AGB for north-east China from 2013-2020

Accurate monitoring of aboveground biomass (AGB) is essential for forest carbon accounting and climate change mitigation, yet remains challenging in heterogeneous boreal forests. This study developed a novel framework that synergistically integrates multi-source satellite data—optical (Sentinel-2/Landsat-8), Synthetic Aperture Radar (Sentinel-1/Radarsat-2), and spaceborne altimetry (ICESat-2), to significantly improve AGB estimation in Northeast China. This study incorporates a forest-type-specific (coniferous vs. broadleaf) modeling to describe the structure of the forests. Vegetation was first delineated using NDVI and refined with a canopy height threshold (>1.2 m). A Support Vector Machine classifier, utilizing spectral, backscatter, and textural features, generated a forest-type classification map. In mixed forests, AGB was computed as the sum of AGB according to different forest type. Among four evaluated regression algorithms, Support Vector Regression (SVR) delivered optimal performance, achieving an R² of 0.76 and an RMSE of 18.48 Mg/ha against field measurements, with a mean relative deviation below 10%. Time-series analysis (2013–2020) revealed a regional AGB increase from 157.13 to 192.23 Mg/ha; however, growth slowed significantly during 2016–2018. Crucially, time-series analysis uncovered a significant growth plateau during 2016–2018. This deceleration was strongly correlated with a tripling of burned area, pinpointing fire disturbance as a key driver of carbon stock variability. This work demonstrates that integrating multi-sensor data with ecological realism through forest-type differentiation provides a robust, scalable approach for precise carbon monitoring, offering critical support for forest management and global carbon cycle research.