Change in negative emission burden between an overshoot versus peak-shaved stratospheric aerosol injection pathway

Abstract. Stratospheric aerosol injection (SAI) geoengineering is being investigated as a potential means of temporarily reducing the impact of global warming, allowing additional time for the implementation of climate mitigation strategies. SAI operates by intervening in the radiative energy balance of the Earth system, exerting a temporary direct cooling effect on the climate. However, SAI also indirectly affects global temperature through its impact on atmospheric CO2 levels by influencing the natural carbon uptake efficiency. Most previous research on the carbon cycle under SAI suggests that continuous injections enhance the uptake of carbon, implying a larger number of allowable emissions for a given temperature target relative to a simulation without SAI. However, there are considerable uncertainties regarding the extent and timeline of facilitation or inhibition of atmospheric carbon removal under SAI. In this study, we evaluate the extent of change in negative emission burden (NEB) over the entire trajectory of a peak-shaving SAI deployment (SSP534-sulfur) compared to the baseline overshoot pathway (SSP534-over) that does not involve SAI. We run the SSP534-over scenario on the CNRM-ESM2-1 Earth system model from 2015 to 2249 and compare it to the simulation where, under SSP534-over conditions, SAI is used to maintain 2 °C warming (SSP534-sulfur). The results indicate that carbon effects are reinforced under SAI. While the land carbon reservoir is a carbon sink, SAI enhances the uptake further; when the land acts as a carbon source, SAI enhances the outgassing. Thereby, carbon fluxes associated with SAI evolve over time: the increase in carbon uptake under SAI during the positive emission phase confirms prior studies and substantiates the concept of buying time during SAI ramp-up; later stages of the peak-shaved SAI scenario show the carbon benefit reducing and turning into an additional obstacle, making a phase-out of SAI more difficult by enhancing the carbon removal burden. The findings of this study may be contingent upon the configuration of the injection design and the representation of SAI within the model, as well as the underlying overshoot scenario. Further research is necessary to validate these results using different models incorporating diverse SAI deployment strategies and underlying emission trajectories.