Sunday, September 11, 2005

Researchers improve hurricane predictions

Researchers take 'LEAD' to improve hurricane, tornado predictions

BLOOMINGTON, Ind. -- Indiana University School of Informatics researchers are part of national team devising technology that more accurately predicts and tracks large-scale weather events such as Hurricane Katrina, which left thousands dead and injured and a path of devastation in its wake.

The Linked Environments Atmospheric Discovery project seeks to create a high-speed computing and network infrastructure that would help meteorologists make more timely and accurate forecasts of hurricanes, tornadoes and other dangerous weather conditions. The national effort seeks to build a "faster-than-real time" system that could save lives and help the public take cover and safety officials better prepare for looming natural disasters.

LEAD is funded by an $11 million grant from the National Science Foundation. IU recently received an additional $2 million NSF grant for its participation in the TeraGrid project to support LEAD and other "science gateways." The funding comes on the heels of IU receiving $4.4 million from the NSF to help improve TeraGrid, an advanced national computing network that allows scientists across the nation to share data and collaborate.

Dennis Gannon is IU's principal investigator, and is joined by co-principal investigator, Beth Plale, Ph.D. Both are faculty members in the School's Department of Computer Science.

"Our goal is to build an adaptive, on-demand computer and network infrastructure that responds to complex weather-driven events," says Gannon, professor of computer science. "A typical scenario will involve constant monitoring of stream data from ground sensors detecting humidity, wind and lightning strikes."

The system also would pool and analyze data received from other sources such as satellites, visual reports from commercial pilots and NEXRAD, a network of 130 national radars that detect and process changing weather conditions.

"Data mining tasks will compare this data to historic patterns," says Plale, assistant professor. "When the conditions are right for the formation of a severe storm, the system will be able to launch hundreds of simulations at the same time. This results in a far more accurate forecast."

As additional sensor information becomes available, it will be used to kill off simulations that no longer have value, refocusing the system's ability to process finer-grid simulations that are more realistic. Newer and smaller NEXRAD radars now under development will provide weather information more rapidly and further narrow simulations.

"The result, for example, would be a prediction of a tornado that is accurate and timely enough to save lives," says Gannon. "This scenario is well beyond the state of the art."

Such forecasts would be of tremendous value for government and public safety officials who plan recovery support for disasters such as Hurricane Katrina, adds Plale. "A more accurate forecast also can reduce the uncertainty administrators face when issuing evacuation orders."

Other institutions involved in LEAD are the University of Oklahoma (lead institution), Howard University, Colorado State University, Millersville University, University of Alabama, University of Illinois, University of North Carolina and the University Corporation for Atmospheric Research-Unidata Program. ###

Contact: Joe Stuteville
jstutevi@indiana.edu 812-856-314 1Indiana University

For more information about the Linked Environments Atmospheric Discovery, go to
lead.caps.ou.edu/.

more at
and or and or and

No comments:

Post a Comment