Researchers developing 'smarter' methods for forecasting floods

Published

Researchers at the Ƶ are using artificial intelligence to more accurately – and quickly – forecast floods.

And they’re doing it well.

UL Lafayette – along with entities such as tech giant Google and the U.S. Department of Defense – are among the top finishers in the Igniting Innovation Award competition earlier this month. The American Council for Technology-Industry Advisory Council gives the award for innovative research that benefits people and communities.

Twelve projects were recognized, including an overall winner. UL Lafayette was among eight finalists. Awards were also given in several specialized categories.

The University’s “Artificial Intelligence and Machine Learning for Flood Prevention and Forecasting” project implements a data mining prototype to discover and curate large amounts of data. Rainfall records, river and surface water levels, soil moisture content and other information is culled from dozens of sources, including satellites, gauges and field sensors.

The data is integrated and interpreted with machine learning, a form of artificial intelligence. The systems are capable of “learning” from data over long periods of time and making predictions without being specifically programmed to do so.

“Advances in artificial intelligence and data science are enabling us to develop innovative methods for flood forecasting and mitigation by taking massive amounts of information and using it for community-scale applications,” said Dr. Emad Habib, a professor of civil engineering. He directs the University’s

Habib is leading a group of University scientists – including undergraduate and graduate students – on the ongoing research project. 

Dr. Mohamed ElSaadani, a research engineer at the flood center, said the technology “has an enormous capacity to determine patterns and trends over time.”

“It provides highly accurate information that can prompt quicker responses to flooding events, and help guide decisions about road closures, evacuation mandates, and other public safety considerations,” he said.

The forecasting model’s capabilities aren’t limited to flood prediction and mitigation – or a specific geographic region.

“The cloud-based system can be shared and adapted to help make determinations about drought, landslides, wildfires or even land usage, which are all affected by variables such as rainfall and soil moisture,” ElSaadani said.

In addition to the and Louisiana Watershed Flood Center, the University’s , , and the are contributing to the project.

The research was sponsored by the CGI through the National Science Foundation Center for Visual and Decision Informatics at UL Lafayette. CVDI is the only NSF center in the nation that focuses on data science, big data analytics and visual analytics.

View a list of projects and learn more about the

Photo caption: Ƶ researchers are using artificial intelligence to more accurately – and quickly – forecast floods. They've created a data mining prototype that places UL Lafayette – along with tech giant Google and the U.S. Department of Defense – among the top finishers in the Igniting Innovation Award competition. Dr. Emad Habib (left) and Dr. Robert Miller of the University’s Louisiana Watershed Flood Center are shown by the Vermilion River discussing water gauges. Photo credit: Doug Dugas / Ƶ