instruere...inlustrare...delectare Disputations

Monday, November 30, 2009

Science and data, i

A lot of people are spun up over the material stolen from the Climate Research Unit of the University of East Anglia. Some are spun up over the material, some are spun up over the stealing, some are spun up over others being improperly or inadequately spun up themselves.

I don't know from the CRU, and I've forgotten pretty much everything I learned in the numerical weather prediction class I took in grad school. So I'll just offer a few thoughts as data for reasoning about all this.

First, climate science differs from the kind of science most people have experience with from high school (or even most undergrad science labs). The scientific experiments conducted by students tend to be done more or less directly on the physical system of interest. If you want to see how inelastic collisions work, you make things collide; if you want to see what K + H2O ->, you drop some potassium in some water and take notes. A hundred years later, someone else can do the same experiment and compare their notes with yours.

For the most part, though, climate science experiments aren't experiments on the climate itself; the subject of the experiments is a particular climate model. The question the experiment is designed to answer is whether the model can be used to correctly answer questions about the real-world climate.

And if the questions we want correctly answered are about the real-world climate fifty years from now, there's really no way of knowing for sure today that a model is adequate.

What can be known, though, is whether a model is adequate to answer questions about the future if certain assumptions are true. Then the work is to determine whether those assumptions are true (even better is to change the model to reduce the amount of assuming that has to be done).