Well, we have pretty good understanding of how the system works---but the time evolution is very sensitive to the initial conditions, in the sense that two systems, starting from arbitrarily close initial states, can evolve all over the available phase space. Since we can't know the initial conditions with infinite accuracy, we can't predict the outcome either. This is the nature of chaotic systems---chaos has a quite precise mathematical meaning.
Finding consistent transition points have been the bane of science for a long time.
Simple systems can be characterised with measurements and controlled tests.
Our true understanding of the atmospheric system is very incomplete. There is a big gap still between what we know, what we think we know and how the system really works.
We are getting closer, but we are far from consistent prediction.