A 10m Aboveground Biomass Density Dataset for France, Italy, and Spain in 2020

This dataset provides 10m aboveground biomass density (AGBD) maps for France, Italy, and Spain for the year 2020. AGBD is expressed in Mg ha⁻¹. The dataset was generated using multi-product consensus samples derived from five existing AGBD products, including GEDI L4A, Pan-European forest maps, ESA CCI AGB, Forest biomass map of EU27 for 2020, and maps of forest biomass and increment in Europe, and an optimized machine-learning modelling framework. The mapping framework integrates multi-source remote sensing, topographic, climatic, and ancillary variables, including Sentinel-1, Sentinel-2, PALSAR, digital elevation models, growing-season hydroclimatic variables, canopy height, and land-cover information. Training samples were selected using pixel-level consistency among multiple AGBD products to reduce product-specific uncertainty and provide relatively reliable reference samples for model training. The final AGBD maps were produced at 10 m spatial resolution using an optimized XGBoost model. The dataset includes gridded AGBD estimates and associated uncertainty layers. The main AGBD layer represents the predicted mean vegetation AGBD, while the uncertainty layer represents the half-width of the 90% prediction interval, calculated as (q95 − q05) / 2. Users should note that the dataset represents modelled AGBD for 2020 and is not a direct field inventory product. The estimates are dependent on the input remote-sensing products, training-sample selection strategy, model structure, and spatial resolution. Caution is recommended when using the dataset for local-scale validation, fine-scale management decisions, or areas where independent field observations are unavailable.