Cloudy with a Chance of Chaos

Weather isn’t random, but bring a jacket anyway.

July 2022

Reading time min

Comic illustration of an annoyed Sun looking up at an angry Cloud.

Illustration: Gary Taxali

In 1969, mathematician and meteorologist Edward Norton Lorenz was running weather models when he needed to restart some computer calculations. He re-entered a variable value to the thousandths. The computer, however, had originally run the computations using values that went to the millionths. The tiny changes in the initial conditions of his models gave him wildly different results. The disproportionate outcomes, he said, were like the flap of a butterfly’s wings causing a tornado weeks later. He’d discovered the seminal mathematical principle of deterministic chaos: the butterfly effect.

It was fitting that the phenomenon emerged from the mathematically chaotic atmosphere, which changes drastically in response to infinitesimal shifts. Today’s climate scientists and meteorologists use powerful supercomputers to account for as many influences as possible when delivering weekly forecasts, issuing flood warnings and mapping hurricane paths, but they’ll never be able to predict the weather with complete accuracy. 

Despite how it may feel sometimes, “the weather is not actually random,” says Aditi Sheshadri, an assistant professor of earth system science. The fluid dynamics equations she uses to predict the movement of the atmosphere around the earth are precise but also limited by available data. To gather enough information for near-perfect forecasts, “you would have to put a sensor at every millimeter squared of the atmosphere,” Sheshadri says. Impossibly small sensors would block out the sky; you’d be breathing them in.

Sheshadri applied the theories of eddy formation to both warmer and colder climates. The news was not good for meteorologists.

Today, meteorologists can predict the weather about 10 days out, but that window is at risk. Last year, Sheshadri and her team used one of Stanford’s supercomputers, called Sherlock, to investigate the effects of climate change on weather forecasting. Weather in the midlatitudes is governed by large-scale storms—or eddies—so Sheshadri applied the theories of eddy formation to both warmer and colder climates. The news was not good for meteorologists. “The eddies tend to grow more quickly, and thus propagate error more quickly, in a warmer climate,” Sheshadri says. 

Advances in computation are making incremental improvements to the accuracy of the 10-day forecast, according to Sheshadri, but science will be no match for Mother Nature. Still, as Lorenz knew well, small changes make a big difference.

Kali Shiloh is a staff writer at Stanford. Email her at kshiloh@stanford.edu.

You May Also Like

© Stanford University. Stanford, California 94305.