Fusing Sentinel-1 and Sentinel-2 Data With Machine Learning for Large-Scale Detection of Coastal Erosion and Accretion

It is essential to analyze changes along the coast accurately to manage and develop sensitive shoreline areas in a way that lasts. This work presents a comprehensive multi-sensor data fusion method for detecting coastal erosion and accretion on China's dynamic southeastern coast on a large scale and in an automated manner. We overcome the limitations of single-sensor approaches and changing environmental conditions by combining high-resolution Sentinel-1 Synthetic Aperture Radar (SAR) and Sentinel-2 optical images within a single geospatial analytical framework. We employed advanced image preprocessing, robust coastline delineation using spectral indices and edge detection, and machine learning-based change analysis on multi-temporal datasets spanning 2020-2022. The combined approach enabled the accurate recording of shoreline movement and the measurement of both erosion losses and accretion gains over one year and multiple years. The results indicate that the coastal land area has expanded significantly, with accretion processes consistently outpacing erosion during the research period. Spatial analysis revealed areas of change that have been ongoing for a long time, which are crucial for mitigating hazards and informed conservation efforts. The suggested process demonstrates the value of combining sensors in a way that facilitates monitoring of the coastal zone and enables the development of practical, scalable solutions for the remote sensing of rapidly changing coastal systems.