Anyone who has planned a picnic knows that current weather forecasts are not very reliable more than a week or so ahead. But by looking at conditions higher in the atmosphere, scientists may be better able to predict weather patterns weeks or months into the future, according to results published today in the journal Science.

Conventional methods of weather forecasting concentrate primarily on analyzing conditions in the troposphere, the lowest layer of the atmosphere where clouds and storms appear. Little attention is paid to the overlying stratosphere because major changes in its state typically take weeks or months to evolve. But Mark P. Baldwin of Northwest Research Associates and his colleagues present evidence that air circulation in the stratosphere can in fact hold clues to long-term weather behavior. They determined that temperature and pressure fluctuations transmitted from the troposphere up to the stratosphere do not simply die away as previously suspected but instead accumulate and cause subtle changes in stratospheric conditions. The changes are eventually fed back to the troposphere weeks later through a mechanism dubbed "stratospheric memory." To investigate this phenomenon, the scientists studied the Northern Annular Mode, a system of regular air movements that spans much of the northern hemisphere, and analyzed weather records as far back as the 1950s. They found that unusual behavior in the troposphere was associated with similar anomalies in the stratosphere that occurred 25 days earlier.

Having demonstrated this correlation in behavior, the scientists now hope to further explore the interaction between the two layers, which they admittedly dont yet fully understand. Still, the relationship may be enough to aid in weather predictions because meteorologists rely on two different methods to come up with extended forecasts. One is to enter detailed information on current conditions such as temperature and wind velocity into computer simulations. The other method is to examine historical weather data and make predictions based on statistics. Although a more detailed understanding of the physics of the two layers is necessary to improve the computer models, the stratospheric effects can simply be used as another factor to incorporate into statistical predictions. In addition to forecasting the weather, the authors hope that these insights will lead to improved models for global warming, ozone depletion and the effects of volcanic eruptions.