Computers Enhance Flood Prediction
Two days before before the heavy rains that led to flooding along the Ohio River began, forecasters at the federal Ohio River Forecast Center knew that something big was up. But they waited 24 hours before sending out their warning.
They had to. A present, forecasting is not advanced enough to pinpoint whether heavy rains will swell one creek or a bigger river 10 miles away. But they hope that will change.
By building better computer models of the interaction between water in the air (studied by meteorologists) and water on the ground (studied by hydrologists), forecasters may soon be able to give river communities more time to prepare for potential floods.
"This is a neat new time," says John Schaake, chief scientist for the Office of Hydrology of the National Weather Service. "Even though you may not know where something is going to happen, the chances of something significant happening are very, very high. If you know that and can quantify that, then you know how to tell people how much to be concerned."
But until they get better predictive tools, forecasters have to be careful not to send out too many false alarms, says Mark Fenbers, senior meteorologist at the Ohio River Forecast Center in Cincinnati, otherwise, people will no longer pay attention to any warnings.
The link between air and ground water came in 1993, after the record flooding of the Mississippi River. Researchers found that how wet the land was had an important effect on whether more rain fell. Researchers at the National Weather Service knew there was some land-surface effect on the atmosphere, and the Mississippi River data allowed them to quantify that effect and incorporate it into their computer modeling, which has led to improvements in the service's forecasts.
"It's a brand-new concept, coupling all these things together," says Norman Miller, a hydrometeorologist at Lawrence Livermore National Laboratory in Livermore, Calif. "In the past, there's been two communities - the hydrological community and the meteorological community - and it's only in the past few years have both sides have gotten around to talking to one another."
Dr. Miller is involved with one of the most sophisticated attempts to better forecast flooding along three California rivers. Researchers there are modeling dozens of variables along the rivers - the flow of water through soil, the level of urbanization, the type of bedrock, and so on and marrying that to sophisticated predictions about rainfall and snow melt. By creating more precise mathematical formulas, they hope to give communities 48-hour advance warning for flooding instead of the more usual 24 hours.
At a new demonstration project in western Pennsylvania, the National Weather Service hopes to use computers to run through various rainfall scenarios and predict the flooding impact of each. Through this, the service hopes to be able to tell river communities that there is a 60 percent chance that a particular river will crest within the next 24 hours. A parallel demonstration project in Des Moines aims to offer the same kind of forecasts over a much longer range of time.
"The earlier you know something, the more time you have to save property and to save lives," says Mr. Fenbers.