Introduction

The possibility of deterministic weather prediction was suggested by Vilhelm Bjerknes as early as 1904. Around the time of the First World War, Lewis Richardson actually attempted to produce such a forecast by manually integrating a finite-difference approximation to the equations governing atmospheric motion. Unfortunately, his calculations did not yield a reasonable forecast. Moreover, the human labor required to obtain this disappointing result was so great that subsequent attempts at deterministic weather prediction had to await the introduction of a high-speed computational aid. In 1950, a team of researchers, under the direction of Jule Charney and John von Neumann at the Institute for Advanced Study, at Princeton, journeyed to the Aberdeen Proving Ground, where they worked for approximately 24 h to coax a one-day weather forecast from the first general-purpose electronic computer, the ENIAC.

1

The first computer-generated weather forecast was surprisingly good, and its success led to the rapid growth of a new meteorological subdiscipline, “numerical weather prediction.” These early efforts in numerical weather prediction also began a long and fruitful collaboration between numerical analysts and atmospheric scientists.

2

The use of numerical models in atmospheric and oceanic science has subsequently expanded into almost all areas of active research. Numerical models are currently employed to study phenomena as diverse as global climate change, the interaction of ocean currents with bottom topography, the evolution of atmospheric pollutants within an urban airshed, and the development of rotation in tornadic thunderstorms.