Weather Predictions and Chaos

The mathematical field of chaos arose from Edward Lorenz's early attempts to use computers for long range weather forecasts.

My last blog entry was about eclipses as an example of the predictive power of science. I was able to watch the eclipse however because another prediction was wrong. The weather report for my location predicted rain or snow but was wrong. It cleared up just in time to see the eclipse and the clouds moved back in when the eclipse ended.

In principle weather should be as predictable as eclipses. Weather also results from basic laws of physics. Weather however results from a much more complex system. We therefore do not have the computing power needed to accurately predict the weather. Accurate weather predictions would require a computer as complex as Earth's atmosphere.

This complexity makes weather a good example of a chaotic system. It is in fact the original example. In the early 1960's, Edward Lorenz started using computers to try to compute long range weather forecasts. It didn't work at the time, and still doesn't work, because our atmosphere is too complex. Very minor, almost immeasurable, differences in initial conditions can compound into significant differences. That is essentially the definition of a chaotic system and the origin of the term "butterfly effect". In a chaotic system, a butterfly in Brazil might cause a tornado in Texas.

This chaotic compounding of very small effects makes it impossible to accurately forecast weather. Science can predict things very accurately in principle, but in practice it does not work for extremely complex systems. That is the basic idea behind the mathematical field of chaos.