GAMI is an updated dataset providing global forest age distributions for 2010 and 2020 with 100-meter resolution, improving upon the MPI-BGC forest age product. Utilizing machine learning, specifically XGBoost, the estimates are based on over 40,000 forest inventory plots, biomass/height data, remote sensing, and climate data. The dataset incorporates Landsat-based disturbance history and uses multiple XGBoost models with varied hyperparameters to address aleatoric and epistemic uncertainties. Twenty realizations of biomass data with controlled perturbations simulate natural variability, offering robust statistical measures and confidence intervals. Additionally, GAMI provides age class fraction products at different spatial resolutions for custom analyses.

The full description of the data and the updates to version 1.0 (Besnard, 2021, https://doi.org/10.17871/ForestAgeBGI.2021) is provided in the associated data description file.