91av

FORECASTING the weather is tricky, but there is a silver lining. Chaos theory
reveals when you can predict the weather with confidence—and when to give
up and go home.

D. J. Patil at the University of Maryland in College Park and a team that
includes James Yorke—who coined the term chaos theory—looked at
forecasting data from the US National Weather Service. The service uses computer
models of the world’s weather systems to create a range of different forecasts,
all based on tweaking temperature, wind, humidity and pressure to see the most
likely weather conditions for the next few days.

For example, the service takes a single weather feature, such as the Gulf
Stream, and puts the current measurements into its model. Then it tweaks the
wind by a tiny, random amount, and runs the model forward by a few days. Chaos
theory predicts a tiny change can have a dramatic effect, so the service
produces lots of forecasts to find out which outcome is the most likely.

The team looked at five such tweaks in the wind to see how different the
forecasts would be. Their results, which will appear in a future issue of
Physical Review Letters, were surprising. Most of the planet was unaffected
by the tiny changes in the wind. But they had a huge effect in just a few
spots—covering about 20 per cent of the globe. For these “hot spots”, it
was incredibly tricky to predict the weather over more than a day or two.

Further experiments showed the same effect for temperature, precipitation and
pressure. Patil says the hot spots drift eastwards at about a third of the speed
of a storm, and tend to cluster at about 30° north and south of the equator.
Forecasters in these moving chaotic areas shouldn’t even pretend to know what
will happen in a week’s time.

Jan Barkmeijer of the European Centre for Medium-Range Weather Forecasts in
Reading, Berkshire, isn’t much surprised by the results. “We already know that
some areas are more sensitive than others. But this may be a more efficient way
to locate those areas,” he says.