APPLICATION OF STATISTICAL METHODS IN WEATHER PREDICTION

Abstract

computing machines is destined to be most contributory and rewarding. He will be the key individual in helping theoreticians to improve their short-range models by suggesting why and how initial ones went wrong; he will provide ideas for a numerical attack on long-range forecasting and will be indispensable for giving the most effective weatherwise interpretation of the machine-produced prognoses. In weather prediction we are confronted with one of the most difficult scientific problems of our times. Moreover, viewed from the standpoint of the practicing forecaster, it is a problem which has not received the attention it merits. The translation of research results into field application has proceeded at what sometimes seems to be an agonizingly slow pace. Some of this can be explained by the very complexity of the problem we are trying to solve. We are working with an incredibly complicated thermodynamic and hydrodynamic system of truly colossal proportions. The medium with which we are concerned is nonhomogeneous, com-pressible, turbulent, and inherently unstable, and it is strongly influenced by heat sources and sinks which are to a certain extent a function of the state of the system. The medium surrounds a rotating globe which constitutes a boundary whose characteristics are, also to a certain extent, determined by the state of the system. Basically, the problem is this: Given an initial state of this capricious and intricately interrelated hydrodynamic and thermodynamic system, how can a future state be predicted? In spite of the complexities listed above, and others which might be enumerated, reassuring progress is being made. Recent advances toward a better understanding of the hydrodynamical aspects of the problem, coupled with the development of high-speed computing equipment, promise to make the present era one of the most exciting in all meteorological history. Fortunate, indeed, are we who are working in this field in these times, and great is our responsibility to see that none of the tools and techniques of present-day science are overlooked in attempts to solve a problem which, if mastered, not only would be richly rewarding in personal satisfaction but would also materially benefit our collective national economies. What, then, does statistics have to offer that might be of value? It is with some diffidence that the present writer ventures an opinion, since he can claim no professional competence as a statistician and speaks only from the viewpoint of a synoptic meteorologist who has experienced the singular and not infrequent frustrations 806 PROC. N. A. S.