Caig: leveraging AI to unravel marine heatwave impacts on hurricane evolution and coastal resilience against compound flooding

The National Science Foundation Office of Integrative Activities awarded the University of Alabama $1,309,231 on January 1, 2027, under the Geosciences program (CFDA 47.050) to investigate how marine heatwaves influence rapid intensification of tropical cyclones and resulting compound coastal flooding hazards.

The project develops Bayesian deep learning methods and physics-informed artificial intelligence integrated with numerical models to quantify predictive uncertainty and enable near-real-time hazard assessment. Research focuses on storm surge, rainfall, and wind changes associated with tropical cyclones intensifying over unusually warm ocean waters, using Mobile Bay, Alabama, and Galveston Bay, Texas, as complementary study regions. A tiered modeling framework will benchmark advanced machine learning approaches against operational and nonlinear baseline models to ensure robust, reproducible predictions transferable across coastal environments. The work includes training graduate students and a postdoctoral researcher in artificial intelligence and coastal hazard science, with all data, models, and software released openly to support future research and innovation.

Performance runs from January 1, 2027, through December 31, 2029, at the University of Alabama in Tuscaloosa, Alabama. The award funds fundamental research aimed at improving emergency planning and coastal resilience by providing actionable risk information to decision-makers facing increasingly destructive tropical cyclones and compound flooding threats in warming climate conditions.

Abstract

Tropical cyclones are becoming more destructive as the climate warms, especially when they pass over unusually warm ocean waters that can cause storms to strengthen rapidly. These fast-changing storms leave coastal communities with less time to prepare against the ever-increasing risk of flooding. This project will improve understanding of how these events combine to create greater flood hazards and will develop new tools to better anticipate their impacts. The research will support more informed emergency planning and coastal resilience by providing actionable risk information for decision-makers. IT will also train graduate students and a postdoctoral researcher in artificial intelligence and coastal hazard science while making all data, models, and software openly available to encourage future research and innovation. This project will investigate how prolonged ocean warming events influence the rapid intensification of tropical cyclones and the resulting compound coastal flooding hazards. Using mobile BAY, alabama, and galveston BAY, texas, as complementary study regions, the research will quantify changes in storm surge, rainfall, and wind associated with tropical cyclones that intensify over unusually warm waters, or marine heatwaves. The project will develop bayesian deep learning methods to quantify predictive uncertainty and integrate physics-informed artificial intelligence with process-based numerical models to enable accurate and computationally efficient near-real-time hazard assessment. A tiered modeling framework will benchmark advanced machine learning approaches against operational and nonlinear baseline models to ensure robust, reproducible, and transferable predictions across diverse coastal environments. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the foundation's intellectual merit and broader impacts review criteria.- subawards are not planned for this award.