An improved approach to estimate the natural land carbon sink

The natural land carbon sink (SLAND) absorbs roughly 25-30% of anthropogenic CO2 emissions, playing a critical role in offsetting climate warming. In the Global Carbon Budget (GCB), SLAND is estimated using simulations that isolate the environmental response of undisturbed land. However, these simulations assume fixed pre-industrial land cover, failing to represent today’s human-altered landscapes. This leads to a systematic overestimation of forest area, and thus sink strength, in regions heavily altered by human activity. We present a new process-based approach to estimate SLAND using Dynamic Global Vegetation Models. Our corrected estimate reduces SLAND by 0.42±0.20 PgC yr-1 over 2014-2023, from 3.19±0.86 to 2.77±0.69 PgC yr-1 (13% smaller). We incorporate the new SLAND with emissions from land-use change from bookkeeping models, to estimate a net land sink of 1.44±0.98 PgC yr-1, which aligns closely with atmospheric inversion constraints. This downward revision of SLAND introduces a positive imbalance in the GCB, with sources now exceeding sinks. This suggests that other components of the budget, such as the ocean sink, may also be misestimated. Our findings motivate renewed scrutiny of global carbon sink partitioning and the assumptions underpinning carbon budget assessments.