|
Its because weather modelling attempts to be predictive, whereas climate modelling is statistical.
I dont want to push the analogy too far, but its sort of like the difference between predicting the behaviour of a single individual within a mob, and the behaviour of the mob itself. You dont know if any one particular guy will chuck a brick through a window, but you do know reasonably certainly that some bricks will be thrown through some windows, and experience will tell you roughly how many.
Predictions based on chaotic outcomes are not all equally uncertain. Dont forget that chaos theory does not imply randomness: on the contrary, it relies on a definite chain of events, one causing the next. The chaos arises from the fact that the possible outcomes are so sensitive to the starting conditions that tiny variations lead to widely varying outcomes. The butterfly really does cause the hurricane.
Modelling chaotic systems relies on breaking it down into the smallest number of units as are capable of being processed, and stepping them through to an outcome. Im not exactly certain what is used in weather prediction, but its something like taking the wind, temperature and pressure at a number of heights at specific locations, applying basic thermodynamic rules to every one over a given period of time, and feeding the results into the next iteration. To be anything like accurate, youll need locations no more than a few metres or so apart, readings at heights of at most every hundred feet or so, and time intervals of a second or less.. Calculating that lot over a few million square km to forecast the weather for next Thursday is no trivial task.
Compared with that, climate does not look so difficult. Climate is average weather measured over years and decades, and even taken over the entire Earth its nowhere near as chaotic as weather. The main uncertainties in climate prediction are not its chaotic nature, but a proper understanding of the mechanisms involved, and these are improving all the time. |
|