Drivers of the natural CO2 fluxes at global scale as simulated by CMIP6 simulations

The implementation of the Paris Agreement should translate into a decrease of the growth rate of atmospheric CO2 in the coming decades due to the reduction in emissions by signing countries. However, the detection of this decrease and its attribution to mitigation measures will be challenging for two reasons: 1) the internal variability of the Earth system may temporarily offset this signal and 2) countries may not maintain their promises. Unless absolute transparency on emissions is adopted by all signing parties, without a robust estimate of the impact of internal variability on the atmospheric CO2 changes, there is no independent way to verify their claims. 

Historical reconstructions and future predictions of global carbon cycle dynamics with predictive systems based on state-of-the-art Earth System Models (ESMs) represent an emerging field of research. With the continuous improvement of ESMs and of these predictive systems, these tools might have the potential of becoming skillful enough in their predictions to represent a useful instrument for policy makers in their effort to monitor and verify the progress of the Paris Agreement’s implementation. 

Here we analyze the main sources of the atmospheric CO2 concentration variability at inter-annual timescale due to internal climate processes in three ESMs, which are used in carbon cycle prediction systems: EC-Earth3-CC, IPSL-CM6A-LR, and MPI-ESM1-2-LR. These results are then compared to the available CMIP6 simulations database.

Investigating the surface CO2 fluxes, we find that land flux inter-annual variations are 10 times higher than ocean flux variations. This has direct consequences in terms of predictability since the land surface processes are generally less predictable than the ocean ones. The regions contributing the most to the variations are Australia, South America and sub-Saharan Africa, suggesting that those are the most important regions to simulate correctly in order to constrain the atmospheric CO2 variations. Interestingly, all those regions are linked to tropical SST variations resembling El Niño Southern Oscillation variability.

Investigating the ocean CO2 fluxes, we find that the regions contributing the most to the global CO2 variations are the Southern Ocean followed by the tropical Pacific.

Therefore, from the analysis of the CMIP6 simulations, we conclude that the main internal driver of the global atmospheric CO2 fluctuations is the tropical Pacific. If the ratio between land and ocean CO2 variations is realistically simulated by the CMIP6 ESMs, this implies that the predictability of the atmospheric CO2 variations due to internal climate processes is tied to the predictability of the tropical Pacific.