An efficient and temporally-transferable method for tidal flat elevation inversion based on homogeneous inundation frequency

Accurate long-term tidal flat elevation mapping is essential for understanding coastal morphodynamics, assessing hazard resilience, and supporting coastal management. However, previous approaches typically rely on separately modeling for each temporal snapshot, leading to inconsistent results over time. To address this limitation, this study proposes an efficient long-term tidal flat elevation inversion method based on inter-annual statistically homogeneous inundation frequency. A phase–tide–based weighted inversion algorithm is developed to correct sampling-induced bias and generate statistically homogeneous annual inundation frequency (AIF) maps, thereby stabilizing the AIF–elevation relationship and enabling a temporally-transferable inversion model for long-term reconstruction within a given region. To further ensure efficiency, a robust rule-based land–water segmentation strategy is integrated. The method was validated across five representative coastal regions in China (Yellow River Delta, Subei Bank, Yangtze River Estuary, Sansha Bay, and Lianzhou Bay), yielding integrated RMSE values of 0.20 m to 0.63 m. Notably, high accuracy was maintained across years using a unified inversion framework without year-specific recalibration, proving the method's strong temporal robustness and computational efficiency. By transforming tidal flat inversion from discrete year-by-year calibration toward statistically stabilized and transferable modeling, the proposed method provides an operationally feasible solution for long-term coastal topographic monitoring.