A Two‐Step Approach to Extracting Sandy Beaches Through Integrating Spatial Semantic Information From Open‐Source Geospatial Datasets

Abstract

Sandy beaches, widespread in coastal areas, provide valuable economic and ecological benefits. However, a substantial proportion of sandy beaches are undergoing erosion caused by marine disasters and human interventions. Advanced sandy beach extraction (SBE) approaches are indispensable to sandy beach observation and protection. This study proposes a novel two‐step SBE approach using remote sensing images and digital elevation models. First, sea–land segmentation is performed as a preparatory work. We model sea–land segmentation as an optimization problem and develop an improved NSGA‐II, SCS‐NSGA‐II, to solve it, considering both topographical and spectral costs. Second, a region growing algorithm is applied to generate the final sandy beach extents. The assessment results verify that (1) our approach effectively reduces the false‐positive rate, thereby resulting in more accurate SBE results compared with existing approaches. (2) SCS‐NSGA‐II ensures the diversity of individuals in spatial patterns and exhibits superior performance compared with NSGA‐II in this task.