The terminology is chaos theory. The models use non-linear PDEs (the Navier-Stokes equations on a rotating sphere plus considerations for radiative transfer, turbulence, and liquid phase change). It is an initial value problem and slight differences in the initial state will result in differences in the simulation. Since you can’t know the initial state perfectly, there is a limit to predictability. Ensembles take advantage of this by making multiple runs with slightly altered initial states to define the envelope of possibilities. That is the short version.
Quote:
Originally Posted by mntlblok
Aren't these various models examples of that "butterfly effect" thing - how initial assumptions (if off even a tiny bit) get magnified out into the future?
|