Climate response to off-equatorial stratospheric sulfur injections in threeEarth system models - Part 2: Stratospheric and free-tropospheric response

The paper constitutes Part 2 of a study performing a first systematicinter-model comparison of the atmospheric responses to stratospheric aerosolinjection (SAI) at various single latitudes in the tropics, as simulated bythree state-of-the-art Earth system models - CESM2-WACCM6, UKESM1.0, andGISS-E2.1-G. Building on Part 1 (Visioni et al., 2023) we demonstratethe role of biases in the climatological circulation and specific aspects ofthe model microphysics in driving the inter-model differences in thesimulated sulfate distributions. We then characterize the simulated changesin stratospheric and free-tropospheric temperatures, ozone, water vapor, andlarge-scale circulation, elucidating the role of the above aspects inthe surface SAI responses discussed in Part 1. We show that the differences in the aerosol spatial distribution can beexplained by the significantly faster shallow branches of the Brewer-Dobsoncirculation in CESM2, a relatively isolated tropical pipe and older tropicalage of air in UKESM, and smaller aerosol sizes and relatively strongerhorizontal mixing (thus very young stratospheric age of air) in the two GISSversions used. We also find a large spread in the magnitudes of the tropicallower-stratospheric warming amongst the models, driven by microphysical,chemical, and dynamical differences. These lead to large differences instratospheric water vapor responses, with significant increases instratospheric water vapor under SAI in CESM2 and GISS that were largely notreproduced in UKESM. For ozone, good agreement was found in the tropicalstratosphere amongst the models with more complex microphysics, with lowerstratospheric ozone changes consistent with the SAI-induced modulation ofthe large-scale circulation and the resulting changes in transport. Incontrast, we find a large inter-model spread in the Antarctic ozoneresponses that can largely be explained by the differences in the simulatedlatitudinal distributions of aerosols as well as the degree ofimplementation of heterogeneous halogen chemistry on sulfate in the models. The use of GISS runs with bulk microphysics demonstrates the importance ofmore detailed treatment of aerosol processes, with contrastingly differentstratospheric SAI responses to the models using the two-moment aerosoltreatment; however, some problems in halogen chemistry in GISS are alsoidentified that require further attention. Overall, our results contributeto an increased understanding of the underlying physical mechanisms as wellas identifying and narrowing the uncertainty in model projections of climateimpacts from SAI.