The Low latency Global Carbon Budget is an activity of the Global Carbon Project. It provides an early estimate of the global CO2 budget for the most recent year, using near-real-time data and fewer, faster models than the annual Global Carbon Budget.
Each edition estimates fossil CO2 emissions, the growth of CO2 in the atmosphere and the uptake by the ocean and the land.
The first low latency budget covered 2023. That year the CO2 growth rate hit a record high while fossil emissions barely grew, which implied a large decline of the land carbon sink. The low latency budget gives an early signal of such changes in the land and ocean sinks.
The 2023 budget was published in National Science Review in October 2024. Three more editions have since been published, for July 2023 to June 2024 (iScience), for 2024 (National Science Review) and for July 2024 to June 2025 (Advanced Science). The 2025 edition is in review.
The activity is led by Philippe Ciais and Piyu Ke at the Laboratoire des Sciences du Climat et de l'Environnement (LSCE) in France. Each edition is a collaboration of land, ocean and atmospheric inversion modelling groups. Each paper and its authors are cited on the Key Results page of that edition.
The annual Global Carbon Budget 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 follows the same structure but uses fewer models, chosen for speed and calibrated against the Global Carbon Budget ensemble. Its numbers are preliminary and are superseded by the Global Carbon Budget when that becomes available.
Two editions are prepared each year. The calendar-year edition covers January to December and is prepared in the first months of the following year. The July–June edition follows the usual timing of El Niño and La Niña events, which develop in the second half of one year and decay in the first half of the next. It is prepared in the second half of the year.
Low latency Global Carbon Budget Leads
Philippe Ciais
Laboratoire des Sciences du Climat et de l'Environnement (LSCE), France
philippe.ciais@cea.fr
Piyu Ke
Laboratoire des Sciences du Climat et de l'Environnement (LSCE), France
piyu.ke@lsce.ipsl.fr
Contributing Institutions
- Laboratoire des Sciences du Climat et de l'Environnement (LSCE/IPSL), France
- University of Exeter, United Kingdom
- Tsinghua University, China
- Institute of Tibetan Plateau Research, Chinese Academy of Sciences, China
- Peking University, China
- Wageningen University, the Netherlands
- NASA Jet Propulsion Laboratory, USA
- Colorado State University, USA
- NOAA Global Monitoring Laboratory, USA
- Spark Climate Solutions, USA
- Microsoft Research, China
- Columbia University, USA
- Leipzig University, Germany
- Max Planck Institute for Biogeochemistry, Germany
- CICERO Center for International Climate Research, Norway
- The University of Hong Kong, China
- Laboratoire de Météorologie Dynamique (LMD/IPSL), France
- Earth System Science Interdisciplinary Center, University of Maryland, USA
- Cooperative Institute for Research in Environmental Sciences (CIRES), University of Colorado Boulder, USA
- Seoul National University, Republic of Korea
How the budget is made
The budget follows the structure of the Global Carbon Budget. Fossil CO2 emissions are either added to the atmosphere or taken up by the ocean and the land. Every term is estimated in two independent ways. The bottom-up budget uses process models and data-driven products. The top-down budget uses atmospheric inversions of OCO-2 satellite observations.
Fossil CO2 emissions are the average of two near-real-time estimates. Carbon Monitor is based on daily activity data in six sectors. The Global Carbon Budget method is based on preliminary energy statistics and projections. The growth of CO2 in the atmosphere comes from the NOAA globally averaged marine boundary layer record (the 2023 edition also reports the Mauna Loa record), with growth rates from OCO-2 and from the inversions as cross-checks.
The land sink is estimated with dynamic global vegetation models driven by ERA5 weather. Their mean level is calibrated to the median of the Global Carbon Budget 2023 models over 2019–2022, so the low latency models supply the year-to-year anomalies. Fire emissions in the models are replaced by the mean of the GFED4.1s and GFAS satellite products.
The ocean sink comes from machine-learning emulators trained on each ocean biogeochemical model and each surface-ocean fCO2 data product of the Global Carbon Budget. The emulators use observed predictors, such as sea-surface temperature, sea ice, wind speed and atmospheric CO2, to extend each product to the latest month.
The top-down budget uses global inversions that assimilate OCO-2 column CO2 retrievals (CAMS in the 2023 edition; CAMS, CMS-Flux and GONGGA from July 2023–June 2024; CTE added from July 2024–June 2025). Their fluxes are corrected for the natural riverine carbon loop so that they can be compared with the bottom-up estimates.
The budget imbalance is what remains when the atmospheric growth and the ocean and land sinks are subtracted from fossil emissions. A positive imbalance means a missing sink. Other models and data, including GRACE terrestrial water storage, are used to check the anomalies, but they do not enter the central budget values.
| 2023 | Jul 2023–Jun 2024 | 2024 | Jul 2024–Jun 2025 | 2025 | |
|---|---|---|---|---|---|
| Period | Jan–Dec 2023 | Jul 2023–Jun 2024 | Jan–Dec 2024 | Jul 2024–Jun 2025 | Jan–Dec 2025 |
| DGVMs | ORCHIDEE, JULES, OCN | JULES, OCN, ORCHIDEE-MICT, LPJ-EOSIM | JULES, OCN, ORCHIDEE-MICT, LPJ-EOSIM | JULES, ORCHIDEE, LPJ-EOSIM | JULES, ORCHIDEE-MICT, LPJ-EOSIM |
| Inversions (OCO-2) | CAMS | CAMS, CMS-Flux, GONGGA | CAMS, CMS-Flux, GONGGA | CAMS, CMS-Flux, GONGGA, CTE | CAMS, CMS-Flux, GONGGA, CTE |
| Ocean emulators trained on | GCB 2022 | GCB 2022 | GCB 2022 | GCB 2025 | GCB 2025 |
| Cross-checks | GRACE water storage | SiB4; 19 DGVM emulators; GRACE | SiB4; GRACE | 20 DGVM emulators; GRACE | GRACE; BESS |
| Imbalance uncertainty | — | — | ±1σ | Monte Carlo, 95% interval | Monte Carlo, 95% interval |
ORCHIDEE and ORCHIDEE-MICT refer to the same low latency ORCHIDEE configuration; model names are given as in each paper.
Funding
The low latency budgets were supported by the European Space Agency Climate Space RECCAP2-CS project (ESA ESRIN/4000144908) and ESA Carbon-RO (4000140982/23/I-EF); the CALIPSO and CLARiTy projects supported by Schmidt Sciences, LLC; the European Union's Horizon Europe research and innovation programme (grant agreement No 101081395, EYE-CLIMA); NASA (80NM0018D0004) and the NASA Orbiting Carbon Observatory Science Team; NOAA cooperative agreement NA22OAR4320151; GENCI/TGCC and SURF computing allocations; and further national grants listed in each publication.