Effectiveness of stratospheric solar-radiation management as a function of climate sensitivity

If implementation of proposals to engineer the climate through solar-radiation management (SRM) ever occurs, it is likely to be contingent on climate sensitivity. However, modelling studies examining the effectiveness of SRM as a strategy to offset anthropogenic climate change have used only the standard parameterizations of atmosphere–ocean general circulation models that yield climate sensitivities close to the Coupled Model Intercomparison Project mean. Here, we use a perturbed-physics ensemble modelling experiment to examine how the response of the climate to SRM implemented in the stratosphere (SRM-S) varies under different greenhouse-gas climate sensitivities. When SRM-S is used to compensate for rising atmospheric concentrations of greenhouse gases, its effectiveness in stabilizing regional climates diminishes with increasing climate sensitivity. However, the potential of SRM-S to slow down unmitigated climate change, even regionally, increases with climate sensitivity. On average, in variants of the model with higher sensitivity, SRM-S reduces regional rates of temperature change by more than 90% and rates of precipitation change bymore than 50%. The Royal Society has defined SRM as techniques that ‘attempt to offset effects of increased greenhouse gas (GHG) concentrations by causing the Earth to absorb less solar radiation’1. The most plausible large-scale method is to increase the loading of lightscattering aerosols in the stratosphere (SRM-S; ref. 1). A number of atmosphere–ocean general circulationmodel (AOGCM)modelling studies suggest that SRM can compensate for many of the temperature and precipitation changes associatedwith global warming, even at the regional level2–4, though these regional compensatory effects are not uniform4,5. These previous studies have used models in which the climate’s equilibrium sensitivity to GHG forcing (henceforth, CS) reflects near-median estimates of CS. However, both observationally constrained and expert-elicited estimates of CS have a substantial ‘high tail’6,7, and it is arguablymore likely that if SRM is deployed it will be becauseCS, and the impacts fromclimate change, turn out to be higher than current best estimates. Here we examine the effectiveness and side-effects of SRM-S across a range of CS to check if use of themeanCS biases our understanding of SRM. Evaluating the effectiveness of SRM-S requires first specifying the conditions inwhich itmight be implemented and the effects that would be desired. There are various scenarios under which SRM might be employed. From a conventional policy viewpoint in which SRM is one of a portfolio of strategies alongside mitigation and