Introduction
In an era where satellite observations of Earth are commonplace, it’s easy to forget that only a few decades ago, the amount of information available about the state of Earth’s environment was limited; observations were infrequent and data were sparsely located.
As far back as the late 1950s, there were primitive numerical weather prediction (NWP) models that could produce an accurate (or what a forecaster would call “skillful”) forecast given a set of initial conditions. However, the data available to provide those initial conditions at that time were limited. For example, the weather balloon network circa 1960 only covered about 10% of the troposphere and did not extend into the Southern Hemisphere, the tropics, or over the ocean.
Weather forecasters of the pre-satellite era typically relied upon manual analysis of plotted weather maps, cloud observations, and barometric pressure readings when making forecasts. They combined this limited dataset with their own experience issuing forecasts in a particular area to predict upcoming weather and storm events. While those pioneering forecasters made the most of the limited tools available to them, poor data – or simply the lack of data – inevitably led to poor forecasts, which usually weren’t accurate beyond two days. This time duration was even less than that in the Southern H …