Enhancing Coastal Resilience Using Hyperlocal Water Level Sensor Networks and AI

Coastal communities face escalating threats from tidal flooding, storm surges, and a rapidly changing climate. Effective disaster preparedness and response hinge on accurate, real-time coastal monitoring and predictive capabilities. Traditional monitoring approaches, often regionally focused and lacking the density needed to observe impacts of intensifying spatial variability, yield sparse datasets that limit the potentially powerful techniques of AI and machine learning for forecasting and risk mitigation. Our work addresses this challenge by leveraging a network of over 130 affordable, hyperlocal water level sensors deployed across 15 states. This network has amassed over 2 million hours of observations, providing a rich, high-resolution dataset. Here, we introduce three innovative examples that merge these hyperlocal datasets with modern AI techniques. First, we present a forecasting approach that integrates diverse oceanic and meteorological datasets to enhance temporal water level predictions beyond tidal harmonics. We then present the technique of “flood profiling” which is useful for predicting the maximum depth, duration, and rate of change of coastal flooding in advance. Finally, we introduce an application that generates spatial water level predictions, providing real-time water level estimates at any point along a coastline. Together these examples demonstrate the transformative potential of hyperlocal sensor networks and AI for coastal management.