National scale sub-meter mangrove mapping using an augmented border training sample method

This study proposes a novel Semi-automatic Sub-meter Mapping Method (SSMM), which incorporates an innovative automated sample collection strategy to obtain sufficient and spatially representative training samples. The method integrates Sentinel-2 and Google Earth imagery and selects nine key features to enhance the spectral separability between mangroves and other land-cover types. Based on this framework, the first Large-scale Sub-meter Mangrove Map (LSMM) was produced. Based on this map, the total mangrove area in China is estimated at 28,253 ha, and 40,035 mangrove patches are identified. The LSMM achieved an overall accuracy of 97.56% and an F1 score of 0.98, demonstrating significant advantages in boundary delineation and in detecting fragmented and small patches. The LSMM provides reliable fine-scale spatial information to support national-scale mangrove monitoring and coastal ecosystem management.