What is the Low latency Global Carbon Budget?
The Low latency Global Carbon Budget is an early, twice-yearly assessment of the global CO2 budget. It estimates how much CO2 is emitted by fossil fuels and cement, how much stays in the atmosphere, and how much is taken up by the ocean and the land.
It combines near-real-time fossil emission estimates, dynamic global vegetation models (DGVMs) driven by ERA5 weather data, machine-learning emulators of ocean models and data products, and atmospheric inversions constrained by the OCO-2 satellite. It is an activity of the Global Carbon Project.
How does it relate to the annual Global Carbon Budget?
The Global Carbon Budget (GCB) is the reference assessment. It is published every year with complete data up to the previous year and uses large model ensembles.
The low latency budget covers the most recent months. It uses fewer models, chosen for speed and calibrated against the GCB ensemble, and gives an early signal of changes in the land and ocean sinks. Its numbers are preliminary and are superseded by the GCB when that becomes available.
How is it produced so quickly?
Each component uses inputs that are available within one to six months of the observations. Fossil emissions come from daily Carbon Monitor estimates, ERA5 reanalysis weather drives the DGVMs, fire emissions come from GFED4.1s and GFAS, and the OCO-2 inversions follow within a few months.
Machine-learning emulators extend the ocean models and data products to the latest month.
Why do the vegetation models and the atmospheric inversions give different numbers?
The two methods are independent. DGVMs simulate land fluxes from weather and CO2 and never see atmospheric CO2 measurements. They can over-react to drought, for example because they lack groundwater access, drought-induced tree mortality and the slow decay of deadwood.
Inversions are fitted to satellite CO2 observations, but they depend on atmospheric transport, prior fluxes and where the satellite can see.
We report both, and their mean, so that agreement and disagreement between them are visible.
What is the budget imbalance?
In the bottom-up budget, the imbalance is fossil emissions minus the atmospheric CO2 increase and the land and ocean sinks estimated by models. A positive imbalance means that the models underestimate the sinks, or that emissions are overestimated.
The top-down budget closes almost by construction, because the inversions are fitted to atmospheric observations. Since the July 2024–June 2025 edition, the imbalance is given with a 95% interval from Monte Carlo propagation.
Why are some editions July–June instead of calendar years?
El Niño and La Niña events usually develop in the second half of one year and decay in the first half of the next. A July–June window captures the carbon-cycle response to such an event in one period. It also allows an update in the middle of the year.
Why is 2015–2022 used as the reference period?
It is the common period before 2023 that is covered by all the satellite-based inversions and the models. Using the same years for every method means that anomalies can be compared like for like.
How are the uncertainties calculated?
Values given as mean ± 1σ are the ensemble mean and the standard deviation across the models or inversions. This is a measure of spread, not a full uncertainty.
In the 2025 edition, changes between years and anomalies relative to 2015–2022 are calculated member by member before averaging, so that the persistent biases of each model cancel.
The combined land or ocean sink is the mean of the bottom-up and top-down estimates. Its uncertainty is half the root-sum-square of their two standard deviations.
What is the net land sink?
The net land sink is the net CO2 exchange between the land and the atmosphere. It includes photosynthesis, respiration, fires, rivers and land-use change emissions. In Global Carbon Budget terms it is close to the natural land sink minus land-use change emissions.
On this website a positive value is uptake by the land, and a negative value is a net source to the atmosphere.
Why do numbers for the same year differ between editions?
Each edition uses updated inputs, for example revised NOAA growth rates, updated fossil emission estimates and longer inversion runs. Some editions also use a different set of models. For instance, the OCN vegetation model was used until the 2024 edition, and the CTE inversion was added from the July 2024–June 2025 edition.
Always cite the edition you use. The latest edition is the current best estimate, including for earlier years.
Can I use the data and figures?
Yes. Data and figures are released under the Creative Commons Attribution 4.0 licence. Please cite the paper of the edition you use and follow the Fair Use Statements in the Data Hub.
For output of individual models, please contact the leads of this activity, who will put you in touch with the modelling groups.
What happened to the land sink in 2023 and 2024?
In 2023 the global net land sink fell to 0.44 ± 0.21 GtC yr−1, the weakest since 2003. The largest abnormal loss was in the Amazon during the drought of the second half of 2023, with a further loss in Southeast Asia and extreme fire emissions in Canada.
From July 2023 to June 2024, during the 2023/24 El Niño, the net land sink weakened by 2.24 GtC yr−1. The tropics accounted for 97.5% of this anomaly.
In 2024 the DGVMs estimate a net land source of 0.65 ± 0.85 GtC yr−1, the weakest net land flux since 1987, while the OCO-2 inversions give a small sink of 0.29 ± 0.80 GtC yr−1. The Amazon and central Africa had the largest source anomalies, and the northern land sink (north of 30°N) continued its decline since 2015, to 0.90 GtC yr−1.
Had the land sink recovered by mid-2025?
Only in part, according to the July 2024–June 2025 edition. The net land sink was 1.06 ± 0.40 GtC yr−1. This was stronger than during the El Niño, but 0.49 GtC yr−1 weaker than the 2015–2022 mean.
Most of the deficit came from losses in late 2024. In the DGVMs, July to December 2024 accounted for 86% of the annual deficit.
Whether the land had recovered by early 2025 depends on the method. The DGVMs retain a weak deficit, whereas all inversions indicate fluxes close to the reference mean and a stronger-than-normal northern sink in late spring 2025.
The calendar-year 2025 edition, which is in review, finds that the net land sink rebounded to 2.36 ± 0.16 GtC yr−1 in 2025, 0.71 ± 0.13 GtC yr−1 above the 2015–2022 mean.