Accurate estimates of aboveground biomass (AGB) and AGB change are crucial for quantifying how rapid climate change and increasing disturbances impact the carbon cycle in Arctic and Boreal regions (ABRs). Quantifying these impacts is critical for guiding international climate policy and benchmarking ecosystem models. Currently there is a lack of regionally consistent and field-calibrated maps that cover multidecadal time spans with high resolution for the entire North American (NA) ABRs. We here address this gap by producing annual AGB maps with 30-meter resolution covering the entirety of Alaska and Canada from 1984 to 2022.

Our maps are based on synthetic surface reflectance estimated from time series of Landsat Collection 2 data that was temporally segmented with the Continuous Change Detection and Classification algorithm. We compiled roughly 50,000 ground plots with forest measurements and AGB estimates from Canada and Alaska to calibrate Landsat-derived features. We compared performances of decision-tree and deep-learning based models, as well as global and ecoregion-specific calibration approaches. Using the best model, we produced wall-to-wall AGB maps for NA ABRs (approximately 1120 Mha) and derived decadal AGB change maps.

Of the models tested, XGBoost performed best on the hold-out test data (R-squared= 0.76; mean absolute error (MAE) = 37 Mg/ha). The ecoregion-specific models did not outperform the global-calibrated model with an exception for the Arctic Tundra (AT), which therefore was only applied in the AT (R-squared= 0.67; MAE= 2.3 Mg/ha). We observed a 5.25% increase in total AGB stocks between 1984-2022 from 95.94 Pg to 100.98 Pg. Specifically, AGB increased in the AT, Hudson Plain (HP), Taiga (TG), Northern Forests (NF), and Eastern Temperate Forests (ETF) ecoregions by 5%-26%. The rate of AGB gain has been increasing in the AT, HP, and TG over the past decades, while decreasing in the NF and ETF during the last decade. The AGB change map also suggested a net loss of AGB in Marine West Coast Forest. Substantial AGB changes coincided with the footprints of fire, harvest, and other disturbances. By providing almost four decades of annual AGB maps and analyzing detailed AGB change dynamics, this study increases our understanding on how climate change and disturbance impact the carbon cycle in the NA ABRs.