A 30-m, Multi-decadal Mangrove Cover Fraction Dataset for China (1990-2023)

Methods
This dataset was generated using a supervised machine learning framework on the Google Earth Engine platform. The primary input data consisted of time-series Landsat surface reflectance imagery (1990-2023), from which annual composites based on 32 spectral features were created for each target year. A Random Forest model was trained and validated using 1,711 high-quality reference Mangrove Cover Fraction (MCF) samples. These reference samples were generated via detailed photo-interpretation of very-high-resolution (VHR) and unmanned aerial vehicle (UAV) imagery, guided by a stratified random sampling design to ensure representativeness. The final China Mangrove Cover Fraction (CMCF) was produced by applying the cross-validated Random Forest model to the annual Landsat composites for each of the six target years (1990, 2000, 2010, 2015, 2020, and 2023) to map sub-pixel MCF at 30 m resolution. The spatial extent of the mapping was constrained by the CAS_Mangroves v2.0 dataset.
README: Description of the data and file structure
This dataset provides estimates of mangrove cover fraction (MCF) across China’s coastal regions for seven distinct years between 1990 and 2023. The data is provided at a spatial resolution of 30 meters.

The dataset consists of 6 GeoTIFF (.tif) files. The file naming convention is CMCF_YYYY.tif, where YYYY represents the year of the observation.

Access information
Other publicly accessible locations of the data:

CMCF_1990.tif: Mangrove cover fraction map for the year 1990.
CMCF_2000.tif: Mangrove cover fraction map for the year 2000.
CMCF_2010.tif: Mangrove cover fraction map for the year 2010.
CMCF_2015.tif: Mangrove cover fraction map for the year 2015.
CMCF_2020.tif: Mangrove cover fraction map for the year 2020.
CMCF_2023.tif: Mangrove cover fraction map for the year 2023.
Data was derived from the following sources:

Format: GeoTIFF
Spatial Resolution: 30 meters
Coordinate Reference System (CRS): WGS 84
Pixel Values: The values in each pixel represent the percentage of mangrove cover. The values range from 0 to 100, where 0 indicates 0% mangrove cover and 100 indicates 100% mangrove cover.
NoData Value: ‘Nodata’