Code, predictor metadata, and source information for: A long-term (2000–2022), high-resolution (0.005°) aboveground biomass dataset of global grasslands

This repository provides the analysis code and source metadata used to construct and validate the global grassland aboveground biomass (AGB) dataset described in the associated publication. It includes:• Python scripts for random forest model training and AGB map generation• MATLAB scripts for spatial and temporal analyses• A compiled table summarizing all site-level field AGB data sources and access information These materials support transparency and reproducibility of the modeling and validation workflow. Large raster predictors and outputs are not included due to file size constraints.