Rising levels of discontent

By Terence Kealey,Vice-Chancellor of Buckingham University (THE TIMES, 20/08/07):

When the weather forecasters make their predictions, why don’t they also predict river water levels? This summer, for example, when they forecast rain, why couldn’t they also tell the people of Tewkesbury what the River Avon was going to do?

There is a science of river water prediction – hydrology – but unfortunately it has not improved at the same rate as weather forecasting. In a paper published in the May issue of the Bulletin of the American Meteorological Society, Edwin Welles, of the US Weather Service, has shown that river forecasting hasn’t got any better for 20 years.

Two decades ago river forecasters could make reasonably accurate predictions of water flows for up to two or three days ahead, but for no more; today they can still make reasonably accurate predictions for two or three days ahead, but for no more. Yet in that time meteorology, having adopted new technologies such as Doppler radar, has made significant improvements, and weather forecasts are now more reliable than in the days of Michael Fish – he of the 1987 storms.

Welles explains that hydrology has stagnated because hydrologists do not engage in “verification”. But verification is embarrassing. Under it, an independent team of scientists notes your predictions, then they see what actually happens, and then they inform you of your errors. No one would enjoy such an audit. Yet the meteorologists verify. Meteorology is a difficult science, but thanks in part to verification it has made its significant improvements.

Hydrology is another difficult science. A river is fed by myriad streams and reserves, and water flow depends on complex interactions between precipitation, farming practices, housing and other land uses, groundwater, temperature, evaporation, and numerous other factors. In the face of such complexity we have learnt not to expect too much of hydrology and have been grateful for even the vaguest river forecasts.

But Welles says that we’ve been too nice to the hydrologists, and that we should demand better. The weather forecasters were driven to verification and every other tool of improvement by the scorn of the general public (viz the hapless Fish), so we should now create a new science of hydrology forecast verification, and we should torture the hydrologists by publicly highlighting their mistakes until they, too, improve.

Contrary to myth, advances in sciences rarely depend on genius. Good ideas are quite common, and they often arise in many people’s minds simultaneously. Most scientific advance emerges out of novel techniques (such as verification) or out of novel technologies (such as the Doppler radar that improved meteorology). If we want reliable river forecasts, let us heed Edwin Welles’s call for verification.