A robust and fully automated morphology-based approach for revealing changes of China’s coastal tidal flats from 2016 to 2024

Accurate and consistent monitoring of tidal flats is fundamental for understanding coastal dynamics and supporting sustainable coastal management. However, existing methods still face major limitations, such as incomplete removal of interference from coastal aquaculture ponds and reclaimed areas, frequent misclassification of spectrally similar wetlands, and heavy reliance on labor-intensive post-classification corrections, which hinder their scalability and automation at large spatial scales. Here, we proposed a robust and fully automated hierarchical classification framework (TF-LMMO) that integrates morphological operation, Laplacian edge detection, spectral index composite, and dense time series Sentinel-2 imagery to delineate coastal tidal flats across China. This approach markedly minimizes the misclassification of aquaculture ponds and intermittent wetlands as tidal flats, without relying on auxiliary datasets or manual corrections. Applying TF-LMMO, we generated 10 m resolution national tidal flat maps for 2016 and 2024, achieving overall accuracies above 95% (Kappa > 0.91). Results indicated that China’s tidal flats increased from 7,622.48 km2 in 2016 to 8,214.95 km2 in 2024, representing a net increase of 592.47 km2. The new tidal flat dataset, with extremely high accuracy, provides a robust baseline for quantifying coastal geomorphic change and offers critical support for evaluating the effectiveness of ecological restoration, blue carbon accounting, and progress toward coastal sustainability goals under accelerating anthropogenic and climatic pressures.