Annual Fractional Cover Maps of Marsh Types across Three HUCs (Sapelo, Doboy and Altamaha) along the Georgia Coast, USA (1995–2024)

This dataset provides annual fractional cover maps of four marsh cover types — freshwater marsh, brackish marsh, salt marsh, and unvegetated area — for three estuaries (Sapelo, Doboy, and Altamaha Hydrologic Unit Codes) along the Georgia coast, USA, spanning 1995 to 2024. Fractional cover was estimated at 30-m resolution using Light Gradient Boosting Machine (LightGBM) models trained on high-resolution marsh type maps derived from National Agriculture Imagery Program (NAIP) imagery and applied to harmonized growing-season Landsat 5 TM, Landsat 7 ETM+, and Landsat 8 OLI surface reflectance data. Input features included spectral band statistics, vegetation indices, and tidal flooding frequency calculated for each growing season (April–October). All Landsat data were harmonized to the Landsat 8 surface reflectance standard through a two-step cross-sensor calibration procedure to ensure temporal consistency. The LightGBM models achieved R² values of 0.70–0.92 and RMSE values of 0.05–0.16 when validated against NAIP-derived fractional cover for 2018. Each annual map represents the fractional cover (0–1) of freshwater marsh, brackish marsh, salt marsh, and unvegetated area for each 30-m pixel. The study area encompasses approximately 44,900 ha of marsh across the three HUCs, which are part of the Georgia Coastal Ecosystems Long Term Ecological Research (GCE-LTER) domain. This dataset supports research on long-term marsh dynamics, marsh type replacement, and coastal ecosystem responses to sea-level rise and freshwater variability.