This year's El Nino phenomenon generated one surprise after another, but a handful of California weather experts are delighted to find that their computer model accurately predicted the state's climate pattern throughout the rainy season. "We thought that we could predict more precipitation for this year, but we were surprised that the predicted trends were so accurate," says Su-Tzai Soong, a UC Davis associate professor of atmospheric science and an authority on regional weather and climate prediction. Since 1992, Soong, along with UC Davis Professor Bryan Weare and colleagues at UCLA and the Lawrence Livermore National Laboratory, has been building a computer model that could provide not daily weather forecasts, but long-term predictions of whether the upcoming rainy season would be wet, dry or normal. The multi-layered model starts with sea-surface temperature measurements provided by the National Weather Service. Those figures are fed into UCLA's global atmospheric model, which predicts climate variables at every 250 kilometers (155 miles). At UC Davis, Soong further refines the prediction with the Mesoscale Atmospheric Simulation, which predicts climate variables at every 20 kilometers (12.4 miles). With surprising accuracy, the model predicted the 1997-98 rainy season would start with a wet period in November, followed by a dry period in December. Furthermore, the model forecast the two periods of persistent rainfall in January and early February that caused widespread flooding in the Central Valley. However, because the predicted wet and dry periods at the beginning of the season lasted longer than anticipated, the heavy precipitation periods came one month later than expected. Soong and colleagues plan to produce a similar computerized prediction this fall in hopes that such forecasts will provide early warning of impending floods and droughts.