This article is addressed to the layman with a science education. It is thus not addressed specifically to the meteorologist, but it contains certain information about the weather services, and the impact of numerical weather prediction on them, that I believe will interest those in the meteorological community who have not had close associations with the service side of our house.
The weather services are principally concerned with weather and river forecasting and serve virtually all of the general public and the entire economy of our nation. Numerical weather prediction is at the core of the complex set of forecast services, and more accurate forecasts depend on progress in numerical weather prediction more than on anything else. This is as true now as it has been during the 20-odd years since operational numerical weather prediction began. Progress has been, and still is, paced by advances in computer power. Numerical weather prediction has seen a succession of four generations of computer hardware, and during its 20-year era has halved the overall errors in wind forecasts at all levels up to 12 000 m, and has improved the accuracy of forecasts of whether or not precipitation will fall. One of the most important forecasts, however, that of how much precipitation will fall, has been little improved for 15 years, although there was a dramatic improvement around the turn of the 1960s. Recent research results, however, indicate that the greatest improvements will be made in this area during the coming decade.
1 This article is being published in the book Geophysical Predictions (National Academy of Sciences, Washington, D.C., 1978). The book covers predictions of a wide variety of phenomena in the atmosphere, hydrosphere, and solid earth.