Flood models that planners trust for bridges and dams keep missing the mark as climate change accelerates, according to new research from Cornell University. The study found artificial intelligence can better project future flood risks than traditional methods, especially as extreme weather intensifies.
Researchers tested six flood models using 1,000 years of simulated climate data and found AI-based models outperformed physics-only approaches. Hybrid models also showed promise, but all methods struggled under future climate conditions. The findings appear in the January 2026 issue of the Journal of Hydrology.
The team says regional flood estimates proved far more reliable than site-by-site forecasts. They warn current tools may not adequately protect communities without major upgrades.
FingerLakes1.com is the region’s leading all-digital news publication. The company was founded in 1998 and has been keeping residents informed for more than two decades. Have a lead? Send it to [email protected].