It’s four days before a major flash flood will hit your local river. Authorities notify you and your neighbors, giving you all ample time to prepare, or at least enough time to move to higher ground. This isn’t just a disaster preparedness obsessive’s fantasy. IBM and researchers at the University of Texas at Austin have developed a piece of flood prediction technology that can accurately warn of impending disaster at up to a hundred times faster than other flood prediction models.
The researchers are currently testing the technology on Texas’s 230 mile-long Guadalupe River (and over 9,000 miles of its tributaries), where every hour it has been able to generate 100 hours of accurate future river behavior based on weather predictions and precise maps of the river system. “We’re taking in a series of data, running it through
a simulation and analytics engine, and getting an output of the flow and depth of the river. If you compare the flows
and depths to the topography of the [surrounding] land, you can make a prediction of where flooding will
occur,” explains IBM researcher Fadi Gebara.
The simulation and analytics engine is so efficient because it relies
on two other pieces of advanced IBM technology: the ultra-fast POWER7
microprocessor and Deep Thunder, a weather modeling system that is far more accurate than your local weatherman.
Currently, the gold standard for flood prediction is HEC-Res,
a piece of simulation software from the Army Corps of Engineers. But
using HEC-Res “to get an understanding of a very local river would take
an enormous amount of time,” says Gebara.
Gebara believes that IBM’s technology could be used for more than
just flood prediction; a modified version could also analyze future
droughts and water surplus–a particularly attractive prospect for
The code base for IBM’s software is already usable reliable, but Gebara worries that it might not initially get to the places that need it the most. In the U.S., the government provides detailed mapping of rivers and the land sitting next to it–information that is inputted into IBM’s engine for more accurate predictions. But this isn’t the case everywhere. So even though the technology could easily be deployed in, say, flood-prone Pakistan, it won’t work well without detailed data inputs. “The issue is making sure you have good initial conditions to feed [the engine],” explains Gebara.
Still, IBM’s flood prediction software will probably pop up soon in cities such as Rio de Janeiro, which already relies on IBM’s water and weather analytics know-how. And when the flood waters start rising, residents will be well prepared.