"Don't blame the butterfly", by Robert Matthews. New Scientist, 4 August 2001, pages 25-27.
Why are weather forecasts so often wrong? Received wisdom says that "butterfly effect" is to blame: So sensitive is the weather to slight changes in conditions that even the flapping of a butterfly's wings can cause the weather to change in a distant part of the globe. But recent research by a mathematician collaborating with meteorologists has shown that errors resulting from the computer models used to make weather predictions are as much or perhaps more to blame for inaccurate forecasts than is the butterfly effect. The mathematician, David Orrell of University College London, even nailed down a mathematical theorem about how "model error" grows over time. The next challenge is to try to reduce model error, but as the article points out, many atmospheric events, such as turbulence, are simply very difficult to model.
--- Allyn Jackson