Sunday, 13 December 2009

ITS THE MATHS STUPID!

With all the furore over the climategate emails, it's easy to overlook the simple truth which is that the maths of climate is impossible to solve. This article by a mathematician puts its effectively but simply, and yet the pretence goes on that we not only understand but have solved the mysteries of forecasting and controlling our climate. What utter folly and misplaced hubris.

3 comments:

Dan Olner said...

Nobody is trying to "solve" the maths of climate, any more than equations are solved in weather models. They have no analytic solution - hence why they're simulated, and hence why meteorologists use 'ensembles' of predictions with a range of initial conditions.

As to whether you can predict anything in complex systems: the best analogy I always come back to - today may be warmer than yesterday; does that mean Spring is coming? No? Why not? Because that's not the right data - to look for a trend, you need to use a larger sample. So what you can deduce depends on what scale (both time and space) you're looking at. A nice intro to inference in complex systems - lecture 1 from:

http://www.cscs.umich.edu/~crshalizi/weblog/556.html

Look out for what statistical inference Shalizi picks up from the simplest chaotic system, a basic logistic map equation.

You may still want to argue that the climate system's boundary conditions are too complex to make easy inferences - fair enough, that's a separate argument to have. But attacking climate models on the basis that they can't "solve" the climate makes no mathematical or modelling sense. They attempt to infer what they can, based on what we know - this is reflected in the very wide confidence intervals in climate models.

Derek Tipp said...

Perhaps "solve" is the wrong word to use and I should instead have used "predict". If computer climate models do not predict over the short term, then we should not rely on them over the long term. The article I linked to explained the situation very well. Scientists are, at the present time unable to make useful predictions as the variables are so poorly understood. (And they probably never will be able to predict climate over the long term.)

Indeed the level of certainty expressed by the IPCC may well be their undoing in the end.

Tim Auld said...

Given that you don't understand the climate and that your life and the lives of your progeny depends on it, why do you advocate changing the composition of the atmosphere significantly?