Uncertainty in Regional Monsoon Responses to Solar Radiation Modification Is Often Greater Across Models Than Across Strategies
Abstract. Proposed solar radiation modification (SRM) strategies designed to stabilize global-mean surface temperature could alter regional monsoon systems. We examine changes in monsoon onset, retreat, duration, peak timing, active-season precipitation, wet-day frequency, and wet-day intensity across six monsoon regions in four Earth system models (CESM2, UKESM1, E3SMv3, and MIROC-ES2H). We compare projections under SSP2-4.5, a moderate greenhouse-gas emissions scenario, with three SRM strategies designed to keep global-mean surface temperature at 2020–2039 SSP2-4.5 mean (nominally 1.5 °C above pre-industrial levels): stratospheric aerosol injection (SAI) at mid-latitudes, seasonal SAI at high latitudes, and marine cloud brightening (MCB) over mid-latitude oceans. Under SSP2-4.5, active-season precipitation increases in most monsoon regions, accompanied by delayed retreat and, in the East Asian and Australian-Maritime Continent monsoons, earlier onset and longer rainy seasons. Both SAI strategies reduce the magnitude of the late-century precipitation increases projected under SSP2-4.5. However, MCB retains larger positive precipitation anomalies than the SAI strategies in several regions, consistent with its distinct spatial forcing pattern and hydrological response. Monsoon duration is most often the leading contributor to active-season precipitation changes, while wet-day intensity plays a more model-dependent secondary role. Overall, differences among models exceed differences between SRM strategies in most region–metric combinations. The South Asian monsoon is a notable exception: differences among SRM strategies exceed intermodel differences for most monsoon metrics, indicating that outcomes in this region are particularly sensitive to the choice of SRM intervention. These results underscore the importance of multi-model ensembles for evaluating regional SRM impacts, as conclusions based on individual models may overstate or obscure differences among simulated intervention strategies.