(Phred) wandered out of a yurt and mumbled:

A fairly serious problem with a model based on past weather is that,
if we assume climate is changing, then the basis of the model is
largely irrelevant. Mind you, that mightn't matter so much in terms
of "averages" in the short term future, but when it comes to
predicting extremes (e.g. "100-year events" etc.) all bets are off.

Modern climate models are not based on historical extrapolation from
past records. They are based on numerical solutions to partial
differential equations that govern the fluid dynamics for motions of
the atmosphere (and sometimes oceans). Since they're not
extrapolations from observations they aren't strongly tied to any
particular climate regime. Indeed, many of the same models applied to
greenhouse warming have been used to study climates of the distant
past; see e.g.,
http://www-lsce.cea.fr/pmip/publicat...rp111_009.html




--
"Music is, after all, a highly complex form of mathmatical
encryption. It's so powerful that in modulates human
emotion on soundwaves." Mike Spence