: Satellite-derived waterlines (SDWs) have become a key source of beach morphological information widely used for coastal monitoring. Nevertheless, the robustness of the extraction methods and the resulting SDW accuracy may be substantially compromised in macrotidal and/or energetic coasts with complex intertidal morphology. This study evaluates the performance of 12 combinations of water indices (MNDWI, NDWI, AWEINSH, SCoWI) and thresholding methods (thresholding at 0, Otsu, and Otsu modified) for SDW extraction from Sentinel-2 imagery at two macrotidal beaches. A robust multi-date validation framework was developed by comparing hundreds (285) of SDWs (2015–2024) against simultaneous reference waterlines photointerpreted from Sentinel-2 and independently benchmarked using very-high-resolution imagery across 255 km of European coastline. The assessment of the SDWs was carried out in two macrotidal beaches of the Atlantic coast in northern Spain (El Puntal and Salinas) with complex intertidal morphology that present differences in terms of the nearshore hydrodynamic conditions, the frequency and extension of the foam, and the sand color. Results show strong site-specific differences: El Puntal exhibits consistently lower offsets, whereas Salinas shows much larger variability and occasional extreme biases (>50 m). Across sites, SCoWI combined with the Otsu modified thresholding method provides the lowest sensitivity to hydrodynamic or morphological conditions at both sites (5.7 m and 7.6 m RMSE in El Puntal and Salinas, respectively), constituting the most robust solution. Correlation analyses reveal that tidal level dominates offset variability in Salinas, especially during low and ebb tides when wet intertidal areas are misclassified as water. To a lesser extent, wave action also introduces uncertainty in some index–threshold combinations. The findings highlight the primacy of the initial water/land segmentation step over subpixel refinement in macrotidal settings and support the development of site-specific flagging strategies to filter unreliable SDWs.
No takes yet. Share an insight, caveat, or question.
Cabezas-Rabadán et al. (2026) studied this question.
Synapse has enriched 5 related papers. Consider them for comparative context: