The defeat of meteorology as an art form

One of the great, unreported, silent battles of the last half-century was lost about 10 years ago - or won, perhaps, depending…

One of the great, unreported, silent battles of the last half-century was lost about 10 years ago - or won, perhaps, depending on your point of view. It was the long struggle for supremacy between the traditional weather forecaster with his hand-drawn charts, his intuition, and his long experience, and the theoretical meteorologist with a number-crunching Cray or IBM. The boffins won, and weather forecasts were much better, but the artistic school of meteorology had gone forever.

A weather chart produced by such a veteran was a work of art. Armed with pencil and rubber, he (and they were all "hes" in those days) would start by sketching in the fronts, using a previous oeuvre for his initial guidance. His first guess would be refined by reference to variations in temperature or humidity over the region of interest, and then confirmed by areas of persistent rain.

Then he would draw the isobars, the lines of constant pressure, carefully nudging each one in to a smooth and graceful curve. The exercise was something akin to Jonathan Swift's technique for poetry:

Blot out, correct, insert, refine,

READ MORE

Enlarge, diminish, interline;

Be mindful when invention fails,

To scratch your head and bite your nails.

A forecaster of this old school had at his disposal an expertise summed up by one practitioner as "a special facility for intuitively and quickly weighing up all the influences evident from the weather map, rather than a penchant for engaging in the profound study of atmospheric physics". As computers began to play their part in forecasting, he would look at his weather chart like and old grandmaster gazing at the chessboard, determined to prove that his hard-learned expertise was more than a match for anything the latest number-cruncher could achieve. And so it was, for many years.

Computers, however, do it all by numbers. Programmed with the equations which specify the atmospheric processes, and given the pressure and temperature at a certain spot in the atmosphere, they can calculate expected values of these elements at some future time.

When this same operation is carried out for hundreds of points on the weather chart, a new weather map is rapidly constructed. Then the computer does it again - and again and yet again. It moves forward step by step, hour by hour, until it arrives at a forecast perhaps several days ahead of the original observations with which it was provided. The results at first were disastrous. Then about 30 years ago the computer could now and then match the veteran forecaster in predictive skill. But now there is no contest: the computer is facile princeps, undisputedly.