National-scale satellite monitoring of tidal flat dynamics along China's coast

Tidal flats occupy the shifting boundary between land and sea, where tides, sediment supply, and human use continually reshape coastal landscapes. These unvegetated wetlands support habitat conservation, shoreline management, and restoration planning, but their long-term trajectories remain difficult to monitor. Existing large-scale tidal flat products often contain unstable time-series signals and misclassification errors, especially in estuarine and deltaic regions where turbid waters, tidal-stage differences, and vegetated wetland boundaries complicate mapping. To provide a consistent national record, we developed S2TidalFlats, an automatic and interpretable satellite-derived framework for 30 m tidal flat mapping along China’s coast from 1990 to 2020. Using Google Earth Engine, we processed 10,012 Landsat 5 and 8 images and applied a non-overlapping three-year compositing strategy to capture low-tide conditions while filtering invalid and anomalous pixels. The framework achieved an overall accuracy of 97.94%. Our results show that China’s tidal flat area declined by 36.9%, from 10,887.26 km² in 1990 to 6,869.20 km² in 2020. The loss rate slowed markedly after 2005, while the national distribution remained dominated by a limited number of large coastal provinces. By reducing confusion with turbid waters and abnormal pixels, S2TidalFlats provides a consistent basis for tracing multi-decadal tidal flat dynamics across diverse coastal settings. This national dataset supports assessments of coastal wetland change, long-term human disturbance, and restoration priorities along China’s rapidly changing coast.