Next Article
AI can improve flood forecasts by 6 times
Technology
Researchers at the University of Michigan found that mixing artificial intelligence (AI) with current flood models can boost prediction accuracy by up to six times.
Their hybrid system, according to Dr. Tran, was trained on more than 42,000 past US flood events, and blends deep learning with NOAA's National Water Model to correct model output errors and, according to Dr. Tran, can provide streamflow and rainfall forecasts for up to 10 days.
Potentially saving lives and protecting homes
Floods impact over a fifth of people worldwide, but early warnings can cut deaths nearly in half.
This smarter AI tool could be adapted globally, helping communities get faster alerts and prepare better—potentially saving lives and protecting homes where flooding is a constant risk.