CLIMATE SCIENCE & POLICY
A new phased research plan attempts to define what scientists would need to learn before any decision on stratospheric aerosol injection—and why evidence alone may never settle the argument.
Reflective, a San Francisco nonprofit that finances solar geoengineering research, has published a detailed plan for the experiments, scientific studies and technical capacity it believes would be required before governments could make an informed decision about stratospheric aerosol injection.
The underlying idea has been examined for roughly half a century. By placing reflective particles high in the atmosphere, researchers hope it might be possible to imitate part of the cooling that follows major volcanic eruptions and temporarily counter some greenhouse-driven warming.
Hundreds of studies have explored that possibility, yet decisive gaps remain. Scientists still cannot say with confidence how efficiently the method would cool the planet, which materials would perform best, how particles would behave after release or what regional and environmental side effects might follow. Until now, no common programme has mapped how to close those gaps.
Reflective’s SAI Research Roadmap is an attempt to turn that diffuse uncertainty into a sequence of work. Its estimate is striking: coordinated research could take about ten years and cost roughly $370 million. Without coordination, the same learning process might stretch towards two decades and nearly $1.4 billion.
The plan is contentious because it includes outdoor releases that grow dramatically in scale. Reflective says research is not an endorsement of deployment. Critics answer that technical knowledge cannot resolve the prior political question: who would control a technology capable of changing conditions across the whole planet, whose interests would guide it, and who would bear the unwanted effects?
Why the Timetable Matters
After Dakota Gruener established Reflective in late 2023, it rapidly became an increasingly significant funder and organiser in the field. It has raised more than $20 million from charities and individual donors, distributed about $4 million among several dozen research groups, and supported its own projects, including an open-source simulator and a digital space for collaboration.
Gruener’s central concern is timing. Climate risks may force consequential choices sooner than normal research systems can supply dependable answers. The road map is intended to focus scientists, philanthropies and public agencies on the work with the greatest decision value, so that uncertainty is reduced deliberately rather than during a political emergency.
That speed argument is why coordination matters. Separate projects can duplicate modelling, use incompatible assumptions, or postpone shared monitoring infrastructure. A common sequence would allow later experiments to build on earlier measurements and let funders see which questions block the next decision. Reflective is effectively treating research design as a shared capability rather than a collection of isolated papers. The projected savings depend on institutions agreeing about priorities early, while results can still redirect the programme.
An earlier Reflective project, the SAI Uncertainties database, catalogued scientific unknowns and engineering barriers that stand in the way of even a small intervention. One fundamental issue is the material itself. Researchers must compare sulfur dioxide with other possible gases or particles, then understand the chemistry and physics each would encounter in the unusually dry stratosphere.
Distribution is equally important. Aerosols that spread and remain suspended could reflect sunlight efficiently; particles that combine into larger clumps may fall rapidly into the lower atmosphere. Those outcomes change how much material would be needed, how long it would persist, and what harms could accompany it. The new plan builds a route from questions like these towards measurements capable of answering them.
From Models to Staged Releases
The opening phase, described as foundational knowledge, stays largely within laboratories and computers. It calls for additional modelling and controlled experiments on possible consequences for regions and ecosystems, including effects on ocean circulation, ice sheets and agricultural yields. The goal is not merely to calculate a global average temperature change, but to understand how benefits and risks might be distributed.
At the same time, researchers would expand instruments and observation systems for measuring the stratosphere as it exists today. A stronger baseline is essential: without knowing normal chemical and physical variation, investigators could struggle to distinguish the signal of a later experiment from background change. Reflective budgets $30 million to $75 million for a phase lasting two to three years, while some monitoring would continue afterwards as well.
The following stage moves outdoors. Aircraft would carry sulfur dioxide into the stratosphere, adding ten metric tons on four occasions spread across two seasons. Including supporting work, the projected budget reaches $70–$150 million in total across an estimated four to eight years. Reflective estimates that it could cut uncertainty around cooling efficacy—the temperature effect produced by each ton of sulfur—by about 25%.
A later experimental stage would be vastly larger, adding 25,000 tons over a single season at least once and perhaps twice. Including related research, that phase is projected to take four to 11 years and cost from $270 million to $1.1 billion. The plan says it could reduce uncertainty about cooling efficacy by approximately 66%.
These figures do not represent a simple promise to proceed. Some activities could overlap, and each step is meant to depend on results from the one before it. The final research phase would exist only if full-scale geoengineering were ultimately undertaken: continuing observation would compare real-world behaviour with models and search for consequences that were missed or underestimated.
Gruener describes the document as an initial version, specific enough for researchers and critics to challenge. Reflective plans to revise it as specialists identify weak assumptions, missing questions or better experimental designs. A feedback system attached to the project is supposed to make that revision process explicit rather than leaving the plan fixed.
Most importantly, the later phases are separated by decision gates. If a release fails to produce the expected signal, reveals troubling impacts or leaves crucial questions unresolved, the next stage should not begin. In Reflective’s framing, a successful research programme must be able to generate a scientifically justified instruction to stop, not only a path towards larger tests.
The Questions Science Cannot Settle
Yet a carefully ordered programme does not make outdoor testing politically acceptable. Proposed field experiments have repeatedly met resistance. Harvard’s SCoPEx effort and the United Kingdom’s SPICE project were both abandoned after opposition from environmental groups, local actors or policymakers, illustrating how quickly a limited scientific test can become a proxy battle over eventual deployment.
The international non-use campaign goes further. Since 2002, hundreds of academics have supported calls to prohibit outdoor solar geoengineering experiments and establish an agreement against use. They argue that no plausible governance system could guarantee fair control of a tool whose climate effects would cross every border.
Aarti Gupta of Wageningen University, who helped initiate the campaign, places authority ahead of engineering. For her, the primary questions concern who develops stratospheric aerosol injection, who decides when and why it is used, and which populations gain or lose. More atmospheric data cannot by itself supply legitimate answers to those choices.
Wil Burns, an American University legal scholar who signed the non-use agreement, also doubts whether intermediate experiments can produce information relevant enough for a responsible decision. Potential damage to the ozone layer or shifts in regional rainfall may not become adequately visible until intervention reaches something close to operational scale. By then, the world would no longer be conducting a modest test.
Burns also frames deployment as a burden on people not yet represented in the decision. If greenhouse-gas emissions continued while increasing aerosol levels merely concealed their warming, future generations could inherit a system they were unable to end safely. A rapid halt could expose the accumulated warming in a sudden jump, the danger commonly called termination shock.
Some scientists consider that scenario less likely than its starkest descriptions suggest and argue that an intervention could be reduced gradually. But the dispute shows why experimental confidence and political legitimacy are different things. Demonstrating a measurable cooling effect would not prove that indefinite operation, international control or the transfer of risk across generations was acceptable.
Why Supporters Still Want Field Tests
Supporters of the road map do not claim that every uncertainty can disappear. They argue instead that public decisions will occur under uncertainty regardless, and that research should reduce the most consequential unknowns while there is still time to design safeguards and reject approaches that fail.
Ilan Gur, formerly chief executive of the UK’s Advanced Research and Invention Agency, praises the plan on those grounds. ARIA funded 21 geoengineering projects last year. Gur’s position is that scientists, policymakers and citizens share an interest in learning efficiently whether an approach might work at all, so money spent narrowing the biggest uncertainties can prevent both false confidence and premature dismissal.
Sebastian Eastham, an Imperial College London researcher leading an ARIA-backed project on a different engineered-cooling method, agrees that Reflective’s outdoor programme would leave important questions open. He nevertheless sees value in creating a concrete framework for deciding amid worsening climate risks. Difficult public choices are rarely preceded by complete knowledge; the relevant standard is whether additional evidence meaningfully improves the choice.
For Eastham, simulations eventually reach a limit. Models can test scenarios cheaply and repeatedly, but some atmospheric processes must be observed rather than inferred. A properly designed release could reveal more about particle behaviour and effectiveness than enormous amounts of additional computing. Refusing every experiment may therefore carry its own scientific cost.
The opposing risk is intellectual repetition: researchers could continue running similar simulations while never learning the facts most important to judging performance or danger. Advocates of carefully bounded fieldwork say the aim is not to normalise deployment. It is to replace assumptions with observations and create evidence strong enough to support either continuation or rejection.
Evidence Before a Climate Emergency
Reflective accepts that solar geoengineering could deepen global inequality. The organisation argues, however, that the comparison cannot stop there, because unrestrained warming is also expected to harm poorer and developing regions disproportionately. Any assessment must compare the risks of intervention with the risks of a hotter world, rather than treating inaction as a neutral baseline.
Gruener also acknowledges that outdoor releases cannot eliminate the scientific unknowns. She contends that the smallest proposed tests could answer important questions with limited environmental exposure. For scale, the smallest release would add 10 tons of sulfur dioxide, an amount below two percent of aviation’s worldwide daily output. It remains tiny compared with a global intervention.
That comparison will not satisfy critics who view the first outdoor test as a political threshold rather than a pollution calculation. Once institutions, aircraft and expertise are assembled, research could create momentum towards use. Supporters respond that refusing to study the mechanism does not prevent a country or other actor from attempting it later; it may only ensure that the attempt is poorly understood.
The road map therefore exposes two different fears. One is that research legitimises a planetary intervention whose governance and long-term obligations may never be fair. The other is that accelerating climate damage could push leaders towards action before basic atmospheric behaviour, technical limits and warning signs have been established.
No research schedule can decide which fear should dominate. What it can do is define the evidence that remains missing, attach costs and time to obtaining it, and create moments when the answer may be to stop. Reflective’s wager is that deliberate investigation now is safer than discovering the same uncertainties during a future crisis, when decisions could be driven by urgency rather than knowledge.
