David, it is very easy to test if two or more entirely different initial states lead to a statistically similar climate. This is one of the points of running ensembles. Another way to see this is that weather states only a few weeks apart bear little correlation to each other. Initial state information is quickly lost among the growing modes that dominate the tropical and midlatitude systems. The analogy is closer to a boiling water pot with continuous energy input than a gently swirling unheated pot. The memory of the initial state is short in these conditions.
↧