opinion & comment COMMENTARY: Solar geoengineering reduces atmospheric carbon burden David W. Keith, Gernot Wagner and Claire L. Zabel Solar geoengineering is no substitute for cutting emissions, but could nevertheless help reduce the atmospheric carbon burden. In the extreme, if solar geoengineering were used to hold radiative forcing constant under RCP8.5, the carbon burden may be reduced by ~100 GTC, equivalent to 12–26% of twenty-first-century emissions at a cost of under US$0.5 per tCO2. ailure to address the accumulation of between a Representative Concentration carbon cycle feedback under assumptions atmospheric carbon is among the most Pathway (RCP) 8.5 scenario and one in that are similar — though not equal — to frequently noted disadvantages of solar which solar geoengineering is used to hold those that would be used to simulate solar F 1–3 geoengineering , an attempt to reflect a radiative forcing at current levels. This is geoengineering to stabilize radiative forcing small fraction of radiation back into space not a complete analysis, but rather a call for under an RCP8.5 scenario. We then combine to cool the planet. The latest US National further research. It is also a call for assessing the two ranges using equal weights and Academy of Science solar geoengineering solar geoengineering scenarios that go well uncorrelated error propagation to yield report1 states it “does nothing to reduce the beyond oft-modelled extreme scenarios that an overall estimate of the contribution 10 build-up of atmospheric CO2”. offset total anthropogenic radiative forcing . of the terrestrial biosphere and ocean of This is not so. Solar geoengineering These rough estimates alone, however, provide 89–283 GtC (Table 1). reduces the carbon burden, and therefore suggestive evidence of the potentially large The direct impact of solar geoengineering ocean acidification, due to the three pathways impact of solar geoengineering on the carbon on the loss of carbon from permafrost soils explored here: carbon-cycle feedback4–8, burden and emissions. is unexplored. We instead include estimates reduced permafrost melting, and reduced Warming can increase the atmospheric from recent intercomparisons of dynamic energy-sector emissions. carbon burden by increasing ecosystem permafrost models that estimate a decrease in While it is appropriate to treat solar respiration, decreasing primary productivity, cumulative emissions under RCP8.5 to range geoengineering as distinct from carbon and decreasing oceanic carbon uptake. from 27 to 122 GtC (ref. 6). mitigation or geoengineering approaches These carbon cycle feedbacks amplify These rough estimates should be that tackle carbon directly9, the impact climate responses to anthropogenic interpreted with caution. Caveats include: of solar geoengineering on the carbon emissions. Point estimates differ widely neglecting the differences between the cycle calls for more integrated research. (see Supplementary Table 1). Special Report on Emissions Scenarios Solar geoengineering or solar radiation We derive an overall range in two steps. (SRES)-A2 (ref. 11) and RCP8.5 (refs 4–6) management (SRM) is, in this sense First we take estimates of 31 GtC (ref. 5) and scenarios; neglecting the fact that simulating alone, arguably a form of carbon dioxide 251 GtC (ref. 4) from the only two models carbon-cycle feedback by eliminating all removal (CDR). that directly simulate the carbon cycle climate change11 is, at best, a rough proxy response to RCP8.5 and solar geoengineering, for solar geoengineering; and ignoring more Carbon impacts of solar geoengineering and combine them with the full range of speculative carbon feedbacks such as sea-bed We calculate the total carbon burden in 2100 results from a study11 estimating the carbon methane hydrates7. Moreover, our subsequent and carbon emissions impacts during the response with and without CO2 impact on rough translation of carbon burden to twenty-first century by estimating, based climate: 42–420 GtC. The latter provides emissions, and vice versa, does not account on diverse prior literature, the difference systematic sampling of the uncertainty in the for changes in ocean buffering12. Table 1 | Reduction in twenty-first-century emissions and in 2100 atmospheric carbon burden in GtC. Source Reduction in twenty-first-century emissions (GtC) Reduction in 2100 burden (GtC) Carbon cycle (burden) 127–404 89–283 Permafrost (emissions) 27–122 19–85 Energy sector (emissions) 24–54 17–38 Total 232–527 162–369 Estimates show the range of the difference between RCP8.5 and a solar geoengineering scenario with constant twenty-first-century radiative forcing. We derive a primary estimate of either burden or emissions (shown in bold) and then convert it using a fixed airborne fraction of 0.7, based on the RCP8.5 CMIP5 multi-model mean25. See Supplementary Materials for details on the calculations. NATURE CLIMATE CHANGE | VOL 7 | SEPTEMBER 2017 | www.nature.com/natureclimatechange 617 ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. ©2017 Mac millan Publishers Li mited, part of Spri nger Nature. All ri ghts reserved. opinion & comment For a given energy demand and fuel SRM as CDR solar geoengineering will be used to offset mix, carbon emissions will rise with If used to offset changes in all warming, a moderate scenario with less temperature, as the efficiency of heat engines twenty-first-century radiative forcing under solar geoengineering is likely a better policy10. declines with rising ambient temperature. an RCP8.5 emissions scenario, our rough For example, using solar geoengineering to Warming will also decrease energy demand estimates suggest that solar geoengineering halve the rate of radiative forcing growth for heating and increase energy demand could reduce the carbon burden in 2100 by might better balance the risks and benefits4,10. for cooling. We use a global estimate of around 160–370 GtC, roughly equivalent Under such a scenario, everything else staying energy demand response to warming in the to reducing twenty-first-century emissions equal, the reduction in carbon burden would 13 residential sector , roughly scaled to cover by 850–1,900 GtCO2 at a mitigation cost of likely be roughly halved as well relative to the the commercial sector as well as transport, US$0.2–0.4 per tCO2. Rather than having no calculations here. Some carbon impacts also along with various estimates of the impact impact on carbon, solar geoengineering may depend on the type of solar geoengineering of climate changes on energy use, to yield a be among the most cost-effective methods of and specific materials used. Our rough cost rough estimate that avoiding the warming limiting the rise in CO2 concentrations and, calculation, in particular, assumes using in an RCP8.5 emissions scenario decreases therefore, the rise in ocean acidification. sulfate aerosols. Resulting stratospheric ozone cumulative emission by 24–54 GtC Even with these carbon benefits, solar depletion may lead to small increases in ocean (see Supplementary Materials). geoengineering cannot substitute for cutting acidification20. Using different compounds21,22 emissions. For one, our rough estimates, may have lower effects or even the opposite Cost-effectiveness using an extreme scenario, show a total effect. Marine cloud brightening similarly Risks, uncertainties, and inter-temporal emissions impact of ‘only’ around 12–26% of has direct implications on emissions, carbon trade-offs make simple cost-effectiveness total twenty-first-century emissions under burden23, and, thus, also ocean acidification24. estimates a poor measure of the RCP8.5 (ref. 17). The third reason is moral hazard: the need overall utility of solar geoengineering. Second, the two primary factors we to consider social and societal responses Narrow calculations of costs make solar identify here, carbon-cycle feedback and beyond the technical calculations. geoengineering, in particular using permafrost release, merely move carbon Sensible policy decisions about stratospheric aerosols, appear ‘too cheap’. within the biosphere. Only the smaller both emissions mitigation and solar Our analysis does not claim completeness. third factor, via the energy sector, prevents geoengineering will be aided by better There are clearly unquantified and perhaps moving carbon from the geosphere. Unlike estimates of the carbon-cycle benefits of solar unquantifiable risks of solar geoengineering. some forms of CDR, no mechanism here geoengineering and of the way the reduction Those may imply that the only relevant removes carbon from the biosphere and puts in carbon burden scales with the amount decision criterion for solar geoengineering it back into the geosphere. Thus, terminating of solar geoengineering and mitigation. deployment is one based on risk–risk solar geoengineering efforts would lead to a A coordinated research effort should aim tradeoffs, not one based on cost–benefit significant adverse carbon impact. to understand the coupling between solar analysis. Increased albedo is not anti-CO2. Third, none of this addresses an oft-cited, geoengineering, CDR, the energy system, But when considering solar geoengineering indirect link via societal responses, often and the ‘natural’ carbon cycle. Policymakers as a means of reducing carbon burdens, under the heading of ‘moral hazard’18. Solar cannot make sound choices without a cost-effectiveness is relevant because the geoengineering may have direct implications sustained, integrated research