Junjie Liu

Accurate assessment of anthropogenic carbon dioxide (CO2) emissions and their partitioning among the atmosphere, ocean, and terrestrial biosphere in a changing climate is critical to better understand the global carbon cycle, supporting the development of climate policies, and project future climate change. Here we describe and synthesise datasets and methodologies to quantify the five major components of the global carbon budget and their uncertainties. Fossil CO2 emissions (EFOS) are based on energy and cement production data. Emissions from land-use change (ELUC) are estimated by bookkeeping models based on land-use data. The global atmospheric CO2 growth rate (GATM) is computed from changes in concentration measured at surface stations. The global net uptake of CO2 by the ocean (SOCEAN) is estimated with global ocean biogeochemistry models and observation-based fCO2-products. The global net uptake of CO2 by the land (SLAND) is estimated with dynamic global vegetation models. Additional lines of evidence are provided by atmospheric inversions, atmospheric oxygen measurements, ocean interior observation-based estimates, and Earth System Models. The sum of all sources and sinks results in the carbon budget imbalance (BIM), a measure of imperfect data and incomplete understanding of the contemporary carbon cycle. All uncertainties are reported as ±1σ. For the year 2025, EFOS increased by 1.2% relative to 2024, with fossil emissions at 10.5 ± 0.5 GtC yr-1 (including the cement carbonation sink, 0.2 GtC yr-1), ELUC was 1.2 ± 0.7 GtC yr-1, for total anthropogenic CO2 emissions of 11.7 ± 0.9 GtC yr-1 (42.8 ± 3.2 GtCO2 yr-1). Also, for 2025, GATM was 4.4 ± 0.2 GtC yr-1 (2.06 ± 0.1 ppm yr-1), 3.6 GtC lower than the record high 2024 growth rate. SOCEAN was 3.2 ± 0.7 GtC yr-1 and SLAND was 3.6 ± 0.9 GtC yr-1, with a small BIM (0.5 GtC yr-1). The global atmospheric CO2 concentration averaged over 2025 reached 425.6 ± 0.1 ppm. Preliminary data for 2026 suggest a decrease in EFOS relative to 2025 of 0.40% (–1.5% to +0.6%) globally, and atmospheric CO2 concentration increasing by 2.2 ppm reaching 427.7 ppm, 54% above the pre-industrial level (around 278 ppm in 1750). Overall, the mean and trend in the components of the global carbon budget are consistently estimated over the period 1959–2025, with a small positive overall budget imbalance, (0.6 GtC yr-1 averaged over the period 1959–2025), although discrepancies of up to around 1 GtC yr-1 persist for the representation of annual to decadal variability in CO2 fluxes. Comparison of estimates from multiple approaches and observations shows: (1) a persistent large uncertainty in the estimate of land-use change emissions, (2) a low agreement between the different estimates on the magnitude of the land CO2 flux, particularly in the northern extra-tropics, and (3) a discrepancy between the different estimates on the mean ocean sink and its trend. This living data update documents changes in methods and datasets applied to this most-recent global carbon budget as well as evolving community understanding of the global carbon cycle. The data presented in this work are available at https://doi.org/10.18160/GCP-2026 (Friedlingstein et al., 2026b).

NASA's Orbiting Carbon Observatory‐2 (OCO‐2) and OCO‐3 have been observing Earth's atmosphere from space since September 2014 and August 2019, respectively. The version 11 (v11) Atmospheric Carbon Observations from Space (ACOS) Level 2 full physics (L2FP) retrieval algorithm has been used to provide estimates of the column‐averaged dry‐air mole fraction of carbon dioxide () from OCO measurements of surface reflected near‐infrared solar radiation. In this work, updates to the ACOS v11 retrieval algorithm are discussed, and the quality filtering and bias correction of the products are described. estimates from OCO‐2 and OCO‐3 are compared to Total Carbon Column Observing Network (TCCON) observations and to atmospheric inversion models, which serve as truth‐proxies for . In addition, a direct comparison of collocated soundings between OCO‐2 and OCO‐3 is performed. Variations in v11 over small areas show improvement in agreement between actual and theoretical uncertainties, suggesting a reduction in systematic errors compared to v10, especially for OCO‐3. There is reduced variance in the new v11 product relative to truth‐proxies, including less divergence in the values with time against models. The collocations of observations between OCO‐2 and OCO‐3 indicate an overall mean difference of <0.2 ppm across the overlapping time range. Furthermore, OCO‐2 and OCO‐3 agree better with each other than they do with the validation sources, suggesting a well conditioned and unified set of concentrations that can be used concurrently for carbon cycle studies to provide a benchmark record spanning September 2014 through August 2026.

Interannual atmospheric CO₂ growth variability reflects changes in land and ocean carbon exchange, but the role of fluxes from regional disturbances such as fire remains unclear. Here we show that regional annual fire-carbon anomalies shifted from partial cancellation in 2001–2010 to reinforcement in 2015–2024. Global fire-emission variance increased about seven-fold despite nearly unchanged mean emissions, with cross-region covariance accounting for 84% of the increase and seasonally separated boreal and tropical anomalies accounting for 44% of the covariance increase. At the same time, global fire anomalies became strongly correlated with whole-atmosphere CO₂-growth anomalies (R² from 0.06 to 0.83). Subtracting fire emissions reduces the year-to-year variance of the atmospheric CO₂ growth rate by 57% in the recent decade, compared with 6% in the earlier one. More generally, our analysis shows that a global Earth-system quantity can become far more variable mainly because its regional contributors shift from cancelling to reinforcing one another, without growing larger individually.

Atmospheric transport uncertainty remains a major source of error in carbon flux estimations, while computational constraints limits the spatial resolution of transport models. Many studies use coarse-resolution GEOS-Chem simulations (4° × 5°) to reduce computational expense, potentially increasing transport errors. The GEOS-Chem High Performance (GCHP) framework enables efficient high-resolution simulations through multiprocessor orchestration. Here, we evaluate the sensitivity of simulated atmospheric CO₂ distributions to model resolution and implications for carbon flux estimation. Model performance improves substantially from C24 (~4° × 5°) to C90 (~1°), with smaller improvements from C90 to C180 (~0.5°). Higher resolution strengthens interhemispheric transport and reduces excessive CO₂ accumulation in the Northern Hemisphere (NH). A two-box model shows that C180 transport implies a 17% weaker NH natural carbon sink than C24, suggesting that the widely-used 4° × 5° configuration overestimates NH land carbon uptake and may partly explain the stronger atmospheric CO2-constrained NH sink in recent studies.

The 2023/24 El Niño strongly reduced land carbon uptake, but the persistence of this anomaly after surface cooling remains uncertain. Here we quantify the July 2024–June 2025 global CO ₂ budget using low‐latency fossil emission estimates, three DGVMs, machine learning ocean flux emulators and OCO‐2‐constrained atmospheric inversions. The atmospheric CO ₂ growth rate was 2.62 ± 0.08 ppm yr ⁻¹ , 6.5% above the 2013–2022 July‐to‐June mean. DGVMs estimate that the net land sink was 1.32 ± 0.19 GtC yr ⁻¹ weaker than the 2015–2022 July‐to‐June mean, whereas combining bottom‐up and top‐down constraints gives a smaller deficit of 0.49 GtC yr ⁻¹ . The annual anomaly is dominated by late‐2024 land carbon losses. Early‐2025 recovery, however, is method‐dependent: DGVMs retain a weak annual land‐sink deficit, while all inversions indicate fluxes close to the reference mean and a stronger‐than‐normal northern sink in late spring 2025. Ocean uptake shows no global weakening. A statistical decomposition links global land flux variability mainly to temperature, with terrestrial water storage contributing more strongly at regional scales. These results identify Northern Hemisphere land‐sink recovery as a central uncertainty in low‐latency carbon‐budget assessments.

Atmospheric inversions provide surface CO ₂ flux estimates based on in situ observed atmospheric CO ₂ mole fractions or satellite‐based column average CO ₂ (XCO ₂ ). Here, we provide a detailed assessment of 14 atmospheric CO ₂ inversions included in the Global Carbon Budget (GCB2024). We develop tools to further assess and use these inversions in global and regional carbon cycle studies including the GCB and the REgional Carbon Cycle Assessment and Processes (RECCAP2) initiative. We show that the global atmospheric CO ₂ growth rate and its interannual variability are reproduced well by all inversions. In contrast to bottom‐up models, inversions provide carbon flux estimates directly constrained by observations. Our ensemble mean estimates of the global sinks for the period 2015–2023 are −1.410.55 PgC yr ⁻¹ for the net land sink (including land‐use change emissions) and −2.970.55 PgC yr ⁻¹ for the global net ocean sink (uncertainties reported as across inversions; estimates include fossil fuel and river flux adjustments). On regional scales, we find significant spread in flux estimates between inversions across regions, and we present criteria and metrics to derive flux‐observation relation constraints or subselect ensembles of inversions based on independent observations. Furthermore, we use the atmospheric inversions to assess the atmospheric growth rate of CO ₂ . We show that the factor used to convert annual observation‐based growth rates to net fluxes is variable over time as a result of atmospheric mixing. Finally, we propose guidelines on how to use the inverse results in global and regional carbon budget studies by the wider carbon cycle community.

Robust information on the spatial distribution of global carbon fluxes is required to project the future trajectory of carbon-climate feedback effects and atmospheric CO ₂ concentrations. Estimates of the latitudinal partitioning of carbon fluxes from top-down atmospheric CO ₂ inverse models currently diverge widely, because of methodological limitations or systematic biases in models or observations. We use airborne CO ₂ observations from the NASA Atmospheric Tomography Mission to evaluate and refine inverse model estimates from the Orbiting Carbon Observatory version 10 Model Intercomparison Project of total CO ₂ exchange for the two-year period of June 2016–May 2018. Applying emergent concentration-flux relationships as constraints reduces zonal total flux uncertainties by 46 to 56% relative to the full v10 MIP ensemble and by 17 to 28% relative to the subset excluding satellite observations over ocean. Subtracting independent estimates of fossil-fuel emissions and air-sea gas exchange results in residual land fluxes with a large northern extratropical sink, a small southern extratropical sink, and a small tropical source. The airborne-derived tropical land source disagrees with a large tropical land sink from process-based terrestrial models combined with estimates of land use emissions and river fluxes, representing an important challenge for our understanding of the global carbon cycle. The large implied northern extratropical sink can be explained either by underestimated land uptake by process models or a combination of process model bias and overestimated fossil fuel emissions.

The atmospheric CO₂ growth rate is the observational anchor of the global carbon budget and is often treated as a well-constrained quantity. Here we show that the widely used NOAA marine boundary layer (MBL) growth rate contains a time-varying whole-atmosphere representativeness error linked to network coverage changes. From 1991 to 2024, changes in network coverage and anomalous atmospheric CO₂ growth state explain 73% of the discrepancy between the NOAA MBL estimate and whole-atmosphere growth rates inferred from atmospheric flux inversions. The discrepancy reaches 0.39 ppm yr⁻¹, or 0.83 PgC yr⁻¹, in 2024—a year of record atmospheric growth and the most negative carbon-budget imbalance on record. We compare NOAA MBL against two complementary whole-atmosphere references: data-driven OCO-2/GOSAT satellite growth rates and the median of six in situ atmospheric inversions. The satellite estimate tracks the inversion median, and both depart from NOAA MBL during anomaly years, with an satellite-estimate RMSE of 0.07 ppm yr⁻¹ against inversions, compared with 0.21 ppm yr⁻¹ for NOAA MBL. Part of the apparent carbon-budget imbalance therefore reflects sampling errors in the atmospheric term, and whole-atmosphere growth rates provide a more robust basis for attributing interannual carbon-cycle variability.

Accurate assessment of anthropogenic carbon dioxide (CO ₂ ) emissions and their redistribution among the atmosphere, ocean, and terrestrial biosphere in a changing climate is critical to better understand the global carbon cycle, support the development of climate policies, and project future climate change. Here we describe and synthesise datasets and methodologies to quantify the five major components of the global carbon budget and their uncertainties. Fossil CO ₂ emissions ( E _FOS ) are based on energy and cement production data. Emissions from land-use change ( E _LUC ) are estimated by bookkeeping models based on land-use data. The global atmospheric CO ₂ growth rate ( G _ATM ) is computed from changes in concentration measured at surface stations. The global net uptake of CO ₂ by the ocean ( S _OCEAN ) is estimated with global ocean biogeochemistry models and observation-based f CO ₂ -products. The global net uptake of CO ₂ by the land ( S _LAND ) is estimated with dynamic global vegetation models. Additional lines of evidence are provided by atmospheric inversions, atmospheric oxygen measurements, ocean interior observation-based estimates, and Earth System Models. This year, we introduced corrections on the E _LUC , S _OCEAN and S _LAND estimates. The sum of all sources and sinks results in the carbon budget imbalance ( B _IM ), a measure of imperfect data and incomplete understanding of the contemporary carbon cycle. All uncertainties are reported as ± 1 σ . For the year 2024, E _FOS increased by 1.1 % relative to 2023, with fossil emissions at 10.3 ± 0.5 GtC yr ⁻¹ (including the cement carbonation sink, 0.2 GtC yr ⁻¹ ), E _LUC was 1.3 ± 0.7 GtC yr ⁻¹ , for total anthropogenic CO ₂ emissions of 11.6 ± 0.9 GtC yr ⁻¹ (42.4 ± 3.2 GtCO ₂ yr ⁻¹ ). Also, for 2024, G _ATM was 7.9 ± 0.2 GtC yr ⁻¹ (3.73 ± 0.1 ppm yr ⁻¹ ), 2.2 GtC above the 2023 growth rate. S _OCEAN was 3.4 ± 0.4 GtC yr ⁻¹ and S _LAND was 1.9 ± 1.1 GtC yr ⁻¹ , leaving a large negative B _IM (−1.7 GtC yr ⁻¹ ), suggesting that the total sink or G _ATM is strongly overestimated in 2024. The global atmospheric CO ₂ concentration averaged over 2024 reached 422.8 ± 0.1 ppm. Preliminary data for 2025 suggest an increase in E _FOS relative to 2024 of +1.0 % (0.2 % to 1.7 %) globally, and atmospheric CO ₂ concentration increasing by 2.1 ppm reaching 425.6 ppm, 53 % above the pre-industrial level (around 278 ppm in 1750). Overall, the mean and trend in the components of the global carbon budget are consistently estimated over the period 1959–2024, with a near-zero overall budget imbalance, although discrepancies of up to around 1 GtC yr ⁻¹ persist for the representation of annual to decadal variability in CO ₂ fluxes. Comparison of estimates from multiple approaches and observations shows: (1) a persistent large uncertainty in the estimate of land-use change emissions, (2) a low agreement between the different methods on the magnitude of the land CO ₂ flux in the northern extra-tropics, and (3) a discrepancy between the different methods on the mean ocean sink. This living data update documents changes in methods and datasets applied to this most-recent global carbon budget as well as evolving community understanding of the global carbon cycle. The data presented in this work are available at https://doi.org/10.18160/GCP-2025 (Friedlingstein et al., 2025c).