Automated Coastline Mapping Using Sentinel 2-based NDVI on Google Earth Engine: A Decision Support Tool for Diachronic Coastal Monitoring

Abstract. Coastal zone monitoring requires efficient and reproducible methods for coastline extraction across multiple spatial and temporal scales. In this study, we present NDVICoast, an automated decision-support tool developed on Google Earth Engine (GEE) for diachronic coastline mapping using Sentinel-2 imagery. The method leverages the Normalized Difference Vegetation Index (NDVI) as a proxy for extracting the vegetation boundary, which serves as a reliable indicator of coastline position in vegetated coastal environments. The tool automates the complete workflow from image acquisition and preprocessing to coastline extraction at two different dates. The method was validated at two sites in Québec with distinct coastal geomorphologies. At the Saguenay Fjord, the automated extraction achieved a Root Mean Square Error (RMSE) of 4.20 meters compared to in situ data, while at Baie-Trinité, the RMSE was 5.12 meters when compared to a digitized reference coastline. NDVICoast enables rapid processing of large temporal datasets, facilitating systematic coastal change detection. Results demonstrate the tool's capacity to support coastal management decisions through consistent, repeatable, and cost-effective coastline monitoring across multiple time periods.