" against actual records.

They all fail. They all fail.
I'll just quote from the link the same part Ace quoted, because I'm lazy like that:
None of the multiple computer simulations used by a UN climate-change agency for assessments of global warming appears good enough to predict how India’s monsoon will behave, two Indian scientists have said.

The researchers examined 10 simulations of future climate scenarios used by the UN Intergovernmental Panel on Climate Change (IPCC) and found none could reproduce correctly the behaviour of even 20th-century rainfall.

Not a single model could simulate realistically key features of the Indian monsoon...

In attempts to assess impacts of global warming, the IPCC considered 17 models of how climate would evolve as carbon dioxide levels in the atmosphere rose. Some models predict more rainfall over India, but with great uncertainty.

“The models have very serious problems in simulating even 20th century monsoon patterns,” said Madhavan Rajeevan, a senior scientist at the National Atmospheric Research Laboratory, Tirupati, and a co-author of the paper.

“When a model (computer simulation) cannot even show with reasonable accuracy monsoon behaviour in the past, there’s a big question mark over its ability to predict future patterns,” Rajeevan told The Telegraph.
CRU is just too jive to be believed, too jive to get further grant money, too jive to continue operating.

Update: My reply to Charles in the comments about the programmer's remarks in the source code, because I think it's an important point:
Here's the scandal:
The guys writing the software to produce the climate prediction models don't know the structure of the data sets they're dealing with.

5 comments:

Ayrdale said...

And the connection between CRU and (apparent) fudged statistics re NZ's temperature records here...

http://nzclimatescience.net/index.php?option=com_content&task=view&id=550&Itemid=1

...very interesting reading.Note that Dr Salinger worked at CRU, came back to NZ and was suddenly fired by his employer NIWA. I wonder if the above was the reason ?

Eric said...

And yet, here in the US, at least, this story is being mostly ignored, except on conservative blogs.

Mr. Bingley said...

"Those who can't predict the Past are condemned to fudge the Future."

Charles said...

Funny how global warming stopped when Harry took over the data conversion in 95. I can't make heads or tails, but IT types are having a field with this

Eric said...

I was looking through some of the comments in the Fortran source code the other day, and found it stunning that they had no idea what the data structures were.

It's been something like 25 years since I did anything in Fortran, so I wasn't about to try and go through the code itself, but the comments were revealing enough.

Here's the scandal:
The guys writing the software to produce the climate prediction models don't know the structure of the data sets they're dealing with.