The main findings for July 2024 to June 2025, as published in Advanced Science.
Late-2024 carbon losses dominated the year; early-2025 land recovery signals differ between methods
After the 2023/24 El Niño, the CO2 growth rate fell but stayed above average. The land sink strengthened relative to the El Niño year yet remained weaker than the 2015–2022 mean, mostly because of losses in late 2024. Whether the land had recovered by early 2025 depends on the method: models retain a deficit, while all inversions show fluxes close to normal.
- The CO2 growth rate was 2.62 ± 0.08 ppm yr−1 from July 2024 to June 2025, down from 3.66 ppm yr−1 during the El Niño but 6.5% above the 2013–2022 July-to-June mean.
- Combining bottom-up and top-down estimates, the net land sink was 1.06 ± 0.40 GtC yr−1: stronger than in 2023/24, but 0.49 GtC yr−1 weaker than the 2015–2022 mean.
- The annual land anomaly was dominated by late-2024 losses: in the DGVMs, July–December 2024 accounted for 86% of the annual deficit.
- Early-2025 recovery is method-dependent: 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.
- Ocean uptake shows no global weakening. Temperature explains most of the global land-flux variability, while terrestrial water storage matters more regionally.
The budget at a glance
| Quantity | July 2024–June 2025 | Notes |
|---|---|---|
| Atmospheric CO2 growth rate | 2.62 ± 0.08 ppm yr−1 | NOAA marine boundary layer, July–June; 6.5% above the 2013–2022 mean |
| Fossil CO2 emissions | 10.35 GtC yr−1 | Range 10.33–10.37 across two estimates |
| Net land sink | 1.06 ± 0.40 GtC yr−1 | Mean of DGVMs (0.55 ± 0.31) and inversions (1.58 ± 0.73); 0.49 below the 2015–2022 mean |
| Ocean sink | 3.06 ± 0.34 GtC yr−1 | Emulators 3.19 ± 0.65; inversions 2.93 ± 0.17 |
| Budget imbalance | 1.05 ± 0.70 GtC yr−1 | 95% interval −0.44 to 2.23, not significantly different from zero |
Values as published, for July 2024–June 2025. Uncertainties are ±1σ; the budget imbalance uncertainty comes from Monte Carlo propagation.
Models and data used
- Fossil CO2 emissions. Carbon Monitor and the Global Carbon Budget method; monthly split from IEA-EDGAR
- Atmospheric growth rate. NOAA MBL (June–July); OCO-2 and GRESO growth rates in a sensitivity test
- Land, bottom-up. 3 DGVMs driven by ERA5: JULES, ORCHIDEE, LPJ-EOSIM; emulators of 20 GCB 2025 DGVMs as a benchmark
- Ocean, bottom-up. Machine-learning emulators of 10 ocean models and 9 data products (GCB 2025)
- Top-down. 4 OCO-2 inversions: CAMS, CMS-Flux, GONGGA, CTE
- Anomaly reference. 2015–2022, July–June windows
Publication
Ke, P., Ciais, P., Yao, Y., Sitch, S., Li, W., Du, X., et al. (2026). Low Latency Global Carbon Budget Reveals Late 2024 Carbon Losses and Contrasting Early 2025 Land Sink Recovery Signals. Advanced Science, e77369. https://doi.org/10.1002/advs.77369