Originally posted by Steve_M:
quote:
1. Models are a time evolution so can't just be compared to current observations if they are to have a credible basis.
What I mean is that for a model to be "successful" then it has to represent in detail the effects of current warming when the effects of current forcings are added (such as warmer troposphere, colder stratosphere...)
They are used to reproduce previous climates (20th C climate is a good candidate because of the reasonable amount of observations).
And they are used to predict future climates. Obviously, we've not had much of a future, but the models in the 1980s did a very good job of predicting the current global temperature.