Forecasting became a science once it was treated as an initial value problem: measure the atmosphere's present state precisely, then apply the laws of physics to compute what comes next. Borrowing equations from another field of physics and adapting them to moving air laid the groundwork for computer weather prediction.
When an established agency refuses a new approach, an outsider can still prove it through cheap improvised experiments. Turned down by the bureau, one researcher spun pans of water in its basement and uncovered the large waves that steer upper-level winds.
Tools such as wind-profiling balloons, radar, computers and satellites drove the science forward as much as the scientists did. They revealed atmospheric structure no one had seen, which forced theory to be rethought, which in turn improved forecasts.