Predicting floods with AI: How Concordia researchers are helping safeguard cities
The model allows users to have a sense of anticipated flood water levels, helping them design effective flood mitigation strategies
As Quebec grapples with yet another round of torrential rain and flooding, the importance of accurate flood forecasting has never been clearer. To tackle this growing threat, researchers at Concordia University’s Gina Cody School of Engineering and Computer Science have developed a tool that leverages artificial intelligence and digital twin technology to predict floods with greater accuracy.
The team’s model, LocalFLoodNet, uses graph neural networks—an advanced type of computer program designed to understand complex relationships, like how water moves through rivers and lakes—to forecast future water levels. Their research focuses on Terrebonne, a Montreal’s suburb, where they combined this model with an interactive digital twin of the city. This digital twin—a highly detailed virtual copy of the real place—allows users to simulate various flood scenarios, offering valuable insights for disaster prevention and urban planning.
Charalambos Poullis: "Our approach not only improves flood forecasting accuracy but also helps design effective mitigation strategies."
But what exactly is a digital twin? Imagine a virtual version of a city that looks and behaves just like the real thing, all on a computer. Researchers can simulate, or mimic, heavy rainstorms within this digital twin and observe how water would move through streets and rivers. This virtual testing ground helps them figure out the best ways to prevent real-life flooding.
“Our approach not only improves flood forecasting accuracy but also helps design effective mitigation strategies—plans to reduce or prevent damage from floods—tailored to different urban environments,” explains Computer Science and Software Engineering professor Charalambos Poullis, one of the lead researchers.
The research, which has the potential to be applied to cities around the globe, was published in Nature’s Scientific Reports and includes contributions from Gina Cody School researchers Naghmeh Shafiee Roudbari, Shubham Rajeev Punekar, Zachary Patterson, Ursula Eicker and Charalambos Poullis.
Learn more about the Department of Computer Science and Software Engineering at the Gina Cody School.