Collaborative Research: CAIG: Transforming Understanding and Prediction of Nearshore Processes via AI-Powered Simultaneous Mapping of Bathymetry, Waves, and Currents

Coastal storms erode beaches and increase flood risk, threatening communities, infrastructure, and ecosystems along the nation's shorelines. Predicting how beaches and nearshore waters will change during a storm requires knowing the bathymetry (shape and topography of the seafloor) just offshore and near the shoreline, but this information is notoriously difficult and expensive to collect, particularly during the very storms when it matters most. This project will develop new artificial intelligence (AI) tools that estimate underwater beach shapes, waves, and currents directly from remotely sensed images, without requiring divers or boats to enter hazardous surf conditions. The resulting open-access system will give federal agencies, coastal engineers, and local planners a more accurate and affordable way to predict beach erosion and design coastal defenses, supporting the national interest in reducing storm damage and protecting vulnerable coastal communities. The project will also create hands-on training opportunities that prepare students to work at the intersection of coastal science and AI, building the workforce needed for AI-driven approaches to environmental and infrastructure challenges. This collaborative project develops physics-informed neural networks (PINNs) and AI-powered computer vision algorithms to simultaneously estimate nearshore bathymetry, waves, and currents from optical and lidar remote sensing data of the surf zone. The resulting Physics-informed, Remote-sensing-driven and AI-powered Modeling (PRAM) system combines deep learning with physical constraints from wave mechanics to produce high-resolution predictions of conditions beneath the ocean surface, including orbital velocities and undertows, using surface observations alone. The research team will field-validate the system across multiple coastal sites to address key science questions concerning wave breaking and roller dynamics, wave skewness and asymmetry in the inner surf zone, and the relationship between hydrodynamic forcing and bathymetric change. By enabling observation of surf-zone processes that are difficult or impossible to measure directly during storms, the project will advance fundamental understanding of nearshore dynamics while establishing a transferable AI methodology applicable to other complex inverse problems in remote sensing and environmental monitoring. 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.