The Gulf Coast is losing land due to rising sea levels and stronger storms. Scientists need to track shoreline changes, but low-cost, reliable monitoring is surprisingly difficult. Conventional satellite imagery methods have significant limitations—traditional water detection indices struggle with setting appropriate thresholds to accurately separate land from water.

The problem is especially challenging along the Gulf Coast due to its complex geography: dense vegetation, muddy water, and intricate coastlines with highly fragmented coastline like the Mississippi River delta near New Orleans create irregular shorelines that are extremely difficult for computers to identify accurately.

To address these limitations, we developed a machine learning approach that analyzes satellite images and automatically detects shoreline locations over time. Our system is specifically designed to handle the Gulf Coast's challenging conditions, including fragmented landscapes, vegetation, and storm-affected areas.

Our approach distinguishes between permanent erosion, temporary storm-induced changes, and tidal variations—critical for identifying areas experiencing higher erosion rates where coastal managers can provide immediate protection. We're building an easy-to-use online tool allowing users to select specific coastal areas and time periods to visualize shoreline changes. This will help communities, scientists, and policymakers make better decisions about protecting vulnerable coastal areas and planning for the future.