Cortana Intelligence Suite hopes to help predict floods and prevent disaster

Kellogg Brengel

Looking for more info on AI, Bing Chat, Chat GPT, or Microsoft's Copilots? Check out our AI / Copilot page for the latest builds from all the channels, information on the program, links, and more!

Microsoft Research Director Kristin Tolle published an article on the Microsoft Research blog about an interesting and powerful example of how Microsoft’s Cortana Intelligence Suite can impact and even save lives.

The story starts in the wake of Austin, Texas’s severe floods of October 2013. During a massive downpour, a stream gauge monitoring Onion Creek failed to report a rise in water levels. Accordingly, emergency responders diverted resources to areas that had triggered flood alerts. But the Onion Creek gauge was defecitve, and the banks of Onion Creek were actually overrun. Consequently, over 500 homes were flooded and five people lost their lives.

Because of incidents like Onion Creek and many more, researchers, federal agencies, emergency responders, and businesses have partnered to make a national standard for flood data to help forecasting and emergency response. After developing systems for some of the water regions involved in this project, Kristin Tolle and a Microsoft Research colleague presented on the Cortana Intelligence Suite to students at the National Center Summer Institute (NCSI).

Cortana Intelligence Suite is a cloud-based suite of tools Microsoft provides to help organizations analyze vast amounts of data to gain valuable insights. We’ve reported on these tools doing everything from helping companies manage fleets of repair vehicles to predicting maintenance needs for elevators. But after presenting at the NCSI event, some students wondered if this suite of analytics tools could prevent situations like the Onion Creek disaster.

Flowchart for how SHEM works via Microsoft Research
Flowchart for how SHEM works via Microsoft Research

Tim Petty of the University of Alaska went on to create what he calls streamflow hydrology estimate using machine learning or SHEM. Using Cortana Intelligence Suite, SHEM is able to act when a stream gauge, like the one at Onion Creek, fails. Petty’s modeling of historical data and his use of machine learning allows SHEM to predict water levels even when a gauge fails.

There is more technical information in Tolle’s Microsoft Research post about how the experiment uses Boosted Decision Tree Regression and how the team is developing the project. But the important point is that Cortana Analytics Suite could help responders forecast flood conditions and deploy emergency resources accordingly to minimize the loss of life and property. While still a project in development, it looks to be an example of how machine learning can transform lives in ways we had not thought of before, much like what was recently said by a Microsoft executive at an AI conference last Thursday.