Coastal change analysis using satellite imagery and machine learning

This report describes the development and application of a shoreline extraction workflow using Sentinel-2 satellite imagery. The study focuses on identifying shoreline positions through remote sensing techniques and processing the resulting data into a structured and usable geospatial format for coastal analysis.
The methodology includes data acquisition from Sentinel-2 imagery, followed by preprocessing steps such as cloud masking, scene selection, and band selection to improve water–land discrimination. Shoreline detection is performed using spectral-based classification methods to separate land and water pixels. Post-processing techniques are then applied to refine the extracted shoreline, including noise reduction, smoothing, and organisation of outputs into consistent spatial datasets.
The workflow is implemented using geospatial tools within a cloud-based processing environment and GIS software to enable efficient and reproducible analysis. The resulting shoreline datasets support the analysis of coastal features and changes over time and demonstrate the potential of Earth observation data for shoreline monitoring and geospatial applications.