Carbon dioxide removal: trade-offs and lags in large perturbed parameter simulations

Carbon Dioxide Removal (CDR) is now widely discussed for offsetting residual greenhouse gas emissions or even reversing climate change. For example, all emissions scenarios of the Intergovernmental Panel on Climate Change that meet the “well below 2°C” warming target of the Paris Agreement include CDR. Ocean alkalinity enhancement (OAE) may be one possible CDR where the carbon uptake of the ocean is increased by artificial alkalinity addition. Here, we apply the Bern3D-LPX and the UVic Earth system models of intermediate complexity in observationally-constrained large perturbed parameter ensembles to investigate the effect of massive OAE on modelled carbon reservoirs and fluxes. OAE is assumed to be technically successful and deployed as an additional CDR in the SSP5-3.4 temperature overshoot scenario. 

Trade-offs involving feedbacks with atmospheric CO₂ result in a low efficiency of an alkalinity-driven atmospheric CO₂ reduction of -0.35 [-0.37 – -0.33] mol C per mol alkalinity addition (skill-weighted mean and 68% c.i.). The alkalinity-driven ocean carbon uptake is partly offset by the release of carbon from the land biosphere and a reduced ocean carbon sink in response to lowered atmospheric CO₂ under OAE.
We further apply the Bern3D-LPX ensemble in idealized simulations, in which ΔSAT increases first to ~2°C and then declines to ~1.5°C, to investigate lags in surface air temperature change (ΔSAT). In these simulations, ΔSAT lags the decline in CO₂-forcing by decades, depending on the equilibrium climate sensitivity of the respective ensemble member.
Finally, we use the Bern3D-LPX ensemble simulations and the, in comparison to earlier studies with the Bern3D-LPX model, updated and longer observational records to assess climate metrics such as the transient climate response to emissions, the transient climate response, and the equilibrium climate sensitivity. 

Our results suggest that massive OAE, if technically and socio-economically achievable, might be able to lower atmospheric CO₂ but considering the trade-offs and lags, not emitting carbon is preferable.