Solar radiation modification (SRM)—also known as solar geoengineering or solar radiation management—is a set of hypothetical interventions that would reflect a portion of sunlight into space or allow more infrared radiation to escape into space. 1 SRM includes both global and regional interventions. The best-known approach is stratospheric aerosol injection (SAI), which would disperse reflective aerosols in the stratosphere to scatter a fraction of incoming sunlight and lower global temperatures (Figure 1 ).
Figure 1: Proposed methods for modifying the Earth’s radiation budget [2]SRM is considered the only known approach that could reduce global temperatures or limit temperature increases within a few years of deployment, assuming inadequate mitigation continues. 3 Most researchers in this space see SRM as a possible “last resort” in case emissions reductions fail to avoid the most dangerous climate scenarios. At the same time, meaningful deployment of SRM would require substantial lead time and is not a “break glass in case of emergency” option.4
Critically, SRM would mask warming rather than address its root cause: greenhouse gas (GHG) emissions. SRM would not address effects of rising CO2 , such as ocean acidification, and would have unintended consequences, such as effects on ozone.5 SRM would not reliably address tipping points, given the large uncertainties in the effects of SRM on the drivers of different Earth system tipping elements.6 Temporarily reducing warming to 1.5°C or 2°C through SRM in a warmer future world would result in different levels and types of climate damage from a world in which temperature targets are met via GHG mitigation.7 Therefore, we still need to drive down emissions regardless of any potential future SRM (non-)deployment decision.
SRM should not substitute for deep decarbonization. Instead, we think that if SRM were deployed safely, responsibly, and justly, it could help reduce near-term warming while slower-acting emissions cuts and carbon dioxide removal take effect.8 However, this “buying time” logic carries risks. On cost-effectiveness grounds, SRM could substitute for the most expensive and difficult mitigation measures rather than supplementing them. Governance that prevents this kind of mitigation deterrence is therefore essential, not optional, to any responsible case for SRM.
Significant uncertainty about SRM’s effects carries high costs. According to Harding, Belaia, & Keith (2022), overconfidence in SRM’s potential could lead to insufficient emissions reductions and climate damages of up to $115 trillion USD. Meanwhile, underconfidence forgoes opportunities to mitigate climate damage cost-effectively using SRM, with costs reaching $95 trillion.9 Because both overconfidence and underconfidence carry substantial risks and costs, the authors argue that research to reduce this uncertainty is itself highly valuable. They estimate the value of this information at up to $10 trillion.10 These estimates rely on a simplified model with a single global decision-maker and are sensitive to key assumptions. Although the authors caution that these estimates could be off by a factor of three or more, they argue the core conclusion is robust to this uncertainty: bias is costly in both directions.11
We chose to investigate the role of philanthropy in advancing SRM governance and research based on an assessment of scale (high), feasibility (medium), and funding need (high) , as described in our Climate Risk Management . Our work on SRM targets the quality of decision-making around it, not any particular decision. We do not advocate for deployment, since the risks of it going badly are real and impossible to eliminate, and the evidence base is far too thin. But we also do not want a premature ban to foreclose the field and prevent the research needed to make good judgments later.
Good governance matters for three reasons. First, it guards against moral hazard—the risk that pursuing SRM could weaken the incentive to cut emissions—by keeping mitigation and SRM discussions institutionally linked so that interest in SRM does not translate into reduced pressure for emissions cuts. It also keeps mitigation the priority by framing SRM as a last-resort supplement rather than a substitute. Second, it lowers the risk of physical harm from SRM itself, since poorly designed, unilateral, or abruptly halted deployment could lead to climate harms that governance is meant to anticipate and prevent. Third, it builds toward a more just process. SRM research today is fragmented and dominated by a small set of well-resourced actors in high-income countries (HICs), so investing in more inclusive, transparent governance now means that any eventual decision is made deliberately and with broader legitimacy.12 Timing matters because governance takes years to build, while interest in SRM is growing quickly.
Because SRM remains controversial and poorly understood, we believe philanthropy's clearest role is to open political pathways for research and governance—by engaging the institutions, funders, and individuals who can legitimize the research field. This groundwork can help build the research governance infrastructure SRM currently lacks, increasing the likelihood that future decisions are transparent, just, and well-informed. We also see value in strengthening low- and middle-income countries’ (LMICs) capacity to shape SRM research and governance, since LMICs will likely shape future governance and deployment.
The rest of this report describes our assessment of the field and where we think philanthropy can improve how future decisions are made.
Refining Our Research Scope We categorized SRM interventions into four approaches: space-based, stratospheric, cloud-based, and surface-based.13
Space-based approaches, such as space mirrors, would involve placing reflective surfaces in space between Earth and the Sun to block some incoming sunlight.Stratospheric approaches center on stratospheric aerosol injection (SAI), which would increase the number of small reflective particles in the stratosphere to boost the reflection of incoming sunlight.Cloud-based approaches take three proposed forms. Marine cloud brightening would add particles to the lower atmosphere to increase the reflectivity of low-lying clouds over particular regions of the ocean. Mixed-phase cloud thinning would target clouds containing both ice crystals and water droplets, thinning them to reduce the amount of heat they trap. Cirrus cloud thinning would modify the properties of high-altitude ice clouds to increase the atmosphere's transparency to outgoing thermal radiation.Surface-based approaches would take four broad forms: urban, agricultural, desert, and ocean. Reflective roofs and crops would increase surface albedo, an approach that is global in theory but produces only modest cooling even at scale.14 Increasing desert albedo with reflective materials would cool the areas where it is applied, but modeling suggests it would severely disrupt rainfall across the tropics and subtropics.15 Ocean albedo change would increase the surface’s reflectivity using microbubbles or reflective foam, with effects confined to particular regions.We ruled out space-based interventions on feasibility grounds and surface-based interventions on scale grounds.16 We therefore focus our assessment on stratospheric and cloud-based approaches. In general, stratospheric and cloud-based approaches differ in mechanism, geographic extent, longevity, and technological readiness. Theoretically, a deployment scenario could combine SRM approaches to leverage their distinct benefits. However, little to no research has examined their benefits, risks, or optimal mix of combined approaches.17
Stratospheric Aerosol Injection How SAI would work: In a hypothetical global deployment, an actor would inject highly reflective aerosols or their precursors at an altitude of at least 20 kilometers, or about 12 miles, above the subtropics in both hemispheres.18
Materials: Sulfate is the most studied option, since large volcanic eruptions inject sulfur dioxide (SO2 ) gas into the stratosphere, giving researchers a real-world precedent to study.19 However, sulfate aerosols carry known risks, such as contributing to ozone depletion, which allows more harmful ultraviolet radiation to reach the Earth. Researchers have therefore considered other substances such as calcite (calcium carbonate), titanium dioxide, and alumina in hopes of reducing those risks.20
Cooling effects: Climate models suggest that, hypothetically, well-designed and incremental SAI deployment could fully or partially offset the mean increase in Earth’s temperature, though the actual effect would depend heavily on the deployment scenario and injection strategy.21 Factors that affect cooling include the type of injected material, the pace at which the material converts to sulfate aerosols, and where and when the material is added.22 These sensitivities affect cooling in two ways: the choice of aerosol determines how effectively it scatters sunlight and how long it persists in the stratosphere, while the timing and location of injection determine how stratospheric circulation patterns transport and disperse the aerosols. Because these variables interact, it is hard to say precisely how much material is needed to achieve a given amount of cooling. In principle, SAI might be scaled up to cool the Earth by 2-5°C, but each additional increment of aerosol produces diminishing cooling returns.23
Duration and reversibility: Once injected, aerosols are expected to remain in the stratosphere for one to three years, and SAI could measurably reduce warming within one year of deployment.24 However, SAI’s cooling is impermanent. Without continued injection, its cooling effects would fade, and global temperatures would revert to previous levels, as if SAI had never been deployed.25 SAI would therefore have to be maintained until GHG emissions reach safe levels. One study modeling 355 emissions reduction scenarios found that SAI deployment, alongside emissions cuts, might need to be maintained for over 100 years to limit global warming to 1.5°C.26
Real-world evidence: SAI is the most studied and best-understood SRM approach.27 However, no scientific outdoor experiments have been undertaken to understand SAI’s effects.28 Instead, researchers have relied on natural events as a proxy, with the 1991 Mount Pinatubo eruption as the most frequently cited case study. This eruption injected about 17 million metric tons of SO2 into the lower stratosphere and is estimated to have cooled global temperatures by up to 0.5-0.6°C for three years.29 Volcanic analogies suggest SAI’s effects would likely be uneven and particularly harmful if injections were hemispherically asymmetric.30
Cost: According to Smith (2020), a hypothetical SAI program beginning in 2035 and running through 2100 would cost between $250 billion and $2.5 trillion in aggregate, depending on the Representative Concentration Pathway (RCP) and the cooling target.31 By 2100, annual costs would range from about $7 billion (halving warming under RCP4.5) to about $72 billion (reversing to 2020 levels under RCP8.5).32 Across all nine scenarios, the annual cost to suppress 1°C of warming is within 10% of $18 billion.33 We consider these figures lower-bound estimates because they exclude indirect costs like monitoring and measuring the impacts of SAI.34 Bronsther & Xu (2025) argue that the true “price” of SRM would likely be much higher than its technical cost given SRM’s political and social challenges; they estimate that damages from SAI’s side effects would fall between $0 and $809 billion annually.35
As a point of comparison, Liu et al. (2023), a separate study with its own assumptions, estimates that an SAI program starting in 2045 in a world with moderate emissions would avoid $169 trillion—or about $7 trillion annually—in expected climate damage by 2069.36 This estimate combines SAI with a fixed emissions trajectory representing pledged mitigation, and does not model any feedback between SRM and mitigation ambition. As a result, it captures neither the risk of moral hazard nor the possibility that mitigation could instead be accelerated beyond current pledges, either of which would change SRM’s estimated value.
Who would deploy SAI: According to Horton, Smith, and Keith (2025), the high technical requirements of a “Planetary, Large-scale, Uninterrupted, and Speedy” (PLUS) deployment limit potential actors to around 10 states.37 Narrowing further to account for geopolitical requirements, the same analysis concluded that only the U.S. and China have the capabilities needed to overcome external opposition and implement a PLUS deployment unilaterally.38 The barrier to entry is much lower at a small scale, however. A single state or small coalition could likely deploy SAI at a small scale within a few years using only current technology, well short of the PLUS threshold described above.39
Cloud-Based Interventions Researchers have proposed several techniques that modify clouds to reduce net radiative forcing (Appendix ). These methods are generally earlier-stage than SAI, with most confined to modeling studies or small-scale field experiments. Cloud-based approaches were originally proposed, and remain most commonly discussed in the literature, as a method for global cooling.40 A smaller, more recent strand of research has explored localized, temporary cooling, such as using marine cloud brightening to shade coral reefs during heat waves and weaken El Niño’s effects.41 The feasibility of marine cloud brightening for global cooling remains disputed.
Deployment Risks Across SRM approaches, the effects of deployment are largely unknown and could pose major risks to human and natural ecosystems.42 Because Earth’s systems are complex and connected, our ability to predict or control SRM’s effects is limited.43 Research on SRM’s effects, their regional variability, and the appropriate methods to assess tradeoffs associated with their development and deployment remains nascent.44 Some deployment risks include:
Effects on ecosystems, agriculture, health, and water: Evidence is limited and conflicting on the effects of SRM on natural and human ecosystems, agriculture, health, or water resources.45 For example, one study suggests that SAI may increase some crop yields when compared with a business-as-usual scenario due to scattered sunlight reaching more plants in dense areas.46 Other studies indicate that SAI could negatively affect crop productivity and nutritional value by altering precipitation patterns and reducing sunlight.47 Confidence around SAI’s effects on agricultural yields remains low.48 Regional tradeoffs: Existing assessments suggest both localized and global interventions would affect regions unevenly.49 SRM’s regional effects are still largely uncertain.50 Studies suggest SAI would reduce the magnitude of climate change in most locations, while altering hydrological patterns in a few.51 These uneven effects could create regional winners and losers, raising concerns about social welfare, equity, and justice.52 Because these effects could be traced back to a deploying actor’s choices, they also risk complicating international cooperation by giving states new grounds for interstate blame and conflict.53 Suitability to address climate tipping points: Little is known about how SRM would interact with tipping point drivers.54 There is early consensus that SRM would be ineffective as an emergency response to tipping points, but SRM’s possible role in preventing tipping points remains uncertain.55 Termination shock: If SRM were abruptly halted—removing SRM’s masking effect on warming—temperatures could rise far faster than they would under climate change alone (termination shock). This rapid warming could severely affect biodiversity hotspots.56 One modeling study found that most of the cooling from SRM would be lost within a decade of an abrupt halt, and that a phase-out would need to last roughly 30 years to avoid a rapid rebound.57 Because ongoing emissions-driven warming makes phase-outs longer and impacts emerge faster, the study highlights emissions mitigation as key to managing this risk.58 Uncertainties and risks include how much warming SRM deployment would actually be masking by the time of any halt and the potential to prevent or slow the cessation of an SRM program.Unilateral or unauthorized deployment: Much of the literature assumes a state facing severe climate harms would weigh the benefits of unilateral SRM deployment against globally shared risks.59 However, political scientists argue this framing is too narrow. Deployment could just as plausibly be driven by motives unrelated to climate vulnerability, including geopolitical competition.60 This wider set of self-interested, competitive, or security deployment scenarios remains under-examined relative to more idealized climate modeling literature.61 We think the implications of unilateral deployment grow substantially in a large-scale deployment scenario, where the climate effects would be significant enough to draw international scrutiny. Whether this leads to conflict or to cooperation may depend more on how other countries react than on the original actor’s motive.62 Regional rivalry and attribution disputes: SAI has little ability to target specific regions. The main exception is that deploying more in one hemisphere than the other shifts tropical rainfall patterns, which could favor some regions over others. Cloud-based approaches are inherently more regional, but even these smaller-scale interventions may have effects that reach beyond their target areas.63 Since scientists likely will be unable to prove exactly how much any single deployment affected a given region's weather, we think disputes over blame and harm could persist for decades without a clear resolution.-----
1 “In most proposed SRM approaches, a small additional amount of sunlight is reflected to space, or more of Earth’s infrared radiation is allowed to escape to space (...).” United Nations Environment Programme (UNEP), 2023
2 Figure 1. Proposed methods for modifying Earth’s radiation budget, including SRM methods and cirrus cloud thinning. National Oceanic and Atmospheric Administration (NOAA), 2024
3 “SRM deployment is the only known approach that can (i) cool the planet within a few years, or (ii) limit rates or amounts of temperature increase (i.e. to meet the Paris Agreement target of well below 2°C, preferably 1.5°C), if insufficient mitigation continues (emission reduction and GHG removal).” UNEP, 2023
4 “An intervention resulting in 0.5 or 1°C of cooling would require hundreds of aircraft, and this fleet build-out itself would span another decade or more.” Smith, 2024
5 Acidification: “SRM is not a substitute for mitigation. Impacts from excess carbon dioxide (CO2 ), such as ocean acidification and ecological degradation, would continue.” UNEP, 2023
Ozone: “Based on the observed cooling after large volcanic eruptions and various model studies, stratospheric aerosol injection (SAI) has the potential to reduce global mean temperatures. However, SAI cannot fully offset the wide-spread effects of global warming and produces unintended consequences, including effects on ozone.” NOAA, 2022
6 See Table 1 in Futerman et al., 2025
7 “Because of these differences in climate forcing and response patterns, the regional and seasonal climates of a world with a global mean warming of 1.5°C or 2°C achieved via SRM would be different from a world with similar global mean warming but achieved through mitigation (MacMartin et al. 2018).“ Intergovernmental Panel on Climate Change (IPCC) Sixth Assessment Report, 2022
8 “Our results thus suggest that deploying SRM alongside emissions cuts and CDR would substantially accelerate the end of temperature overshoot.” Boselius, Duffy, & Irvine, 2025
9 “Comparing across the unbiased diagonal, we find that over-optimistic bias about SG – believing SG is more efficacious than reality – can be more costly than over-pessimistic bias. However, both directions of bias can have large costs. For example, the cost of over-optimistically believing SG efficacy is 0° when it is really 45° is around $115 trillion and the cost of over-pessimistically believing SG efficacy is 45° when it is really 0° is lower but still large at around $95 trillion.” Harding et al., 2021
10 “We estimate the value of information that eliminates uncertainty about SG efficacy is up to roughly $10 trillion, comparable to the value of eliminating uncertainty about climate sensitivity.” Harding et al., 2021
11 “Our model has many potential order-unity uncertainties, so our estimates could be uncertain by a factor of three or more. But, even with this uncertainty, and even if research reduces uncertainty by only one-fifth, the value of reduced uncertainty about SG is about a thousand-fold larger than the cost of a SG research programme.” Harding et al., 2021
12 See Figure 1 in Biermann & Möller (2019)
13 Unless indicated otherwise, we use definitions from “BOX S.1 Solar Geoengineering Strategies Considered in This Study.” National Academies of Sciences, Engineering, and Medicine (NASEM), 2021
Mixed-phase cloud thinning: “After a few cloud droplets freeze, the resulting ice particles grow quickly due to the Wegener–Bergeron–Findeisen process (Heymsfield et al 2011, Findeisen et al 2015) and scatter less solar radiation compared to cloud droplets. These large ice particles fall faster than droplets, forming ice virgae or precipitation, which depletes the total water content (TWC) of the cloud. This cloud depletion results in a weaker cloud’s heat-trapping, but also a weaker solar reflectivity. Thus, we are interested in identifying MPRCs with a net positive radiative effect dominated by heat-trapping, for which glaciogenic seeding could produce a net cooling effect. We refer to this geoengineering concept as mixed-phase regime cloud thinning (MCT).” Villanueva et al., 2022
Ocean albedo change: “Increase surface albedo of the ocean (e.g., by creating microbubbles or placing reflective foam on the surface)” IPCC Sixth Assessment Report, 2022
14 "To date, studies indicate limited potential for such approaches, and the efficacy and environmental consequences of these approaches have yet to be carefully studied." NASEM, 2010
15 “Although global desert geoengineering has the potential to generate the largest global average cooling effect, this average masks the fact that most of the 1.12°C cooling is highly concentrated over the desert regions where the albedo increase is applied… For both urban and crop albedo geoengineering, only minimal shifts in precipitation occur, with the exception of equatorial Pacific regions (Figures 3b and 3c), whereas desert geoengineering induces quite extreme changes in precipitation patterns throughout the tropics and subtropics (Figure 3d).” Irvine, Ridgwell, & Lunt, 2011
16 Space-based interventions: "Due to the magnitude of space-based deployment required for such an undertaking, and the enormous cost of putting objects into orbit, these options appear impractical for addressing threats posed by climate change this century.” NASEM, 2021
17 “It should be noted that applying combinations of SAI, MCB, and CCT strategies could offer possible ways to leverage the benefits and reduce the negative consequences of each approach individually. But there has to date been little to no focused study of these combined strategies.” NASEM, 2021
18 “SAI involves injecting highly reflective sub-micron-size particles into the stratosphere, possibly through releases from aircraft that will need to reach altitudes of 20–25 km (lower stratosphere)” UNEP, 2023 . For more information on injecting SAI at multiple different latitudes, we recommend MacMartin et al., 2017 .
For information on SAI and latitude, see Henry, Bednarz, Haywood, 2024
19 “Sulfate is the most well-studied stratospheric aerosol due to its enhancement after volcanic eruptions. As a result, many SAI scenarios have been modelled using sulfate to offset different GHG trajectories set out by the IPCC.” Huynh and McNeill, 2024
20 “For instance, simulations suggest that stratospheric injection of solid materials such as alumina, calcite, or rutile (TiO2) as an alternative to sulfate would enhance shortwave RF while minimizing stratospheric ozone loss and heating.” NASEM, 2021
21 Climate model: “Climate model results indicate that an operational SRM deployment could fully or partially offset the global mean warming caused by anthropogenic GHG emissions” UNEP, 2023
Actual effect: “As indicated by climate model results, a well-designed SRM deployment that ramps in over time may reduce rates or amounts of surface temperature increase, and reduce some changes to the hydrological cycle associated with climate change across most regions.” UNEP, 2023
For examples of different types of cooling scenarios, we recommend the section, “Implementation of SAI” and Figure 2 from Zarnetske et al., 2021 .
22 “As illustrated in several studies, simulations of the relationship between RF and the amount of sulfur added (i.e., the resulting sulfur burden and the ratio of surface area to volume produced) are sensitive to several factors, including: how the sulfur is added (H2S versus SO2 versus sulfate);
the oxidation rate of the SO2; the microphysical description of the gas to particle conversion, particle coagulation, and sedimentation; changes in the large-scale dynamics of the stratosphere (Kleinschmitt et al., 2018; Marshall et al., 2019); and the altitude (Dai et al., 2017; Tilmes et al., 2018b), latitude (Dai et al., 2017; Tilmes et al., 2017), and season of addition (Visioni et al., 2019).” NASEM, 2021
23 “An operational SAI deployment could be scaled up to produce global cooling of 2–5°C, albeit with diminishing returns at higher rates of injections.” UNEP, 2023
24 “The aerosols released by SRM deployments would persist in the stratosphere for 1–3 years for SAI.” UNEP, 2023
“In contrast to the longer timescales of mitigation, an SRM deployment could produce a substantial reduction in radiative forcing in as little as one year.” UNEP, 2023
25 “The cooling effects of proposed SRM options would start to diminish as soon as the SRM deployment is halted.” UNEP, 2023
26 “In all pathways consistent with extrapolated current ambition, SRM deployment would exceed 100 years even under the most optimistic assumptions regarding climate response.“ Baur et al., 2023
27 “SAI is the most studied and best understood of the SG approaches proposed to date.” NASEM, 2021
28 We do not consider previous SAI outdoor experiments to have been scientific. Actual outdoor experiments explicitly aimed at studying SRM have been (1) low-altitude and only loosely analogous to stratospheric conditions (Yuri Izrael), (2) commercial rather than scientific (Make Sunsets), (3) without active materials (SCoPEx), or (4) conducted without producing peer-reviewed scientific output (SATAN).
29 Tons of SO2 : “Eruption columns reached 40 kilometers in altitude and emplaced a giant umbrella cloud in the middle to lower stratosphere that injected about 17 megatons of SO2 (...)” Self et al., 1993
Cooling: “Effects on climate were an observed surface cooling in the Northern Hemisphere of up to 0.5 to 0.6°C (...). The optically dense Pinatubo aerosol cloud caused marked changes in the amount of radiation reaching the Earth's surface; in turn, these changes affected weather and climate over the past 3 years following the eruption.” Self et al., 1993
30 “Thus, our results indicate that much like the response to volcanoes, asymmetrically loading one hemisphere with stratospheric aerosols could induce drought or increased rainfall in the Sahel in subsequent years.” Jacobson et al., 2020
31 “Depending upon the scenario analyzed, aggregate costs for SAI through the remainder of the century can range from roughly $250 billion to nearly $2.5 trillion, with an annual budget in the year 2100 of $7 to $72 billion (all in 2020 USD).” Smith, 2020
32 See Table 6. Cost impact of scenario choices. Smith, 2020
33 “What remains remarkably constant however is the annual cost to suppress 1 °C of warming, which remains within 10% of $18 billion irrespective of the scenario chosen.” Smith, 2020
34 “Moreover, this analysis also excludes any indirect costs for damages that might accrue either from SAI or from climate change itself. Those could be orders of magnitude larger than the aggregate program budget.” Smith, 2020
35 “However, SRM comes with at least four political and social challenges that are not accounted for in such calculations. First, best estimates from Irvine et al. indicate that SAI will harm 0% to 7% of the world’s population as a side-effect12, such that those people would demand compensation from the technology’s winners. Using existing (and conservative) metrics to calculate the amount of side-effect harm for SAI, we estimate that such harms would fall between $0 and $809 billion annually.” Bronsther & Xu, 2025
36 We calculated the total avoided climate damage by subtracting climate damage under scenario “SAI-1.5-2045” from that under scenario “SSP2-4.5.” We divided the total by the number of years of deployment to annualize the avoided damage. See Table 1: NPVs of climate damages, adaptation, and mitigation costs, and total economic effects. Liu et al., 2023
37 “Thus, only a few states—including the US, France, Germany, the UK, Japan, Canada, Russia, China, India, and Brazil—appear capable of developing the fleet of aircraft numbering in the hundreds that would be required to reduce global temperatures by 1°C by midcentury.” Horton, Smith, & Keith, 2025
38 “The fact of bipolarity means that the US and China alone will be sufficiently powerful across multiple domains to both successfully implement a PLUS deployment on a unilateral basis and prevent other non-rival but technically capable states from doing so.” Horton, Smith, & Keith, 2025
39 “Provocative unilateral action could also take the form of a deployment much smaller than the one in this full-scale scenario, placing on order one-tenth as much sulfur or less. Such subscale deployment, using current technology, would be within reach of many more actors and could be started within a few years… The change in climate achievable in this way would not be detectable above natural variability at national scale, so deployers could argue that this was consistent with avoiding trans-boundary harm under customary international law.” Parson and Keith, 2024
40 The concept of marine cloud brightening was introduced in Latham, 1990 , which framed it as a tool for controlling global warming.
41 The Australian Reef Restoration and Adaptation Program’s work on marine cloud brightening is representative of newer, more localized applications.
El Niño: “By exploiting a unique opportunistic experiment provided by the 2019–2020 Australian wildfires, we show that both the modeled cloud brightening and ensuing La Niña–like response can be reproduced by simulating MCB in the SESP.” Wan et al., 2026
42 Unknown effects: “The extent to which these adverse consequences might manifest and to which they might be mitigated or avoided is largely unknown.” UNEP, 2023
Risks: “SRM would be a new and intentional interference in the climate system with potentially dangerous side-effects and risks to human society and ecosystems.” UNEP, 2023
43 “These damages would likely be unevenly distributed across nations. The extent to which these adverse consequences might manifest and to which they might be mitigated or avoided is largely unknown.” UNEP, 2023
44 “A final reason for caution is that for most of these impact areas, there have been very few published studies (only one or two in some cases), and even the methodologies of how to study some of these impact areas are in nascent states.“ NASEM, 2021
45 “Despite a growing literature on the climate response to solar geoengineering—proposals to cool the planet by increasing the planetary albedo— there has been little published on the impacts of solar geoengineering on natural and human systems such as agriculture, health, water resources, and ecosystems.” Irvine et al., 2017
46 “Under SAI, the positive effect of increased diffuse radiation on yields largely compensates for the loss caused by reduced direct sunlight, and in some cases contributes to increased yields.” Fan et al., 2021
47 Precipitation: “Crop productivity is also projected to be reduced where SAI reduces rainfall relative to the scenario without SRM, including a case where reduced Asian summer monsoon rainfall causes a reduction in groundnut yields (Xia et al. 2014; Yang et al. 2016).” IPCC Sixth Assessment Report, 2022
Nutritional value: “Compared to SSP2-4.5, the reduced surface temperatures under SAI tend to decrease protein in maize and rice, with milder impacts on wheat and soybean.” Clark et al., 2025
Reduced sunlight: “SAI will increase the fraction of diffuse sunlight, which is projected to increase photosynthesis in forested canopy, but will reduce the direct and total available sunlight, which tends to reduce photosynthesis.” IPCC Sixth Assessment Report, 2022
48 “Because of the divergence in results from the studies, including change in crop yields and plant productivity, overall confidence is low.” UNEP, 2023
49 “Regional disparities in climate hazards may result from both regionally-deployed SRM options such as GBAM, and more globally uniform SRM such as SAI (Jones et al. 2018; Seneviratne et al. 2018).” IPCC Sixth Assessment Report, 2022
50 “These damages would likely be unevenly distributed across nations. The extent to which these adverse consequences might manifest and to which they might be mitigated or avoided is largely unknown” UNEP, 2023
51 “We found that stratospheric aerosol geoengineering deployed to halve warming might substantially reduce the overall magnitude of climate change, while exacerbating hydrological change in only a small fraction of places.” Irvine & Keith, 2020
52 “Like climate change, SG interventions would impose heterogeneous impacts across the world and over time, which would have important social welfare, equity and justice, social, and political implications.” Aldy et al., 2021
53 “Effective climate mitigation could be made more difficult by high-leverage climate engineering technologies that introduce intentionality and shorter chains of causation to climate politics…This could complicate ongoing efforts to mitigate and adapt, not least by promoting adversarial friend–enemy logics that sour international relations and make collective action on mitigation (and other collective global problems) more difficult, as well as making climate engineering solutions themselves more costly and difficult to implement.” Corry, 2017
54 “The linkages between different kinds of solar geoengineering and the drivers of tipping points are understudied and uncertain.” Global Tipping Points Report, 2023
55 “There is already early consensus that geoengineering techniques would not offer an emergency response to anticipated tipping events (Horton, 2015; Lenton, 2018). However, assessment over whether they might provide pre-emptive measures to support prevention is ongoing, and heavily contested (Gupta et al., 2020).” Global Tipping Points Report, 2023
56 “Sudden termination of SG was found to cause extremely rapid temperature velocities for both land and ocean environments (far exceeding the values predicted for future climate change without SG)—thus placing much greater migration pressure on most species—with particularly acute effects seen in “hotspots” for biodiversity including the tropical oceans, the Amazon basin, Africa, Eurasia, and the polar regions.” NASEM, 2021
57 Lost cooling: “First, as shown in figure 1, the termination of SAI causes around 75% of its cooling to be lost within a decade. However, this percentage is not a fixed number. If SAI is deployed for a long time, as it is in figure 4(h), SAI is slower to lose its cooling in a termination. The most-sustained cooling emulated (320 years) loses only 50% of its cooling in a decade, rather than 75%.” Farley et al., 2024
Phase-out length: “For a phase-out beginning in 2055 under our scenario to meet these safety criteria, it would be approximately 30 years long.”
58 “Since the risks of potential inconsistencies in SAI deployment are linked to the size and duration of an inconsistency, mitigation of GHG emissions is an important factor of risk management during SAI deployment.” Farley et al., 2024
59 “Current literature focuses on actors that are heavily impacted by climate change as the most likely contenders to develop and use the technology…” Young, 2023
60 “In climate-oriented governance parlance, SRM engenders a ‘free-driver’ logic because any capable actor could in theory attempt to engineer a preferred climatic state. For security-related actors, drivers of unilateral development (or use) resided much more in distrust towards (and competition with) other countries than climate risks.” Corry, McLaren, Kornbech, 2024
61 “As a foundation for negotiating governance—whether to enable or constrain the development and deployment of such technologies—such projections are dangerously incomplete and potentially misleading. They tell us little about climatic responses to nonideal—self-interested, malicious, or competing—deployments, and even less about political or cultural responses outside of the climate system.” McLaren, 2025
62 “The main threat from unilateral intervention is the provocation it presents to other states and the resultant risks of destabilization or conflict. Other states will surely object, for various reasons and with varying intensity, ranging from pro forma objection to firm determination to resist. The severity of geopolitical threat from this narrative will mainly depend on these responses by other states, especially major powers.” Parson & Keith, 2024
63 “For any given average cooling, spatially uneven changes to radiative forcing, either by strong north–south imbalance of SAI or by patchy use of MCB or CCT, would likely exacerbate climate changes over larger areas than an even distribution with the same average cooling.” Parson & Keith, 2024
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
%20(1).png)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
.webp)
