1 Introduction

Despite substantial progress in clean technology and increasing policy ambition, the world remains off track to hold warming to the Paris Agreement targets of well below 2 °C and, aspirationally, no greater than 1.5 °C above pre-industrial temperatures (United Nations Environment Programme 2022). In light of this, a growing number of scientists and advocates—including the authors (Diamond et al. 2022; Wanser et al. 2022)—have called for expanding research into solar climate interventions that would utilize tiny solid or liquid particles suspended in the atmosphere (aerosols) to reflect more sunlight away from Earth and thus offset some of the effects of global warming due to greenhouse gas emissions from fossil fuel burning and deforestation (Crutzen 2006; NASEM 2021; United Nations Environment Programme 2023).

Notwithstanding the relatively early state of solar climate intervention research, at least one company has already been founded to market “cooling credits” that purport to offset a given quantity of greenhouse gas emissions with the emission of aerosols or their precursor gases (Temple 2022). In the near term, the science of solar climate intervention is simply too uncertain for such credits to be meaningful market instruments. More fundamentally, however, many of the climatic and environmental effects of greenhouse gas emissions are incommensurate with those from aerosol injections. If solar climate interventions are to be part of society’s portfolio of responses to climate change, they must be complements to, not substitutes for, mitigation.Footnote 1 A market approach to solar climate intervention based on uncoordinated “cooling credits” is not a viable climate solution now and is unlikely to ever be in the future.

2 Uncertain efficacy and inadequate monitoring

A conceit behind “cooling credits” is that they can serve as suitable substitutes for carbon credits and offsets that, at least in theory [and accounting for carbon cycle responses (Zickfeld et al. 2021)], reverse the harms that would have occurred due to some quantity of emitted greenhouse gases by compensating additional mitigation efforts elsewhere or procuring a drawdown and secure long-term storage of CO2. Despite their promise, in practice, the widespread use of carbon offsets has thus far been inhibited by challenges to quantifying carbon drawdown and storage permanence supported by monitoring, reporting, and verification (Babiker et al. 2022). The uncertainties associated with quantifying the cooling from the emission of some mass of aerosol (or precursor gas) and translating this to the warming from some mass of carbon are substantially larger. Indeed, estimating how much greenhouse gas warming has been masked by present-day aerosol pollution (largely via its effects on clouds) is one of the greatest uncertainties in climate science (Forster et al. 2021). Very similar physics and chemistry challenges apply to both understanding the effect of aerosol pollution today and predicting what would happen under a hypothetical future solar climate intervention deployment.

Stratospheric aerosol injection (SAI), in which aerosols or precursor gases would be added to the upper atmosphere (Budyko 1974; Crutzen 2006), would almost certainly be able to produce a cooling effect like that observed after large explosive volcanic eruptions (Hansen et al. 1992; Robock et al. 2013). However, how much material would be necessary to produce the desired level of global mean cooling and how this would vary by injection altitude, latitude, and timing remains highly uncertain (MacMartin et al. 2017; Rasch et al. 2008; Visioni et al. 2017, 2020, 2021). The type and amount of material injected in addition to its seasonality and location would also affect potential side effects.

The cooling ability of marine cloud brightening (MCB), in which sea salt would be sprayed into low-lying clouds to make them more reflective and potentially longer lasting (Conover 1966; Latham 1990; Latham et al. 2012), is less certain than for SAI. Aerosol-driven cloud enhancements have been clearly observed in effusive volcanic eruptions (Chen et al. 2022; Gassó 2008; Malavelle et al. 2017; McCoy and Hartmann 2015; Toll et al. 2017; Yuan et al. 2011) and other “natural experiments” (Christensen et al. 2022) like pollution tracks from international shipping (Conover 1966; Diamond et al. 2020; Durkee et al. 2000; Manshausen et al. 2022; Radke et al. 1989; Russell et al. 2013) and large industrial centers (Hobbs et al. 1980; Toll et al. 2019; Trofimov et al. 2020). Statistically significant detection of regional radiation changes (Seidel et al. 2014) has been more challenging, however, except in ideal conditions for the particularly susceptible stratocumulus cloud regime (Diamond et al. 2020). Whether substantial cooling can be routinely and predictably achieved in other cloud regimes and regions is a major uncertainty for assessing the technical feasibility of MCB (Diamond et al. 2022; Feingold et al. 2022). Questions about the proper size of injected particles also have major implications for the mass of aerosol required for a given cooling (Hoffmann and Feingold 2021; Wood 2021), and the answers will likely vary for clouds under different weather states. Seeding in unfavorable meteorology can even lead to counterproductive cloud evaporation and darkening (Y.-C. Chen et al. 2012; Coakley and Walsh 2002; Zhang and Feingold 2023).

For both SAI and MCB, major investments in monitoring would be necessary to confidently detect that an intervention was working as intended (Feingold et al. 2022; NASEM 2021). This would involve a sustained commitment to maintaining and improving the capabilities of a global observing and monitoring system for Earth’s radiation budget and atmospheric composition including, among other initiatives, expanded balloon and aircraft measurements of stratospheric properties and advances in retrieving cloud and aerosol properties from space- and ground-based sensors.

3 Incommensurate impacts of greenhouse gases and aerosols

Even if these (and many other) uncertainties are narrowed in the coming years (Wanser et al. 2022), the different natures of environmental effects from increasing greenhouse gases and the impacts of reflecting sunlight complicate direct comparisons. Solar climate interventions like SAI and MCBFootnote 2 work by reducing the amount of shortwave radiation from the sun that the Earth absorbs, whereas the greenhouse effect warms by preventing Earth’s longwave (“heat”) radiation from escaping out to space. As a result of this difference, cooling by reflecting sunlight decreases precipitation more than the same cooling from avoided greenhouse gas emissions (Bala et al. 2008) [see illustrative climate model results (Boucher et al. 2019a, b, 2020a, b; Visioni et al. 2021) in Fig. 1a, b]. A well-designed solar climate intervention could plausibly reduce both temperature and precipitation impacts of climate change simultaneously, but only by explicitly aiming not to fully offset greenhouse gas warming (Irvine et al. 2019; Irvine and Keith 2020). The mismatch between changing shortwave and longwave radiation could also alter the distribution of temperature change, for example, between day and night, between seasons, and between the tropics and the poles (Bala and Caldeira 2000; Jiang et al. 2019; Kravitz et al. 2013). The distribution of risks and benefits will therefore differ between mitigation and a solar climate intervention even for the same amount of avoided global mean warming.

Fig. 1
figure 1

Example climate changes under a high greenhouse gas emissions scenario (SSP5-8.5), a lower greenhouse gas emissions scenario (SSP2-4.5), and a scenario in which the warming from the high greenhouse gas emissions scenario is reduced to that from the lower greenhouse gas emissions scenario via stratospheric aerosol injection (G6sulfur). Global mean surface temperature (a), precipitation (b), and surface ocean pH (c) results are shown from simulations of the Institute Pierre-Simon Laplace’s IPSL-CM6A-LR climate model (Boucher et al. 2019a, b, 2020a, b) for G6sulfur (solid yellow lines), in which sulfur dioxide is injected into the stratosphere such that global mean surface temperature is maintained at the level of those in SSP2-4.5 (dashed blue lines) despite actual emissions from SSP5-8.5 (dotted red lines). Changes are shown as departures from the SSP5-8.5 2020–2030 mean values. Even though both SAI and mitigation from SSP5-8.5 to SSP2-4.5 greenhouse gas emissions result in similar temperature evolutions (a), SAI overcorrects the precipitation change (b) and does not ameliorate ocean acidification (c). See Visioni et al. (2021) for details. Although the values shown are for only one model and scenario, any solar climate intervention would produce qualitatively similar results

Full size image

Because solar climate interventions do not directly decrease the amount of CO2 in the atmosphere, they are unable to substantially ameliorate ocean acidification (Fig. 1c). On the bright side, solar climate intervention would reduce the stressor of warming, potentially increasing resiliency in the face of continued acidification. But by breaking the historic link between global temperatures, radiation, and atmospheric CO2, ecological systems may find themselves in environmental conditions for which there is no recent analogue, with as-yet unknown consequences (Zarnetske et al. 2021).

There is also a timescale mismatch between the cooling produced by aerosol interventions and warming from CO2, which can linger in the atmosphere for hundreds to many thousands of years after emission. Aerosol from an SAI deployment would remain in the stratosphere for months to years unless replenished; sea salt from an MCB deployment would leave the lower atmosphere on a timescale of days. Although it is possible, on paper, to use accounting metrics like the “global warming potential” (Forster et al. 2021) over some time period to equate long-term CO2 warming and shorter-term aerosol cooling, their effects will differ in reality and there is no obvious choice for the proper metric.

In addition to the issues above that pertain to all solar climate intervention methods, there are also risks specific to each technique. As examples, chemical and circulation effects of SAI may delay recovery of the ozone hole (Haywood et al. 2022; Tilmes et al. 2008; Tilmes​​​​​​​ et al. 2022) and the patchiness of MCB (which can only be performed where the right kinds of clouds occur) could cause circulation responses with deleterious consequences for precipitation in some regions (Bala et al. 2010; Hill and Ming 2012; Jones et al. 2009). These uncertain negative side effects, likely to vary nonlinearly with the nature of delivery and volume of material, mean that the risks and benefits of a solar climate intervention cannot be calculated simply as the sum of individual inputs. Highly coordinated or centralized activity may therefore be required to minimize risks and maximize benefits under continually evolving environmental conditions. This would run counter to the idea of a marketplace of uncoordinated individual actors with incentives primarily (or only) tied to scale.

Thus, even if a solar climate intervention were to work exactly as its deployer intends, reducing sunlight will not provide a one-to-one offset of greenhouse-gas-driven climate change. Mainstream proposals therefore tend to conceptualize solar climate intervention as a temporary measure to be wound down as mitigation and carbon dioxide removal scale up (MacMartin et al. 2018), which would be inconsistent with the widespread adoption of “cooling credits” that are not tied to the drawdown of atmospheric CO2 and may instead contribute to its continued rise. Although it is possible that some form of market mechanism may be appropriate as part of an overall coordinated strategy—for example by linking shorter-term solar climate interventions and longer-term carbon dioxide removal (Lockley et al. 2019)—it would be imperative that the solar climate interventions complement mitigation and carbon dioxide removal, not substitute for them.

4 Conclusion

Solar climate interventions may one day be critical components of the broader portfolio of climate policies to limit damages from greenhouse gas warming. If they are, however, it should not be through a “cooling credit” mechanism that is unquantifiable in the medium-term and, due to the differences between the environmental consequences of greenhouse gases and aerosols, fundamentally incompatible with the imperative to maximize safety and minimize harm. Although at least one startup has already launched (Temple 2022), policymakers, businesses, and individuals can deter such initiatives by sending a clear signal that there will be no business opportunity for such unsubstantiated “cooling credits” within carbon markets or voluntary offset initiatives now or in the future.