Water physics 101: IBM testing flood prediction technology

Although it's probably too late to help with Hurricane Irene, new modeling software offers a much more granular view into potential river behavior.
Written by Heather Clancy, Contributor

Even though it likely will be too late to help with Hurricane Irene, which threatens oodles of potential flooding along the Eastern seaboard late this week, IBM is testing simulation software that can help better predict floods -- down to a much more specific area of detail.

The technology lives in the same family as IBM Deep Thunder, which is being used by forward-thinking municipalities and emergency management agencies to help make better plans about the impact of extreme weather on roads, water supplies, communications systems, and so on.

The IBM flood prediction software from IBM Research is noteworthy because it can focus on much smaller river branches, rather than just the main forks and water supplies that gets the attention of current generations of flood prediction software. It relies on sensort data as the input, feeding the simulations running on IBM Power 7 systems. The visualization below gives you a sense of the output:

IBM is conducting a live test of the technology with the University of Austin at Texas. Researchers are using the model to predict behavior along the 230-mile Guadalupe River as well as more than 9,000 miles of related tributaries, according to IBM. That information could be used to make evacuation plans much more specific or to help prioritize areas most at risk from flooding.

Said Ben Hodges, associated professor at the University of Texas Austin Center for Water Resources:

"Combining IBM's complex system modeling with our research into river physics, we've developed new ways to look at an old problem. Unlike previous models, the IBM approach scales-up for massive networks and has the potential to simulate millions of river miles at once. With the use of river sensors integrated into web-based information systems, we can take this model even further."

Right now, it takes about an hour of data-crunching time to produce approximately 100 hours of simulations showing likely river behavior.

Related posts:

Editorial standards