2026-10-14T14:23:10Z https://www.scidb.cn/oai

10.57760/sciencedb.31545 2025-11-13T11:02:04Z

2025-11-13 A 10-m Vector Dataset of Mangrove Species Distribution in China doi:10.57760/sciencedb.31545 en This dataset provides a 10-m spatial-resolution vector map of mangrove species distribution along the Chinese coast, produced under an Environment-Enhanced Remote Sensing (EERS) framework. EERS performs feature-level fusion of single-date Sentinel-2 spectral and textural features with long-term climatological and oceanographic variables from WorldClim and Bio-ORACLE. Spectral separability among species was quantified across 36 monthly composites (2018–2020) using the Jeffries–Matusita distance, identifying June 2019 as the most discriminative window. Lightweight, interpretable classifiers were evaluated with spatially blocked validation to map 26 mangrove species merged into 18 classes. Relative to a remote-sensing-only baseline, overall precision increased by 16.7% and AUC-ROC improved from 0.803 to 0.900. Mapping results indicate dominance by Avicennia marina (31.6%), Aegiceras corniculatum (17.0%), and Kandelia obovata (16.8%). Introduced species (Sonneratia apetala and Laguncularia racemosa) together account for ~14%, while threatened taxa such as Nypa fruticans cover <0.05%. Delivered as a vector dataset with an effective 10-m mapping resolution, it offers a current baseline for biodiversity assessment, invasion monitoring, and climate-informed coastal ecosystem management. Mangrove; Species distribution; 10-m resolution; Environment-Enhanced Remote Sensing (EERS); Sentinel-2; WorldClim; Bio-ORACLE; Jeffries–Matusita distance; Interpretable machine learning Kan Zhiyi Hu Huijuan Yu Weiwei Chen Guangcheng Chen Bin RESTRICTED dataset http://www.doi.org/10.1007/s41651-026-00254-7 Science Data Bank