December 1999

Professor Receives $1.25 Mil. to Study Climate Variation By Emily Yaghmour It's one thing to predict the weather--it's a whole different thing to predict climate. By gathering mountains of data about atmospheric conditions and then entering them into sophisticated computer models for processing, meteorologists can create fairly accurate five-day weather forecasts. Predicting climate--that is, the average weather conditions in a given area over a period of months or years--is far more difficult. The earth's atmosphere is considered to be a chaotic system, which makes it dependent on initial conditions. Because it is impossible to know all the initial conditions that affect climate (and even if we did, we don't have computers powerful enough to process all these conditions), it should be impossible to predict climate. At least that's what scientists thought until Jagadish Shukla and his research team discovered that changes in sea surface temperature largely determine changes in the wind and rainfall patterns in some tropical regions. In other words, changes in the temperature of the ocean's surface have the power to overrule initial conditions in determining climate. This discovery proves "that there is a scientific basis for predicting climate," says Shukla, Institute for Computational Sciences and Informatics, who also serves as director of COLA (the Center for Ocean-Land-Atmosphere Studies), an independent research center in Calverton, Md. "What we want to do now is to demonstrate and enhance the predictability." To help him achieve this goal, Shukla recently received a grant from the National Science Foundation for $1.2 million for a five-year project. He and his colleagues plan to extend the same computer models used for weather prediction to predict climate variation. Models have been used in the past to try to predict climate variation, but these models were statistical, not based on measurable physical conditions and mathematical equations as weather prediction models are. In addition, Shukla plans to search for other conditions that, like changes in sea surface temperature, may help predict climate variation. One condition he plans to investigate is that of land surface conditions--for example, how do snow-covered surfaces or wet soil surfaces affect climate? While he hopes and expects to find other factors and processes that can help predict climate variation, Shukla points out that there are limits. Because we are dealing with a chaotic system, he says, climate predictions will always be based on probability.