University of Texas, IBM eye faster flood prediction system

The University of Texas and IBM used the Guadalupe River as a test bed for faster flood prediction technology.

IBM and the University of Texas at Austin are aiming to predict floods faster with a combination of sensor data, weather patterns and analytics.

A flood map

The two parties aimed to predict the Guadalupe River's behavior 100 times faster than normal speed. Overall, IBM and the University of Texas are trying to provide several days warning ahead of a flood.

In a statement, IBM said traditional flood prediction methods overlooked tributary networks in rivers. These branches are where most flooding starts. In other words, the main stems don't tell the whole flood story.

IBM's analytics simulated the activities of thousands of river branches and can scale out to millions. Combined with weather simulation models, emergency plans can buy time and be more efficiently carried out.

Texas' Guadalupe River, which is 230 miles long and has 9,000 miles of tributaries, served as the test bed for the effort. It takes an hour to simulate 100 hours of river behavior.

Going forward, IBM and the University of Texas will aim to scale the effort, hook into other data streams and look at applications for urban and suburban flash flooding.

This post was originally published on