The size and complexity of the atmosphere make weather prediction a physics problem of excruciating difficulty. By 1997, supercomputers, satellite data collection, and progress in applied mathematics have brought meteorology to the brink of maturity. With their aid, three-day forecasts have become as reliable as one-day forecasts were 15 years ago. Eventually, however, prediction will run into a fundamental limitation: the butterfly effect, discovered in 1963, inevitably spoils detailed forecasts beyond about two weeks in advance. Fortunately this problem will not prohibit long-range climate prediction.
Weather and climate predictions come of age.