CCAV-10m: An Annual Spatiotemporal China's Coastal Wetland Vegetation Dataset Integrating Sentinel-1/2 Observations via Deep Learning

The dataset is derived from multi-source remote sensing imagery of Sentinel-1 SAR and Sentinel-2 MSI, integrating optical and radar information through a deep learning–based classification framework. CCAV-10m covers the entire eastern coastal zone of China at a spatial resolution of 10 m × 10 m and an annual temporal resolution, comprising eight yearly maps (2016–2023).

It distinguishes six dominant coastal wetland plant types: S. alterniflora, Suaeda spp., P. australis, S. mariqueter, mangroves, and T. chinensis. Model validation indicates a high classification accuracy, with an overall accuracy (OA) of 0.916 and a Kappa coefficient of 0.898, demonstrating stable and reliable species discrimination across years. Each annual product is provided in GeoTIFF raster format with spatial reference information and vegetation classification results, compatible with common GIS software (e.g., ArcGIS, QGIS) and remote sensing platforms (e.g., Google Earth Engine).