The 1991 eruption of Mount Pinatubo and a 2020 global shipping fuel regulation are being used as natural experiments to test whether deliberately reflecting sunlight could temporarily cool the planet. As global temperatures edge past the Paris Agreement’s 1.5°C target, governments are seriously considering solar radiation management (SRM)—techniques such as spraying reflective aerosols into the stratosphere or brightening marine clouds. But computer models cannot reliably predict whether these methods would work or trigger dangerous side effects like ozone loss, disrupted rainfall, or altered El Niño patterns. This project tackles that uncertainty head-on. By training machine learning on decades of satellite data, the team will compare what happened to clouds and stratospheric aerosols during real-world events (volcanic eruptions, shipping emission changes) against a computer-generated “counterfactual” of what would have happened without them. This allows them to isolate the SRM-like signal from natural weather variability—something previous models could not do. If successful, the research will provide the first robust observational constraints for SRM models, enabling governments to weigh the risks of climate warming against the risks of deliberate intervention. It is fundamental science with immediate policy relevance.
View original technical description
NASA and the UK Met Office project that 2024 will likely temporarily surpass the Paris Agreement target of +1.5°C warming from pre-industrial levels. To address this warming crisis, international governments are considering solar radiation management (SRM) as a temporary measure to cool the Earth by reflecting more sunlight back into space, thereby buying time for global decarbonization. The two leading SRM approaches are stratospheric aerosol injection (SAI) which creates a " parasol-like" aerosol layer, and marine cloud brightening (MCB) which enhances cloud reflectivity and coverage over oceans. While modelling studies suggest that SAI and MCB could slow down warming by reflecting solar radiation, their effectiveness and potential side-effects remain uncertain. Potential detrimental side-effects include ozone layer depletion, altered precipitation patterns leading to flooding or drought, shifting of the Intertropical Convergence Zone (ITCZ) and altered El Niño-Southern Oscillation (ENSO) events that would have a global impact on extreme weather. Despite these uncertainties, the accelerating pace of global warming is prompting serious consideration of SRM. This is reflected in recent initiatives like the UK’s ARIA Active Cooling program 2024, US NOAA’s Earth’s Radiation Budget-SRM programme 2024, and Australia’s regional MCB efforts to protect the Great Barrier Reef since 2020. Current SRM models are limited in accurately predicting outcomes due to structural and parametric uncertainties, especially in representing key atmospheric processes like aerosol-cloud interactions, the core mechanism behind MCB. However, modern Earth observation networks, which monitor Earth systems holistically during natural events such as volcanic eruptions and wildfires, provide valuable analogues for studying SRM. Thus, it is crucial to "QUantify the Efficacy and risks of Solar radiaTION management using natural analogues" (QUESTION). QUESTION will address three key challenges in solar radiation management. Challenge-1 is ensuring that the scale of natural analogues is large enough (thousands of km) to effectively mimic SRM and assess their global impacts. Challenge-2 is separating SRM signals from other co-varying confounders like greenhouse gases (GHGs) and meteorology. Challenge-3 is a lack of robust climate-relevant scale observational constraints to improve modelling and attribution of SRM’s impact. Large global and regional perturbations during the satellite-era offer opportunities to address these challenges. For example, the Pinatubo-1991 eruption, which injected about 18 Tg of SO2 into the stratosphere, serves as a strong analogue for SAI, while the International Maritime Organization’s 2020 regulation that reduced global shipping sulphur emissions by 80% provides an ideal analogue for MCB-termination. While large-scale analogues helps to better represent SRM (Challenge-1), their challenge is to isolate the effects of SRM from that of weather variability which is different with place and time (Challenge-2). However, our recent development of novel machine-learning application means that Changes-1&2 can now be addressed simultaneously and therefore achieve QUESTION’s ambition. By training machine-learning on long-term satellite data, we can estimate how clouds or stratospheric aerosols would behave without perturbations (counterfactual), such as changes in ship emissions or volcanic activity, thus clarifying the impacts of large-scale perturbations. QUESTION (Theme-3) will enable us to constrain and advance climate models, and to attribute the recent quick warming and weather extremes to the quantified influences of these natural events using robust climate attribution methods (Challenge-3). This research aligns with Theme-1 and Theme-2, providing a strong scientific foundation for informed and responsible climate action through supporting comprehensive risk-risk assessment: comparing detrimental impacts of climate warming scenarios versus SRM deployment.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know