Supercomputers still too slow to forecast storm surges

Meteorologists have excelled in forecasting the track of Hurricane Katrina and others but they still need technological advances in storm surge prediction, experts told a congressional committee on Tuesday.

The National Hurricane Center delivered an accurate forecast "56 hours before the storm came ashore, and that is astounding", said Senator David Vitter, a Louisiana Republican, at a hearing convened by the US Senate Commerce Committee's Subcommittee on Disaster Prevention and Prediction. "That's enough time to drive from New Orleans to New York, twice, with a good night's sleep both times."

A panel of witnesses at the hearing said the supercomputers that meteorologists use to model storms have come a long way over the past few decades. A greater scattering of weather buoys, which transmit observations used in the modelling, has also helped to boost accuracy.

But the computers remain too slow and don't always have a great enough density of data to make accurate predictions, according to Keith Blackwell, an associate professor of meteorology at the University of Alabama's Coastal Weather Center.

Blackwell said: "In order to speed up computer hurricane forecast simulations, [forecasters] end up taking many shortcuts on how important physical processes are calculated or represented in a model", which, he said, can "lead to significant forecast errors and often very little continuity between model forecasts".

Until the modelling systems become more sophisticated, he added, "we're going to have a problem with getting people out of the way because the storm's going to do something very unexpected".

Marc Levitan, director of the Louisiana State University Hurricane Center, said his organisation came up with models that predicted Hurricane Katrina's storm surge had the potential to spill water over the levies that enclose New Orleans.

But none of their models are currently capable of foreseeing the sort of levy breaches that occurred during the disaster, he said.

Levitan added: "Where we're lacking in the prediction sense is... predicting the human consequences. What is the wind damage? What will be the number of buildings flooded, number of casualties, rescues needed?... The state of science is not there today."

Senators at the hearing nonetheless heaped praise on the forecasters' performance. Senator Vitter said he regretted the recovery efforts weren't equally co-ordinated.

He said: "What's most frustrating is... it didn't have to be this way. There was really no element of surprise here. It wasn't just predictable, it was predicted many times over in part by these fine folks before us."

Anne Broache writes for CNET News.com