Assessing temporal trends of forest aboveground biomass density in Japan from 2009 to 2018 under disturbance regimes using multisource remote sensing data

Abstract Understanding long-term aboveground biomass density (AGBD) trends under a changing climate is essential for quantifying forest carbon dynamics. However, management activities such as harvesting and thinning can obscure climate-driven signals. In this study, we developed a national framework to estimate annual AGBD across Japan by aggregating Global Ecosystem Dynamics Investigation (GEDI) footprints into 1000 m grid cells and integrating them with multisource satellite and topographic predictors. We first trained a series of models using GEDI reference data aggregated under different minimum footprint requirements across grid cells. By applying the best performing model, which was obtained when each grid cell contained at least 28 GEDI footprints and showed the highest agreement with National Forest Inventory (NFI) data ( r = 0.87, RMSE = 31.73 Mg ha −1 , RRMSE = 0.18, bias = 25.52 Mg ha −1 ), we generated annual AGBD maps from 2009 to 2018. Using these estimates, we quantified a national mean AGBD increase of 0.21 Mg ha −1 per year, with undisturbed forests showing a stronger increase of 0.43 Mg ha −1 per year, while disturbed areas exhibited a decline of 0.20 Mg ha −1 per year. By integrating GEDI observations, multisource remote sensing data, and annual forest disturbance maps, we successfully characterized long-term AGBD dynamics under different disturbance regimes and revealed distinct climate-driven and disturbance-driven trajectories.