Albedo enhancement by stratospheric sulfur injections: More research needed



Research on albedo enhancement by stratospheric sulfur injection inspired by Paul Crutzen's paper a decade ago has made clear that it may present serious risks and concerns as well as benefits if used to address the global warming problem. While volcanic eruptions were suggested as innocuous examples of stratospheric aerosols cooling the planet, the volcano analog also argues against stratospheric geoengineering because of ozone depletion and regional hydrologic responses. Continuous injection of SO2 into the lower stratosphere would reduce global warming and some of its negative impacts, and would increasing the uptake of CO2 by plants, but research in the past decade has pointed out a number of potential negative impacts of stratospheric geoengineering. More research is needed to better quantify the potential benefits and risks so that if society is tempted to implement geoengineering in the future it will be able to make an informed decision.

At the Fall American Geophysical Union Meeting in 2005, the buzz going around was, “Did you hear about the paper that Paul Crutzen is writing about geoengineering?” My first reaction was, “What is geoengineering?” I wrote to Paul for a copy of the draft, found the idea very interesting, and after reading it, asked him if he was sure he wanted to publish this. Of course, the answer was “Yes” and of course he was right to do it [Crutzen, 2006]. Crutzen stimulated many, including myself, to get involved in geoengineering research. I was intrigued and began working on it, specifically the idea of the creation of an artificial stratospheric sulfate aerosol cloud to emulate those created by large volcanic eruptions.

At my first meeting on this topic, the Managing Solar Radiation Workshop at NASA Ames Research Center, Moffett Field, California, November 18–19, 2006, I was amazed and shocked to find so many engineers and physicists enamored of this idea, and ended up writing down 20 reasons why it might be a bad idea [Robock, 2008]. The hubris of some, who thought that this was just a mechanical or physical problem to solve, and the lack of awareness of the science of climate change and the natural chaotic variability of climate, was very scary. A number of those potential risks were already understood 10 years ago, and were discussed by Crutzen and in the accompanying essays, particularly by MacCracken [2006], but work in the past decade has produced much better understanding and identification of those risks, in particular that temperature and precipitation cannot both be controlled at the same time [e.g., Jones et al., 2013], that summer monsoon precipitation would be reduced [Tilmes et al., 2013], that even if global average temperature could be kept from increasing, there would be cooling and warming in different places [Kravitz et al., 2013a], and that ice sheets melt from the bottom, and changing insolation would not be very effective at slowing their melting [McCusker et al., 2015]. The history of past weather and climate modification attempts provides strong lessons about the difficulty of governance and the dangers of military applications [Fleming, 2010].

My research has led me to summarize what we know, as a list of five potential benefits of stratospheric geoengineering and 27 concerns and risks, and is shown in Table 1, updated from Robock [2008, 2014]. The number of items on each side of the list was never meant to be a metric for deciding whether to ever implement stratospheric geoengineering. It would be possible to produce a list with an equal number of benefits and risks, but each would have different levels of importance. In the current list, items listed under both benefits and risks differ in specificity, scope, and granularity.

Table 1. Risks or Concerns and Benefits of Stratospheric Geoengineering, Updated From Robock [2014]
BenefitsRisks or Concerns
  1. Please also see Robock [2008] for explanations of most items.
1. Reduce surface air temperatures, which could reduce or reverse negative impacts of global warming, including floods, droughts, stronger storms, sea ice melting, and sea level risePhysical and biological climate system
1. Drought in Africa and Asia
2. Perturb ecology with more diffuse radiation
2. Increase plant productivity3. Ozone depletion
3. Increase terrestrial CO2 sink4. Continued ocean acidification
4. Beautiful red and yellow sunsets5. May not stop ice sheets from melting
5. Unexpected benefits6. Impacts on tropospheric chemistry
6. Prospect of implementation could increase drive for7. Rapid warming if stopped
mitigationHuman impacts
 8. Less solar electricity generation
 9. Degrade passive solar heating
10. Effects on airplanes flying in stratosphere
 11. Effects on electrical properties of atmosphere
 12. Affect satellite remote sensing
 13. Degrade terrestrial optical astronomy
 14. More sunburn
 15. Environmental impact of implementation
 16. Whiter skies
 17. Affect stargazing
 18. Human error during implementation
 19. Unexpected consequences
 20. Cannot stop effects quickly
 21. Commercial control
 22. Whose hand on the thermostat?
 23. Societal disruption, conflict between countries
 24. Conflicts with current treaties
 25. Moral hazard—the prospect of it working could reduce drive for mitigation
 26. Military use of technology
 27. Moral authority—do we have the right to do this?

In fact, item number 1 on the benefits side, that stratospheric geoengineering could reduce global warming and many of its negative impacts, may be so important that society in the future may decide to implement stratospheric geoengineering to reduce some amount of warming and live with and adapt to the negative consequences of geoengineering. (The only rational way to do this would be for a limited amount of time while mitigation and carbon dioxide removal from the atmosphere reduce the radiative forcing from greenhouse gases.) Each of the potential benefits and risks needs to be quantified so that society can make informed decisions in the future about how much and what type of geoengineering to implement and for how long.

Some of the topics in Table 1 can be addressed by climate modeling. With Ben Kravitz and others, I have started the Geoengineering Model Intercomparison Project [GeoMIP,; Kravitz et al., 2011, 2013b, 2013c, 2015; Tilmes et al., 2015], in which various scenarios of anthropogenic stratospheric aerosols, marine cloud brightening, and cirrus thinning are being evaluated with climate model experiments as a response to global warming. So far we have had six annual international workshops [Robock et al., 2011; Kravitz et al. 2012, 2013d, 2014, 2016a, 2016b], produced a special section of Journal of Geophysical Research—Atmospheres with 15 papers, a current special combined issue of Atmospheric Chemistry and Physics and Geoscience Modeling Development that is now accepting submissions, and a robust international modeling community conducting standardized climate model experiments, with 36 peer-reviewed GeoMIP publications so far ( The new Coupled Model Intercomparison Project 6 [CMIP6; Meehl et al., 2014] requested additional focused experiments, and in July 2015, GeoMIP6 (named to coincide with CMIP6 nomenclature) was formally made a CMIP6-Endorsed MIP. The GeoMIP6 experiments to be conducted are described by Kravitz et al. [2015] and Tilmes et al. [2015] and in Table 2, where the experiments that international modeling groups have agreed to carry out over the next several years are shaded in yellow. In addition to the standard experiments, GeoMIP6 also establishes a GeoMIP Testbed for new experiments to be conducted by one or a few climate models as demonstration projects for future possible model intercomparisons.

Table 2. All Core GeoMIP Experiments up to This Point, Including the Additional Proposed GeoMIP6 ExperimentsThumbnail image of

Some of the issues in Table 1 can be studied by looking at the analog of volcanic eruptions [Robock et al., 2013], but some cannot be addressed at all by scientific investigation. In 2012, I thought that the governance problems, some of which were discussed by MacCracken [2006], would be insoluble and that stratospheric geoengineering will never be implemented by international agreement [Robock, 2012a], and have yet to change my mind. In fact the more we look at stratospheric geoengineering, the more unlikely implementation becomes because of the associated risks. In particular, risks associated with unknowns, governance, and ethics (18–27 in Table 1) will be very difficult to address. Nevertheless, much is still unknown, and we have an obligation to continue the research.

The ethics of doing geoengineering research also needs to be addressed. Both Lawrence [2006] and Cicerone [2006] made a clear case that we have an obligation to better understand the benefits and risks of potential geoengineering deployment so that policymakers in the future, should they be tempted, would be able to make informed decisions. I agree [Robock, 2012b], provided that outdoor small-scale experiments are subject to environmental regulation and governance. However, as discussed by Robock et al. [2010], large-scale experiments would have to be conducted for decades to distinguish the signal of small injections from the noise of weather and climate variations. This would be no different from actual geoengineering implementation. Furthermore, only by injecting SO2 into an existing sulfate aerosol cloud could the growth of aerosols be studied. Perhaps, after the next large volcanic eruption, this could be tested on part of the cloud, but that would require development of monitoring equipment that could follow the air parcel.

The American Meteorological Society policy statement on geoengineering [American Meteorological Society, 2009], which was subsequently adopted by the American Geophysical Union [2009], recommends “Enhanced research on the scientific and technological potential for geoengineering the climate system, including research on intended and unintended environmental responses.” Strong recommendations for geoengineering research have also come from Keith et al. [2010], Betz [2012], and Government Accountability Office [2011]. The recent U.S. National Academy of Sciences report [McNutt et al., 2015] recommends “an albedo modification research program be developed and implemented that emphasizes multiple-benefit research that also furthers basic understanding of the climate system and its human dimensions.” Yet a U.S. national geoengineering research program has yet to materialize. Now that the stigma of doing the research is over, it would be relatively cheap to evaluate the many suggested techniques, by continued computer modeling and study of analogs, and also by conducting small outdoor experiments, as recommended by Crutzen.

Crutzen started an international research effort on geoengineering, yet much more remains to be learned. All scientists working on geoengineering that I know of make a strong call for mitigation and adaptation to address global warming, and this is also the recommendation of the U.S. National Academy of Sciences report [McNutt et al., 2015]. In fact, a rapid transition to solar and wind power can keep global warming close to the 2015 Paris goal of 2°C above pre-industrial levels [e.g., International Energy Agency, 2016]. So far geoengineering research concludes that there is no safe Plan B, and provides enhanced support for mitigation and adaptation. Additional research support for these efforts will make clear over the next decade whether this current understanding is robust, and it would be irresponsible for the United States and other nations not to make this investment in research.


This work was supported by NSF grants AGS-1157525, GEO-1240507, and AGS-1617844. This paper does not analyze any new data.