Is no news bad news?
posted at 2/5/2012 1:10 PM EST
Still waiting for reader "topaz978" to post references to scientific articles that bear on claims made. The reference with which this thread started was not a scientific publication but an ad from a university PR shop. There is no point to reviewing "numbers" as though they had some meaning--which is what the PR encourages--when technology used to generate numbers can't be validated.
Turbulent fluid circulation is intrinsically chaotic, a behavior unaltered by scales of space and time. Likewise, numerical solutions of time-dependent equations describing atmospheric flows diverge uncontrollably. Rather than finding a way to predict long-range trends of circulations from physics, current modelling for climate prediction imbeds learning technologies.
Outputs of the learning technologies are hypersensitive to information on which they are tuned, leading to a political selection of "correct" data sets. Because there is no analytic theory behind this approach, results lack statistically valid error estimators.
The dodge used so far in attempts to cope with those fundamental problems has been to subdivide space and time ever more finely and to average increasingly large numbers of model runs, hoping somehow an average might mean something. While it could be "philosophically" interesting, to rescue a word, nature doesn't behave that way.
Nature makes, in effect, one run. Of the thousands produced in a typical model experiment, which might that be? [ D.A. Stainforth, et al., Uncertainty in predictions of the climate response to rising levels of greenhouse gases, Nature 433(1):403-406, January, 2005, available from http://www.climateprediction.net/science/pubs/nature_first_results.pdf