To limit global warming to 2°C or 1.5°C above pre-industrial levels, large-scale deployment of carbon dioxide removal (CDR) methods is needed, in addition to strict emission reductions. The preventive capacity of CDR is estimated to be several hundred GtCO2, which would allow for scaling up deployment if needed and protect against unexpectedly high warming this century. Given the impact of CDR on energy-water-land systems, it is important to consider diverse portfolios that include both terrestrial and marine methods. These include, for example: afforestation/reforestation (AR) a ocean alkalinity increase (OAE). Research into these methods and their combinations in the interactive Earth system is crucial.

CDR reduces atmospheric CO2, contributing to biogeochemical cooling. However, it also triggers complex feedbacks between carbon and climate. One of the key effects is the so-called compensation flows. For example, when the terrestrial carbon sink increases (e.g. through afforestation), the ocean sink decreases compared to the no-CDR scenario, and vice versa. These compensation fluxes cause removal efficiency (defined as the reduction in atmospheric carbon divided by the increase in carbon in the relevant sink) is less than 100 %. This has important implications for monitoring, reporting and verification of CDR.

Study based on two Earth system models (MPI-ESM and FOCI) with a setup of 42 simulations for different ranges of AR (0-927 Mha) and OAE (0-18 Pmol) during the 21st century yielded important insights. It was shown that global carbon flux responses are linear, when different CDR methods are scaled and/or combined. This suggests that CDR effectiveness is insensitive to the amount of CDR applied and to the composition of the CDR portfolio. The combination of methods, which is advantageous for risk diversification and compliance with sustainability thresholds, therefore does not compromise the effectiveness of individual applications.

When scaling AR terrestrial carbon (Cland) increased linearly (e.g., doubling AR resulted in nearly doubling Cland), with its increment per 100 Mha being relatively insensitive to the application rate. The removal efficiency for AR reached % in the halfAR and AR 85 scenarios. Similarly, when scaling UAE Ocean carbon (Cocean) increased linearly (doubling OAE resulted in a nearly doubling of Cocean), with the increase per Pmol being relatively insensitive to application rate over the range tested (up to 16 Pmol). Removal efficiency for OAE was slightly higher (86-87 %) than for AR, but the study cautions that this metric reflects the strength of the compensation rather than the efficiency of the method itself.

Combination of AR and OAE in Mixed and halfMixed scenarios also showed linear responses of global carbon fluxes. The reduction in atmospheric carbon (Catmo) was almost the sum of the effects of the individual methods. This suggests that globally there is little interaction between AR and OAE, despite the potential biogeophysical influences of AR. Linearity was also confirmed when scaling the CDR methods within the portfolio (comparison of halfMixed and Mixed). Removal efficiency does not appear to be reduced with the portfolio.

Regarding global warming mitigation, scaling and combining CDR methods resulted in roughly linear increases in global warming mitigation. For example, in the halfAR and AR scenarios, the average temperature decreased by 0.09 ± 0.1 °C and 0.2 ± 0.11 °C in 2090-2099. In the halfOAE and OAE scenarios, the warming mitigation reached 0.14 ± 0.13 °C and 0.22 ± 0.12 °C. The combined halfMixed and Mixed scenarios showed warming mitigation of 0.20 ± 0.13 °C and 0.42 ± 0.14 °C. It is important to note that at the level of individual model grids, the temperature response is not linear, as complex nonlinear biogeophysical and biogeochemical effects come into play.

Although global and regional carbon fluxes exhibit linearity, Caution is needed when considering project-level estimates. Local deviations from linearity exist and are stronger on land, especially in the case of AR, due to complex, nonlinear local and non-local feedbacks. Estimates of sequestration for individual afforestation projects based on fixed carbon densities may not be accurate. For OAE, local linearity is more pronounced in areas of major sequestration. Nevertheless, the models suggest that linear estimates could be useful for rough approximations in scaling OAE in real life, although they require further experimental validation.

Overall, the results suggest greater flexibility in designing sustainable CDR portfolios that include both terrestrial and marine methods. Combining methods does not compromise the global removal efficiency of individual applications.

This study represents a significant step towards understanding the dynamics of CDR portfolios. Future research should further validate these results using a broader range of CDR methods and scenarios to explore potential interactions that could affect linearity. Spring

Study published in the journal nature.com

Glossary of key terms

• Afforestation/Reforestation (AR): A method of carbon dioxide removal that involves planting forests on land that was not previously forested (afforestation) or reforesting land where forests were previously present (reforestation).
• Carbon Dioxide Removal (CDR): Techniques that remove carbon dioxide (CO2) directly from the atmosphere and store it permanently.
• Carbon-climate feedbacks: Processes by which changes in atmospheric CO2 concentrations affect climate and these climate changes in turn affect natural carbon fluxes (e.g. carbon uptake by land and oceans).
• Catmo: Total carbon content in the atmosphere.
• Clan: Total carbon content in soil (including vegetation).
• Cocean: Total carbon content in the ocean.
• Compensating fluxes: Changes in natural carbon fluxes (e.g. between the atmosphere and ocean or the atmosphere and land) that occur in response to CDR and may offset the original removal of CO2 from the atmosphere.
• Earth System Models (ESM): Complex computer models that simulate interactions between the Earth's atmosphere, ocean, land, and biosphere. High-complexity ESMs include more detailed processes and feedbacks.
• Emission-driven simulations: Model simulation experiments where the input is greenhouse gas emissions (such as CO2) and the model dynamically calculates atmospheric concentrations and subsequent climate responses.
• Linearity: In the context of this article, this means that the response of the system (e.g., the change in carbon in the reservoir) is directly proportional to the size or combination of CDR interventions used.
• Ocean Alkalinity Enhancement (OAE): A carbon dioxide removal method that involves adding alkaline materials to the ocean to increase its ability to absorb atmospheric CO2.
• Removal efficiency: The extent to which the application of CDR results in a net reduction in atmospheric carbon. In the article, it is defined as the reduction in atmospheric carbon divided by the increase in carbon in the soil (for AR) or in the ocean (for OAE).
• SSP3-7.0: One of the Shared Socioeconomic Pathways scenarios, which represents a high-emissions pathway, characterized by continued increases in greenhouse gas emissions throughout the 21st century.