An integrative framework for AI-supported coastal hydrodynamics monitoring and analysis

This study introduces an innovative, AI-enhanced framework for monitoring coastal hydrodynamics in regions with limited infrastructure, focusing on Puerto Rico. The goal is to improve real-time analysis of water levels and flood risks caused by climate change-induced sea level rise and storms. The research integrates advanced image segmentation (using the Segment Anything Model), spatial data projection via monoplotting, and pattern recognition through Dynamic Mode Decomposition. Surveillance cameras, enhanced with solar power and wireless connectivity, were strategically deployed along the coast to capture environmental changes. The methodology addresses challenges like low sampling frequency in satellite data, image distortion, and limitations in traditional monitoring systems, offering a cost-effective and scalable solution for vulnerable coastal regions.

The study found that combining AI models with field data enables accurate water level monitoring and identification of coastal hazards like flooding. The SAM model effectively segmented land-water boundaries under varying light and weather conditions, while DMD uncovered significant patterns in water level changes driven by tides and wind. The new monoplotting method improved the precision of coastal mapping, though manual calibration remains a limitation. Results aligned well with NOAA tidal gauge data, validating the framework’s reliability. This approach offers a robust foundation for future coastal resilience strategies, especially in regions facing severe climate threats but lacking traditional monitoring infrastructure.