The Global Carbon Project (GCP) annually compiles an updated global carbon budget that synthesizes the latest estimates of anthropogenic CO2 emissions, land and ocean sinks, and the rate of growth of atmospheric CO2. The remainder between these items, referred to as global carbon budget imbalance, reflects the aggregated uncertainties of the individual component estimates. Accurate quantification of anthropogenic CO2 emissions and their redistribution among the Earth's major carbon pools (atmosphere, oceans, and terrestrial biosphere) is crucial for monitoring progress in climate change mitigation, informing climate policy, and projecting future climate trajectories. Recent research findings point to a significant reduction in this imbalance, suggesting improved scientific understanding of the Earth's carbon cycle.

What is a carbon budget imbalance?

CO2 is released into the atmosphere by fossil fuel combustion, industrial processes and land-use changes. Natural sinks partially offset these emissions, with the oceans absorbing approximately 25% and terrestrial ecosystems 30%, while the remaining 45% accumulates in the atmosphere. At the global level, there is a mass balance between annual emissions, sinks and the annual rate of growth of CO2 in the atmosphere. The residual imbalance, denoted BI, quantifies our inability to close the budget of flows across the Earth’s carbon reservoirs. Mathematically, the imbalance (BI) is defined as: BI = Emissions (fossil fuels + land-use change) – Sinks (ocean + terrestrial + cement carbonation) – Annual rate of growth of CO2 in the atmosphere. This imbalance quantifies the sum of errors in all components of the flow, serving as a metric of data imperfections and gaps in our understanding of the current carbon cycle.

Correction of the rate of growth of CO2 in the atmosphere

Until now, estimates of growth rates derived from marine boundary layer (MBL) bottle measurements have been considered highly accurate, and the imbalance has often been attributed mainly to inaccuracies in other components, particularly in land and oceanic sinks. However, sources emphasize that while MBL measurements are very accurate, errors in estimating the whole-atmosphere growth rate arise from sparse spatiotemporal sampling and methodological assumptions, such as the assumption of instantaneous mixing. The troposphere mixes well throughout the year, while the stratosphere exchanges air with the troposphere slowly, over 2–4 years. Two atmospheric flux inversion models have been used to correct for these inconsistencies: Copernicus Atmosphere Monitoring Service (CAMS) a Jena CarboScope (CS)These models optimize emissions and sinks to match mixing ratio measurements, while incorporating the effects of stratosphere-troposphere exchange (STE) and other atmospheric transport dynamics. These corrections reduced the root mean square (RMS) imbalance from 0.76 PgC yr⁻¹ by up to 25% (to 0.57 PgC yr⁻¹ for CAMS and 0.58 PgC yr⁻¹ for Jena CS)Even in years with significant anomalies, such as the 1991 Mount Pinatubo eruption and the strong El Niño in 1998, which have historically challenged models, the correction resulted in a meaningful reduction in the imbalance extremes. These improvements demonstrate that uncertainties in the rate of atmospheric growth constitute a substantial part of the overall imbalance error and that bottom-up models are more accurate than previously thought.

Improvements in process models and inventories

In addition to growth rate corrections, improvements to the carbon budget components themselves, which the GCP updates annually, have also contributed to reducing the imbalance. An analysis of changes between the 2017 and 2023 GCP reports showed an overall reduction of imbalance by 16% (from 0.91 to 0.76 PgC yr⁻¹)These "bottom-up improvements" include:

• Increasing the size of the model file: For example, the number of dynamic global vegetation models (DGVMs) for land subduction increased from 15 to 20 and the number of ocean biogeochemical models (OBGCMs) from 8 to 10.
• Improving forcing data and model resolution: Ocean models in the GCP 2023 report included models with 75 vertical layers, compared to a maximum of 51 layers in the GCP 2017 report.
• Inclusion of new carbon cycle processes: For example, better accounting for the impact of diffuse radiation on plant productivity, simulating fire ignition and suppression, including peat fires, wood harvesting and nitrogen fertilization. Cement carbonation has also been included in the budget since the GCP 2020 report, which represents a significant CO2 sink.

Total reduction and its significance Combining corrections to the whole-atmosphere growth rate with improvements to the component models between the GCP 2017 and GCP 2023 reports, significant reduction in RMS imbalance from 0.91 PgC yr⁻¹ to 0.57 PgC yr⁻¹, which represents 37% imbalance reductionThis reduced imbalance, which approaches the range of uncertainties in annual fossil fuel emissions (0.5 PgC yr⁻¹) or emissions reductions during the COVID-19 pandemic (0.55 PgC yr⁻¹), suggests that the global carbon budget imbalance may become more pronounced in the future. an increasingly powerful tool for tracking large changes in anthropogenic emissionsOverall, these results indicate improved understanding and increasingly accurate representation of the Earth's carbon cycle through process models and emissions inventories, which should improve future climate projections and monitoring of anthropogenic emissions. The global assessment of the Paris Agreement urgently requires scientific refinement of the understanding of the carbon cycle for effective policymaking. This progress represents an important step forward. JRi