Wall-to-Wall 30 m Above-Ground Biomass Density (AGBD) Dataset for Pakistan's Forests (2020)

Developed the first baseline wall-to-wall 30 m Above-Ground Biomass Density (AGBD) dataset for Pakistan’s forests for the year 2020 using multi-source remote sensing and national forest inventory data. Integrated Landsat-8 optical imagery, ALOS-PALSAR L-band SAR backscatter, SRTM elevation data, and 1,258 REDD+ field inventory plots to model forest biomass distribution across Pakistan. Applied and evaluated multiple machine learning approaches, with the Classification and Regression Tree (CART) model achieving the best performance (R² = 0.91, RMSE = 32.24 Mg/ha, MAE = 18.35 Mg/ha). Generated a high-resolution national AGBD map showing strong spatial variability in forest biomass, with values ranging from 0.36 to 428.42 Mg/ha, a mean of 91.54 Mg/ha, and an estimated total biomass stock of 4.75 Pg. This dataset includes estimated AGBD in megagrams per hectare (Mg/ha) and associated uncertainty in percentage (%), all at 30 m spatial resolution, in GeoTIFF format.