I'm more intrigued to know what is Danish for Sky if Sky is Danish for Cloud.

Yeah, but you're a big-picture guy.

Disappointingly it appears to be Himmel, like in German.

Any images of very confused danes watching a british weather forecast have gone from my head.

One thing, Jo - I accept that predicting the future isn't exactly easy, but surely if they apply the models to history and see if they'd have predicted what actually happened they can judge the effectiveness of specific models? Or am I missing something?

> if they apply the models to history and see if they'd have predicted what actually happened they can judge the effectiveness of specific models?

You're right in theory, yes. What you're missing — or, more precisely, what they're missing — is data. Satellite pictures, for instance, go back, I think, about forty years; high-quality ones more like fifteen years. We now measure temperature in many more places than we did twenty years ago. The historical data from, say, fifty years ago to which we could try applying contemporary models is, by today's standards, woefully inadequate.

And the thing about deterministic chaos — which certainly does apply to our climate models, even if it doesn't apply to the climate itself — is rapid and unpredictable divergence from similar conditions. So, just because a model provides accurate predictions over twenty years for a given set of initial conditions, that doesn't necessarily mean that it will provide accurate predictions over twenty-five years for the same set of initial conditions or even over ten years for another set. (Meteoroligists use this fact all the time now to work out how predictable a weather system is, so, while they can't always give us an accurate ten-day forecast, they can say with some certainty that they can accurately extrapolate this weather system seven days into the future but that weather system only five days.)

So we can apply a model to what we knew of the world's climate in 1950. Let's say, for the sake of argument, that the model correctly predicts the climate of 2000. What do we learn from that? We learn that we have a model that works over a period of fifty years given the initial conditions of 1950. We don't learn that the model works over sixty years. We don't learn that it works over fifty years given the initial conditions of 1970. We don't learn that it works for the large amounts of data available in 2000 that were not available in 1950. We do learn that all these things are a bit more likely; accurate predictions increase our confidence in a model. But they shouldn't increase our confidence to the extent that we actually trust it — especially not when a team of Danes discovers a verifiably true and significant fact that the model completely ignores.

I am not up to speed on all the recent climate modelling work, but my recollection is that some of the classic greenhouse effect models in fact didn't work terribly well predicting the future from the past and had to be fudged to fit, with a model change at some point in the recent past. Not proof of falsehood, but not exactly re-assuring. Perhaps the best current models do better - I don't know.

Well, we now know that every single one of those models was created by people who didn't know of the existence of a significant cause of cloud formation, so it's hardly surprising.

All this being said, the Ancient Greeks and Romans had a superstitious belief in the cosmic specialness of the circle and therefore insisted that all planetary movements be based on combinations of perfectly circular orbits; ellipses were not allowed. That didn't stop them creating models that were so good that their predictions came pretty damn close to today's standards of accuracy — they could make accurate hundred-year predictions, despite knowing sod all about the causes of planetary movement.

Orbits are a bit more predictable than weather, though.

I quote from the fourth assessment report;
"We think we have high understanding only for carbon dioxide and methane, medium for ozone and solar forcing, low understanding of stratospheric water vapor, the direct effect of aerosols and contrail cirrus, not to mention very low understanding of indirect cloud albedo effect and land surface albedo."
We're being prophesised doom to on the basis of computers modelling climate (which is after all an agglomeration of weather) when forcasts past 7 days tend to be a bit woolly, on the basis of diddly squit knowledge of many of the mechanisms. I know we know the basic mechanics (well most of them) pretty well, Newton, Quantum, et al. but scaling up to such a big system is impossible over any decent projection in time. We can predict the next few days fairly well, (cf. the Met Office), but they wouldn't hope to predict say, 3 weeks on Wednesday's weather. What alchemy makes predictions years in advance suddenly accurate enough to demand humanity falls into lockstep with George Monbiot? Nothing but wishful thinking and perhaps a bit of a god complex I suggest.