March 10, 2007


I recommend reading this interview with Dr S. Fred Singer, and the climate change debate. It's got lots of insights on how the science works. And the politics of science...

...Talk about the models. What is a computer model, and what isn't it? What is its purpose in science?

There are many kinds of computer models. But the ones that people mostly talk about these days are the giant models that try to model the whole global atmosphere in a three-dimensional way. These models calculate important parameters at different points around the globe--and these points are roughly 200 miles apart--and at different levels of the atmosphere. You can see that if you only calculate temperature, winds, and so on at intervals of 200 miles, then you cannot depict clouds, or even cloud systems, which are much smaller. So until the models have a good enough resolution to be capable of depicting clouds, it's very difficult to put much faith in them.

But, still, they're playing quite an important role in this debate. Take me through a history of what the models have predicted. You've alluded to this, and how some of their predictions have had to be scaled down. What can models do, and what can't they do?

You have to understand that these models are calibrated to produce the seasons. That is to say, the models are adjusted until they produce the present climate and the seasonal change.

So they're faked, you're saying?

They're tweaked. I think that's a polite way of putting it. They're adjusted, or tweaked, until they produce the present climate and the present short-term variation. You have to also understand there's something like two dozen climate models in the world. And one question to ask is: Do they agree? And the answer is: They do not. And these models are all produced by excellent meteorologists, fantastic computers. Why do they not agree? Why do some models predict a warming for a doubling of CO2, of, let's say, five degrees Centigrade--which is eight degrees Fahrenheit)--and why do other models predict something like one degree?

Well, there's a reason for this. These models differ in the way they depict clouds, primarily. In some models, clouds produce an additional warming. In some models, clouds produce a cooling. Which models are correct? There's no way of telling. Each modeler thinks that his model is the best. So I think we all have to wait until the dispersion in the model results shrinks a little bit--until they start to agree with each other.

What happens when you use these models to try and reproduce past climates, when other forcings are known, like ice ages and so forth? Can they succeed at that?

They fail spectacularly in explaining, for example, why an ice age starts, or why an ice age stops. The most recent result on this was published in early 1999. It's always been known that, for example, the deglaciation--that is, the transition from an ice age to the warm interglacial, which is spectacular--suddenly the ice age ends and the warming starts. And at the same time, you see an increase in carbon dioxide in the record. And these are records taken from ice cores--good measurements....

One trouble with computer models is that every one of them is good at something. If you spend years making a complex model, based on real data, it's going to predict something or other with great accuracy. So it's easy to hold up your model and say, "Look. Crushing irrefutable evidence! Those who dissent are like Holocaust Deniers."

Posted by John Weidner at March 10, 2007 8:34 AM
Weblog by John Weidner