Reflections on a meeting about space-based solar geoengineering

By David Keith, Oliver Morton, Yomay Shyur, Pete Worden, and Robin Wordsworth

The idea that solar geoengineering might make use of space-based devices is not new. While not a common opinion, some have even seen it as preferable to other solar geoenginnering approaches. Mautner and Parks, writing in 1990  [1] claimed that “Climate engineering, which is impractical with Earth-based technology, can be achieved by space-based methods.” However, Earth-based approaches have been the focus of the discussion in the intervening 30 years and have come to be seen as technically quite straight forward. Space-based approaches have received substantially less attention, often being discussed as too expensive, demanding, fanciful or just out there. Occasionally they are specifically excluded—as they are in the charge to the current US National Academy study. For the most part they are simply left unconsidered.

The solar geoengineering community has not demonstrated significant interest in the idea. Only about 2% of articles on solar geoengineering discuss potential space-based methods in detail, and solar geoengineering attracts little interest among space technology enthusiasts. Ideas of similar or greater ambition—the establishment of industries and manufacturing in space (Bezos) or settlements on Mars (Musk)—are treated, rhetorically at least, as relatively near-term prospects, realizable in a generation or two. Justifications for such projects are frequently given in terms similar to those that might be used to justify space-based solar geoengineering; that they are a way to improve the terrestrial environment (by moving much energy intensive industry into orbit, Bezos) or preserve human civilization against calamity (through creating a back-up, Musk). But given the degree to which space enthusiasts are willing to talk of far-out prospects—including the engineering of other planetary environments, most notably through the terraforming of Mars—explicit discussion of solar geoengineering is notable by its absence.

We convened an informal meeting to consider space-based solar geoengineering issues at Harvard on November 25th, 2019. The attendees were scientists, scholars, writers, and entrepreneurs primarily from the US and Europe, with interests in both the development of space and solar geoengineering. The aim was to discuss the two communities’ priors on this subject: are space-based approaches plausible (solar geoengineering), and is geoengineering an interesting application of space-based industry (space development)? The goal was to examine whether those priors were justified and, to the extent that they might not be, ask what level of discussion and what sort of research agenda might usefully follow. These notes, written by the organizing committee, are intended to give a sense of what we came away thinking. They should not be taken as a consensus of all those who attended.

The technological challenge of space-based solar geoengineering is not one which could realistically be met within the coming decades. While various orbital arrangements have been considered, the only one that appeared reasonable to us is a screen of some form that maintains a constant position, relative to the Earth, between the Earth and the Sun. We discounted low-Earth-orbit (LEO) systems because of space debris and aesthetic concerns including flickering sunlight and night-sky pollution.

Orbital mechanics decree that the position of a stable shield would be close to that of the first Sun-Earth Lagrangian equilibrium point (L1), about 1.5 million km sunwards of Earth. Close to this pseudo-location a light-blocking shield can be kept directly between the Sun and the Earth with minimal effort. An L1 sunshield, or L1S, capable of intercepting 1% of the sunlight headed for Earth—an amount that might be expected to counter half the radiative force of a doubling of CO2—would need to have at least an area of nearly 1 million km2, which is to say around that of Egypt, but more likely be many times that size. We roughly divided L1S technologies into two groups: high-tech/low-mass and low-tech/high-mass. A low-mass L1S would use precisely engineered scattering structures that have very high mass-specific scattering efficiency as has been suggested by Angel [2] and Teller [3]. If such high-tech scatterers can achieve a sufficiently low mass per unit area, they might plausibly be manufactured on Earth and launched directly to L1.

A high-mass L1S would, in contrast, use less sophisticated scatters constructed in space using local resources. The most low-tech approach is to continuously supply dust created from asteroids, in such quantities that the amount dwelling around L1 at any given time would provide the necessary blocking (note that the dust loss rate may make this implausible). We discussed new ideas for keeping the dust confined, or at least decreasing its loss rate, including gravitational shepherding by small asteroids and methods of electrostatic of electromagnetic confinement. An alternative would be to use resources from space to build structures capable of actively maintaining position, such as solar sails which use light pressure for manoeuvring. These ideas are, at best, half baked, but they suggest there are a range of potentially important new ideas could be evaluated with little effort.

All L1S options are orders of magnitude more ambitious than anything previously attempted in space. Therefore, any research agenda would be highly dependent on the expansion of space technology particularly in the subfields of: asteroid mining, in-space manufacturing, launch (including electromagnetic launch and laser launch), and large array spacecraft communication. Given forecasts for the growth of space technology we concluded that, though the capability to do so is by no means assured, there are plausible paths of development which would allow construction of L1S to begin around midcentury. Ramping up to a 1% reduction in sunlight over the subsequent half century need not have prohibitive annual expenditures. As with all imagined approaches to solar geoengineering, this would not represent an alternative to carbon-dioxide-emissions reduction, which remains the key to reducing climate risk. But in some circumstances—such as those of a world of net-zero greenhouse-gas emissions which required significant cooling—such a shield might find plausible use.

Such a shield might be built and either if launch costs dropped to a few $100/kg or if in-space manufacturing and assembly capabilities entered a period of exponential growth. If the L1S construction is spread over decades to ramp up radiative forcing, the costs remain very large compared to the suggested costs of terrestrial approaches to solar geoengineering, such as marine-cloud brightening (MCB) and stratospheric aerosol injection (SAI). But costs on the order of $1trn are small compared to many estimates of the costs of climate change. They are also comparable to the costs of large projects undertaken by superpowers. $1trn represents roughly 50 years of NASA’s space budget; America’s F-35 fighter program has costs in the same range. Nonetheless, any serious consideration of such an expensive geoengineering project would have to carefully weigh its benefits to society versus other forms of climate change mitigation and adaptation.

We concluded that space-based solar geoengineering is not a plausible near-term goal or aspiration. But an L1 sunshield could offer advantages over other approaches to solar geoengineering, such as MCB and SAI, and the related cooling technology of cirrus cloud thinning (CCT). Some of the positive points we discussed were:

  • The cooling provided by an L1S approach would be more even across the face of the globe than that readily offered by SAI or CCT, and far more even than that offered by MCB. This would reduce some of the secondary geophysical effects caused by spatially inhomogeneous cooling.
  • The radiative forcing from MCB and CCT is inherently local and short duration (lifetime of a day or less) which means that large-scale MCB or CCT geoengineering incudes the possibly of “weather control” and of sharply un-equal climate impacts. With its year-long timescale SAI does not enable weather control, but does allow substantial latitudinal variation in forcing and thus in climate. If weather control or the ability to cause sharply unequal climate impacts is a geopolitical threat, then the inherent evenness of radiative forcing of an L1S system is a geopolitical benefit.
  • Unlike terrestrial solar geoengineering, the L1S approach does not involve intentional insertion of material into the atmosphere such as aerosols that alter atmospheric chemistry and may damage human health. An L1S modification would, for example, have no direct effect on stratospheric chemistry—including ozone chemistry. It would work instead as a direct reduction of the solar constant or (in more complex designs) an alteration of the solar spectrum. While any alteration in solar flux will cause changes in stratospheric heating rates and thus in stratospheric dynamics, these changes would be smaller than those seen in SAI, and their feedback behavior more predictable. (It is worth noting, though, that an L1S dependent on components launched from Earth would, if it were to depend on chemical rockets, result in significant amounts of material deposition in the upper atmosphere: see concerns, below.)
  • Geoengineering with forms of L1S that block specific infra-red wavelengths in the incoming sunlight, rather than scattering all of it, could have significantly less of an effect on the hydrological cycle than a SAI deployment offering the same radiative forcing. This might allow a geoengineered climate quite closely comparable to the historical climate in terms of rainfall as well as temperature—something which the precipitation-reducing tendency of SAI’s effects on the hydrological cycle precludes. However, such approaches would likely require a significantly larger shield area, with a corresponding increase in cost and complexity.
  • As large space-based projects, L1S approaches would come clearly under established international law, specifically the Outer Space Treaty of 1967. “The exploration and use of outer space, including the moon and other celestial bodies, shall be carried out for the benefit and in the interests of all countries, irrespective of their degree of economic or scientific development, and shall be the province of all mankind.” There is no comparable overarching legal context for SAI, MCB or CCT.

Some areas of concern we discussed were:

  • The required number of launches to space per year could pollute the upper atmosphere with aerosols. This concern could be quantified with further calculations.
  • An L1S could be another level of distraction within the climate change and mitigation efforts dialog.
  • Any L1S proposal would most likely have a significant effect on the commercial sector; the $1trn would represent a great many contracts. This would be a departure from the current solar-geoengineering discussion where commercialization is frowned upon.
  • Such a large project would need to be undertaken either by the largest and richest countries, or as a global effort. This would have potential positive aspects (see above) but would also present major geopolitical challenges.
  • The vulnerability of a space-based solar geoengineering system to destruction by rogue actors is likely to be much higher than for atmospheric solar geoengineering or (obviously) standard climate-change-mitigation strategies like decarbonization and carbon capture. Redeployment of an L1S after destruction would be a major effort. The inherent back-up against termination of SAI provided by the fact that many actors could conceivably have the capacity to undertake SAI would be absent.

Additionally, a little more consideration of such possibilities might usefully be undertaken both in space-policy and space-advocacy discussion which should both consider all the possible uses of outer space. Some discussion of these possibilities in international fora, such as the UN Committee on the Peaceful uses of Outer Space, might also be beneficial, though it is not a priority. Greater consideration of space-based approaches might also improve solar-geoengineering discussion, which could benefit from the wider perspective that the addition of a new technological approach might bring.

These considerations mean that, although further study of L1S approaches should not be a priority, nor should they be ruled out as having no conceivable attraction. Some theoretical studies to refine the parameter space and map out the feasibility of a L1S option would significantly add to the field.



Roger Angel

Jay Apt

Anita Bernie

Billy Derbes

John Deutch

Esther Dyson

Martin Elvis

Jane Flegal

Robert Hoyt

David Keith

Frank Keutsch

Oliver Morton

Ted Parson

Sara Seager

Yomay Shyur

Neal Stephenson

Matt Weinzierl

Adam Wong

Pete Worden

Robin Wordsworth


Call-in Participants

Creon Levit

Douglas MacMartin

Colin McInnes

Paul Wooster


Sunday, 24 Nov

5:00 PM           Arrival and reception

6:20-6:40         Initial welcoming remarks – Goals for meeting

7:00                 Dinner

8:00-8:15         David Keith: Solar geoengineering prospects

8:15- 8:30        Oliver Morton: Space prospects

8:30-9:30         Discussion


Monday, 25 Nov

8:00                 Breakfast

8:45-9:00         Plan for the day + short Q&A about plan

9:00-10:45       Scene setting chalk talks

Roger Angel: L1 options

Yomay Shyur: Non-L1 options

David Keith: Spectrally selective shields

Martin Elvis: Scenarios for space economy

Paul Wooster: Evolution of launch costs

             Rob Hoyt: In situ manufacture

10:30-10:45     Break

11:00-12:00     Group Discussion—What is being overlooked? What is most plausible? What might research look like? Robin Wordsworth moderating

12:00-1:00       Lunch

1:00-1:45         Breakout group discussions—What might a 10-year research plan look like? What are the biggest non-technical concerns?

1:45-2:45         Breakout group reports with skype connection and comments by Doug MacMartin, Colin McInnes, Creon Levit

2:45-3:00         Break

3:00-3:30         What would this mean for space technology and policy?—Oliver Morton moderate a panel with Esther Dyson and Jay Apt

3:30-4:00         What would this mean for geoengineering and climate politics and policy? —Pete Worden moderate a panel with Ted Parson and Jane Flegal

4:00-5:00         Next steps and meeting close—David Keith moderating




M. Mautner and K. Parks, "Space-based control of the climate," Engineering, construction, and operation in space II, pp. 1159-1169, 1990.


R. Angel, "Feasibility of cooling the Earth with a cloud of small spacecraft near the inner Lagrange point (L1)," Proceedings of the National Academcy of Sciences, vol. 103, no. 46, pp. 17184-17189, 2006.


E. Teller, L. Wood and R. Hyde, "Global warming and ice ages: I. prospects for physics based modulation of global change," Lawrence Livermore National Laboratory, Livermore, CA, 1997.


"The Outer Space Treaty," 1967.



David Keith, a professor of applied physics at Harvard’s School of Engineering and Applied Sciences (SEAS) and a professor of public policy at Harvard’s Kennedy School of Government, is the faculty director of Harvard’s Solar Geoengineering Program. Oliver Morton is The Economist’s briefing editor. Yomay Shyur is a Postdoctoral Fellow at the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS). Pete Worden is the Chairman of the Breakthrough Prize Foundation and Executive Director of the foundation’s Breakthrough Initiatives. Robin Wordsworth is an associate professor at Harvard’s School of Engineering and Applied Sciences (SEAS) and an affiliate of the Earth and Planetary Sciences Department (EPS).