Pocket calculator

by Gareth on September 8, 2009

homer.jpgThe National Institute for Water and Atmospheric Research (NIWA) has bought a shiny new supercomputer. For $12.7m they’re getting a IBM Power 575 that will be the fastest climate machine in the southern hemisphere, 14th fastest in the world — and it will get an upgrade in 2 years which will double its speed. Sounds like a good deal to everyone but NZ’s climate cranks. Bryan Leyland, “chair of the economics panel” of the NZ CSC rushed to issue a press release:

“It is a national scandal that NIWA are squandering $12.7 million of taxpayers’ hard earned money on yet another supercomputer. In spite of buying a Cray T3A supercomputer several years ago, their predictions of the climate have been spectacularly wrong. They failed to predict the 1998 El Nino event, the cooling that has been noticeable since 2002 and the increased cooling that has been recorded over the last two years.”

A NIWA spokesman was quick to point out that they bought the Cray in 1999, so would have been hard-pressed to use to it predict an El Niño in the preceding year, but Leyland’s outburst is mainly interesting for two reasons: it’s an amusing public parade of ignorance (a bit like standing in the middle of Wellington wearing a dunce’s hat shouting “Look at me!”), and because he recommends that NIWA give up climate modelling and instead rely on the work of a British forecaster called Piers Corbyn. Let’s start with Leyland’s take on climate models.

Leyland starts by repeating the “cooling since…” crank meme, and complains that climate models failed to predict a fall in temperatures. There is no “cooling”, so Leyland is wrong before he even starts, but he’s also wrong about climate models. Climate models don’t forecast year-to-year variations in global temperature, they’re used to project climate states — the statistics of climate — for future periods under different greenhouse gas scenarios. These statistics are derived from multiple runs of single models, and by averaging across lots of different models. Individual model runs, however, reproduce the range of natural variations, as you can see in this chart from RealClimate:

RCmodelrunchart.jpg

In comparison with the range of global temperatures covered by the individual model runs (all the coloured lines) the actual annual temps look quite smooth. Leyland then demonstrates a nice turn of phrase:

What this tells us is that there is something fundamentally wrong with the climate models that they are using in their existing computer. The $12.7 million new computer will simply give them more wrong answers even faster than before. It will do nothing to solve the fundamental problem – that, all over the world, climate models have dismally failed to predict the climate.

In fact, climate models around the world do a pretty good job of modelling the climate system. They’re not perfect, but they get the big picture right. And we can be confident that as the atmospheric greenhouse gas loading increases, the climate system will change. The world will warm, ice will melt and weather patterns will change. NIWA’s new computer will help us to get a better picture of what might happen here — and that’s an essential component of preparing for the change that’s inevitable. Leyland, however, wants us to return to the dark ages:

NIWA should also take notice of the better than 80% accuracy of weather predictions produced by Piers Corbyn at Weather Action in the United Kingdom. For several years, he has successfully predicted severe weather events in United Kingdom months in advance and recently he has expanded this to hurricane prediction in the Caribbean Sea, the Pacific and Indian Oceans. Before NIWA get any money for yet another supercomputer they should be forced to explain exactly why a small organisation with desktop PCs has consistently outperformed all the supercomputers running climate models all over the world.

Corbyn is a controversial figure in the world of British weather forecasting, in much the same way that Ken Ring is controversial over here. Ring makes forecasts based on astrology and the movements of the moon, whereas Corbyn claims to use a system based on “solar weather“. Both make claims for accuracy that are hard to substantiate. Corbyn is a paid-up member of the Heartland-funded climate denier community, and is on record (here) as concurring “fully with Christopher Monckton and his conclusions” on the hockey stick controversy, which should tell you all you need to know about Corbyn’s stance on climate change. I’ve not attempted to see how successful Corbyn’s forecasting technique actually is (though Stoat and James Annan have both had “discussions” with him, and I dissected Ring’s forecasts in detail here), but I doubt he’s “consistently outperformed” anything. It’s good enough for Leyland though, presumably on the basis that any support, however weak, is good support:

To many climate scientists, the answer is obvious. Piers Corbyn bases his predictions on solar effects, not man-made greenhouse gases. It is high time that NIWA gave up its greenhouse gas induced delusions and followed suit.

When the day comes that the laws of physics are repealed for being too left-wing, Leyland may be able to make a case. In the meantime, the really important question is why Leyland is so unpatriotic? New Zealand has its own fringe forecaster who thinks that global warming isn’t happening and who is prepared to expound at length on the subject. Why is a British heliophile preferred to our very own lunar enthusiast? Perhaps Ken has fallen foul of the tall crank syndrome, or is astrology a star too far for a Heartland-approved crank to espouse? I think we should be told…

[Kraftwerk (brilliant!)]

{ 22 comments… read them below or add one }

Doug Mackie September 8, 2009 at 4:28 pm

Oh, if only NIWA had instead squandered our money on an atom smasher we would know the answer.

From Leyland’s submission to 2008 ETS Select Committee”
“Research is being carried out at CERN (a huge “atom smasher” complex in Switzerland and France) that is expected to confirm that high energy cosmic rays trigger the formation of low-level clouds. If this proves to be the case, then the theory that “manmade carbon dioxide causes dangerous global warming” must be seriously questioned.”

R2D2 September 8, 2009 at 8:12 pm

Climate models,

What can I even began to say about this article by the New Scientist….

http://www.newscientist.com/article/dn17742-worlds-climate-could-cool-first-warm-later.html

Gareth September 8, 2009 at 8:31 pm

You could point out that none of these modellers are disputing the overall direction of change, only discussing the extent to which natural variations are overlaid on the underlying rising trend of CO2 forcing. Latif’s modelling group is attempting to produce short term “forecasts” of climate by feeding real ocean heat content starting information to their GCM. The technique is in very early days, and fraught with difficulty. See this RealClimate post for more detail.

Will next year be hotter than last? Will it be a new record high?

Don’t know. (Don’t care much either, but I might be tempted on a suitably framed bet that next year will set a new record).

Will the decade from 2011 to 2020 be hotter than the decade 2001 to 2010?

Yes. Name your sum.

R2D2 September 8, 2009 at 9:02 pm

“You could point out that none of these modellers are disputing the overall direction of change”

Well that depends, disputing with who? They may not dispute the effect of CO2 on temperature is positive, but they did make these comments:

“Model biases are also still a serious problem. We have a long way to go to get them right. They are hurting our forecasts,”

So I think they are disputing the gradient of expected future warming – in disagreement. Mann and Co at real climate seem to be in disagreement – even offering a bet the authors are wrong.

He also disagrees with some of your recent articles on arctic:”Pope warned that the dramatic Arctic ice loss in recent summers was partly a product of natural cycles rather than global warming. Preliminary reports suggest there has been much less melting this year than in 2007 or 2008.”

I also like these comments; “Breaking with climate-change orthodoxy, he said NAO cycles were probably responsible for some of the strong global warming seen in the past three decades.”

Gareth September 8, 2009 at 10:13 pm

So I think they are disputing the gradient of expected future warming ..

Nope. Read the reference I gave.

Do natural climate features like the NAO play a part in recent Arctic events? Of course. If you read the WWF report you’ll find an excellent discussion of the NAO.

As to this year’s sea ice melt, I’ll be posting soon. I may not win any bets, but the ice has most certainly not “recovered” in any real sense.

R2D2 September 8, 2009 at 11:30 pm

Do you know where I can find a copy of the IPCC first assessment report?

Gareth September 9, 2009 at 8:29 am

You could do your own research, y’know. It’s where you might expect it to be, at the IPCC site.

R2D2 September 9, 2009 at 7:31 pm

I did expect it to be at the IPCC site, but only the 3rd and 4th assessments are (unless I just can’t work the site out?).

I have looked quite a few places but can not find a copy. Thought maybe you would have had a look at one when you were researching your book?

Gareth September 9, 2009 at 7:37 pm

Did you not bother to follow my link, and scroll down the page? Do you require a spoon in order to be fed…?

R2D2 September 9, 2009 at 7:39 pm

Umm yeah I did, maybe you could post the link to the PDF, I can’t see any link on the first assessment report.

Gareth September 9, 2009 at 8:06 pm

Apologies: I assumed there was a link on that page, and there isn’t. I’ve done a quick Google and can’t find anything obvious – it may just be that 1990 was pre-digital publishing, and so pdfs were not/have not been made. You may have to search the index at your local library…

I didn’t read it during research for HT. I did go through a lot of AR3 and 4, though.. ;-)

R2D2 September 9, 2009 at 8:14 pm

I agree PDF was probably not around in 1990, but plenty of early work has since made it to PDF:

http://scholar.google.co.nz/scholar?as_q=&num=10&btnG=Search+Scholar&as_epq=&as_oq=&as_eq=&as_occt=any&as_sauthors=&as_publication=&as_ylo=&as_yhi=1990&as_allsubj=all&hl=en

Yeah it is hard to track down alright.

The reason I am looking is I would like to see what predictions they made for 2009. You are quick to make a bet about what things will be like in 2020, but that is a long time to wait – why not just see if the IPCC was right in 1990?

The cynic in me says if it was completly right the IPCC would have a link to it, if it was completely wrong all sceptics would link to it. Strange that it is niether – maybe it is the middle of the two

Gareth September 9, 2009 at 8:26 pm

You want to look at Hansen’s testimony to Congress in 1988, where he presented some model projections. There’s a look at how they turned out here.

…when asked whether any climate model forecasts ahead of time have proven accurate, this comes as close as you get…

Dappledwater September 9, 2009 at 8:21 pm

He also disagrees with some of your recent articles on arctic:”Pope warned that the dramatic Arctic ice loss in recent summers was partly a product of natural cycles rather than global warming. Preliminary reports suggest there has been much less melting this year than in 2007 or 2008.”
– R2D2

2008 Arctic sea ice volume was the lowest on record, despite having a greater extent than 2007:

http://nsidc.org/news/press/20081002_seaice_pressrelease.html

In particular:

NSIDC Senior Scientist Mark Serreze said, “When you look at the sharp decline that we’ve seen over the past thirty years, a ‘recovery’ from lowest to second lowest is no recovery at all. Both within and beyond the Arctic, the implications of the decline are enormous.”

and:

“Simply put, the natural variability of short-term weather patterns provided enough of a brake to prevent a new record-low ice extent from occurring.”

and:

NSIDC Research Scientist Julienne Stroeve said, “I find it incredible that we came so close to beating the 2007 record—without the especially warm and clear conditions we saw last summer. I hate to think what 2008 might have looked like if weather patterns had set up in a more extreme way. ”

Yes experts are aware both climatic warming and prevailing weather conditions will have an effect on Arctic ice loss. So the quote attributed to Pope seems a bit strange.

R2D2 September 9, 2009 at 8:30 pm

This year is 2009

Pope said previous records (since 1970) may have been “partly a product of natural cycles rather than global warming”

Gareth September 9, 2009 at 9:18 pm

Vicky Pope is head of climate change research at the UK Met Office, and has published several articles pointing out that natural variability is overlaid on a underlying warming trend. She’s not saying there’s no warming, just that we shouldn’t ascribe every weather event to climate change. And she’s correct, of course. But the decline in the Arctic ice is a long term trend, not a wiggle… She addresses some of these issue on this page – which also has a nice graphic showing how modelling has become more complex – and therefore realistic – over the last 40 years.

Dappledwater September 9, 2009 at 10:03 pm

This year is 2009 – R2d2

In which the summer sea ice extent minima is yet to be reached. Making 2008 the most recent complete data. Did you not realize this?.

Dappledwater September 10, 2009 at 12:47 am

Further damning evidence of “alarming” ice mass loss in the Arctic:

http://www.sciencedaily.com/releases/2009/09/090901143321.htm

Doug Mackie September 10, 2009 at 3:12 pm

I wonder if the puter could be used to solve this fiendish conundrum pointed out by one person in their submission to the ETS review:

“Remember today’s computers still can’t tell you which numbers between 1 and 20 add up to 20. The program is trivial – the run time hundreds of years. The very nature of this type of problem belongs to a well known class of problems for which no solution can exist.”

Gareth September 10, 2009 at 5:46 pm

Don’t be a tease, Doug… Who?

CTG September 11, 2009 at 2:47 pm

Does anyone know what algorithm this is talking about?

I wrote a program to do what I think he is saying, i.e. find all possible combinations of numbers less than 20 that sum to 20. It took 160 milliseconds on my laptop – and that was written in Java without any attempt at efficiency. As he describes it there, it doesn’t sound like an NP-complete problem, because it is fairly trivial to discount large portions of the problem space. An NP-complete problem usually is computationally difficult because there is no way to reduce the problem space algorithmically.
I wondered if he was referring to factorisation, as that is often perceived as being difficult or impossible. However, there have are special-purpose devices that are able to factor large semi-primes in days, and clusters of general purpose computers have done some quite large numbers (1000 bits) in a few months – hardly “hundreds of years”.
In any event, I’m not sure what is the relevance of NP-complete problems to climate modelling.

Doug Mackie September 11, 2009 at 11:59 am

“John McK Blundell”

Includes special bonus gibberish:

“No one debates Einstein’s Space Time theory today – why ?
Because we all accept that GPS units which use Einstein’s theory are able to provide us with our current locations. Yet even today we have what is referred to as the ‘ Pioneer Anomaly ‘ whereby some satellites are over time just a little closer to the sun than predicted by current theory. Research continues on this issue from a number of approaches: Poor Data collection, Thermal emissions, A Modified Theory of Newtonian Dynamics ( MOND ) for low accelerations and many others.”

Previous post:

Next post: