The last thirty years have seen the development of comprehensive numerical models of the large-scale circulation of the atmosphere, based on physical principles. Such models are quite skillful at describing the evolving weather up to a few days ahead, despite imperfect theory and inadequate observational data. Yet even a hypothetical perfect model, which exactly represented the dynamics of the real atmosphere, and which used data from the best conceivable observing system, could not produce art accurate forecast of indefinitely long range. Any forecast must eventually lose skill because of the intrinsic instability of the atmosphere itself.
This limitation on the predictability of the detailed evolution of the atmosphere (~weather ~ ) does not preclude the possibility of seasonal and longer-range forecasts of means and other statistical properties (%limate ~). However~ we are only beginning to learn what aspects of climate may be predictable, and what theoretical tools and observational data will be required to predict them.
1. Numerical weather prediction
Weather forecasting was revolutionized shortly after World War II. The transformation which occurred then was the development of computer models, based on physical laws, which could predict the continually evolving large-scale atmospheric flow patterns of middle latitudes. Formerly, this task had been carried out subjectively by meteorologists whose skill was due to physical insight, long experience, and semi-quantitative reasoning.
In short, the process was as much an art as a science, and its most successful practitioners often could not explain just how their forecasts were made or teach others to do as well.
It was thus a considerable shock to meteorology when early numerical weather prediction models turned out to produce forecasts of the same quality as those of the best human forecasters. A comparable feat, not yet achieved, might be a computer chess program which played at grandmaster strength. Why did computers instead prove adept at the notoriously difficult task of forecasting the fickle behavior of atmospheric circulation?
The breakthrough in numerical weather prediction was made possible by several factors. Theoretical advances provided a tractable formulation of the problem in terms of equations of fluid dynamics which were highly simplified but nevertheless captured much of the essential physics governing large-scale atmospheric flow. Observations of the weather aloft, as well as at the earth's surface, had become available as input data to define "present ~ atmospheric conditions as a starting point for the initial value problem.