Numerical Weather Prediction

The proposition that a forecast is fundamentally a computation: given an accurate enough picture of the atmosphere's current state, its future state can be found by numerically solving the equations that govern fluid motion and heat, rather than by comparing today's weather map to similar ones from the past. Vilhelm Bjerknes set out the program in 1904; Lewis Fry Richardson attempted the first real calculation by hand in 1922, producing a badly wrong forecast for reasons later diagnosed as noise in his starting data. Electronic computers made the arithmetic fast enough to outrun the weather itself, and by the 1950s numerical prediction had become the basis of routine forecasting, though Edward Lorenz's later work on deterministic chaos showed it a hard ceiling on how far ahead any such forecast can reach.