
This dataset contains three decades of mangrove forest distribution data for Madagascar, enabling comprehensive analysis of coastal ecosystem change from 1995 to 2025. All products are provided as georeferenced GeoTIFF rasters at 10-meter spatial resolution.
The dataset includes three raster layers: mangrove extent for 1995, mangrove extent for 2025, and a change detection layer classifying pixels into stable mangrove, mangrove loss, mangrove gain, and non-mangrove categories. The 1995 layer was derived from Landsat imagery using supervised Random Forest classification with optical spectral data, vegetation indices, and topographic variables. The 2025 layer employed the same approach with Sentinel-2 optical data, vegetation indices, and topographic variables, enhanced with Sentinel-1 SAR dual-polarization data for improved classification accuracy.
All layers are delivered in compressed GeoTIFF format, georeferenced to WGS 84, with 10-meter pixel resolution covering all mangrove-bearing areas along the coastline of Madagascar. The change detection layer was produced through pixel-by-pixel temporal comparison, providing spatially explicit information on mangrove dynamics over the 30-year period.