An assessment of the infrastructural and temporal barriers constraining a near-term implementation of a global stratospheric aerosol injection program

Models of stratospheric aerosol injection deployment scenarios have often assumed that a global sunscreen could be applied to the Earth on relatively short notice, perhaps in response to a climate emergency. This emergency response framing confuses the timescales associated with the commencement of such a program. Once deployed, stratospheric aerosols could cool the Earth quite quickly, but the most commonly assumed deployment scenarios would require aircraft and other infrastructure that does not currently exist. Given the span required to develop and certify a novel aircraft program and to subsequently build a fleet numbering in the hundreds, scenario builders should assume a roughly two-decade interval between a funded launch decision and the attainment of a target level of cooling.


Introduction
Stratospheric aerosol injection (SAI) is a hypothesized climate management technique that would seek to increase the Earth's albedo, thereby reducing the quantum of energy that the climate system would absorb from the sun.SAI is a controversial (NASEM 2021, Biermann et al 2022) and untested (NASEM 2021) climate intervention that could result in undesirable impacts (Haywood et al 2013, Keith et al 2016) and is not proposed as a substitute for either emissions reductions or adaptation (National Research Council (U.S.) et al 2015, NASEM 2021).However, similar aerosol injections from large volcanic eruptions have long been known to substantially reduce surface temperatures even at points far removed from their origin, an effect that was directly measured and attributed after the eruption of Mount Pinatubo in 1991 (Kirchner et al 1999, Ramachandran et al 2000).There is also increasing confidence that SAI deployment would be both aeronautically feasible (Smith and Wagner 2018) and extraordinarily cheap (Smith 2020) relative to other prospective measures by which to combat climate change or its impacts.
However, to achieve the substantial global cooling targets often imagined for SAI, a large fleet of novel, highaltitude aircraft would be required.Studies of alternative lofting concepts such as balloons, rockets, guns, or tethered hoses conclude that the most efficient and reliable lofting technology would be fixed-wing, selfpropelled, air-breathing jets (McClellan et al 2012, Smith andWagner 2018).Few jets can achieve the required altitudes and those that do carry insufficient payloads (Smith andWagner 2018, Smith et al 2022a).Nor can existing jets can be cost-effectively modified to undertake this mission (Smith and Wagner 2018).Therefore, the design, testing, certification, and fleet manufacturing time spans associated with procuring such a deployment fleet must be accounted for in contemplating both the year in which a deployment program could commence and the year in which it might fulfil its temperature target However, such a developmental span is seldom explicitly taken into account when climate or social scientists consider the introduction of an SAI program (Tilmes et al 2018, Kravitz et al 2019).SAI is often portrayed as a 'break glass in case of emergency' measure that might be rushed into service in response to a climate catastrophe or a surpassed tipping point (Gates 2021), but this can only occur if the required infrastructure has been assembled beforehand.Large commercial aircraft programs generally have a decadal interval between funded launch and 'entry into service'.Governmentally procured programs generally exceed that span, and intergovernmental programs introduce further complications and delays.
Moreover, the developmental span results merely in the first prototype aircraft.If the goal is a substantial quantum of coolingsay to re-cross a tipping point or combat recurrent instances of extreme heatseveral hundred aircraft would be required (Smith 2020), implying a manufacturing process that could consume yet another decade.The upshot is that if would-be geoengineers expect there to be an effective tool behind the glass if and when they break it, they should assume a roughly two decade interval between program launch and full deployment.

SAI scenarios
An assessment of the pathway that may lead to deployment of SAI requires a bit more clarity as to the destination one envisions, since there is a wide spectrum of both objectives and methods that might be chosen.Deployment scenarios may differ in respect of the background warming that they confront, the degree of cooling they target, and the timeframe on which they might commence and ramp up (Brody et al 2024).They are most commonly global in terms of geographic scope, but could instead be regionally targeted.They may be intended to reduce global average surface temperatures by substantial amounts over a long period of time, or could be localized short-term experiments intended to better understand the process engineering and surface impacts of such a program.They could be undertaken by large states or coalitions of them at either an emergency or a deliberate pace.With this multiplicity of deployment variables, it is necessary to narrow our focus before a pathway to deployment can be examined.
For this purpose, we refer to the SAI deployment scenario presented in MacMartin et al 2022.That paper aims to 'advance understanding and development of decision-ready scenarios' for solar radiation management, such that scientists subsequently running Earth System Models to clarify the possible impacts of SAI can converge on scenarios that are useful to future policy makers.Such scenarios need to be 'relevant and plausible', noting that those characteristics are subjective and contestable (MacMartin et al 2022).
With these goals in mind, the paper articulates an SAI scenario with the following features: • A background emissions scenario of SSP2-45 • A start date of 2035 • A model horizon of 35 years, so 2035 to 2070 • A maximum temperature target of 1.5C above the pre-industrial threshold, although alternative scenarios consider even more aggressive targets of either 1.0 or 0.5C above pre-industrial • A 10 year transition period from the temperatures at the start of deployment to the target • SO2 as the aerosol injected • Injection latitudes of 30S/15S/15N/30N • Injection altitude of 21.5 km These are not the only plausible parameters by which an SAI program might be defined.Several recent papers have described deployments at high latitudes to restrain tipping elements or ice loss at the poles (Lee et al 2021, 2023, Smith et al 2022b, Xie et al 2022, Duffey et al 2023, Goddard et al 2023, Sutter et al 2023).Others have considered regional interventions at lower altitudes intended to manage heat waves rather than average annual temperatures (Bernstein et al 2013, Mulena et al 2019, Zhao et al 2019).Small-scale (Golja et al 2021) or largescale/short-duration tests (Dykema et al 2014) might be undertaken to clarify the scientific processes and engineering requirements of subsequent larger/longer deployments.
Prospective deployers could have socio-political rather than climatological goals that might be served by a sub-scale deployment (Keith and Smith 2024) or commercial objectives that would call for micro-scale deployments such as those currently being undertaken by the 'cooling credit' vendor Make Sunsets.Marine cloud brightening, cirrus cloud thinning, or space-based SRM would call for entirely different technologies on different scales and timeframes.In each of these alternative cases, the infrastructural and temporal barriers that might constrain near-term deployment could be different, such that the arguments noted below may not apply.However, we assess that the deployment scenario described in MacMartin et al 2022 represents a specific and very common understanding of what SAI deployment would entail, so we will hereafter focus on what would be required to achieve this scenario.

Platform developmental span
Solar geoengineering studies that illustrated such planetary-scale intervention scenarios often did not explicitly mention the developmental span required to prepare the industrial apparatus generally or the aeronautical infrastructure specifically (MacMartin et al 2017, Tilmes et al 2018).Those that did account for infrastructural preparation often assumed a start date of 2030 (Moss et al 2010), 2033 (Smith and Wagner 2018) or more recently, 2035 (Smith 2020, Zhang et al 2023).However, to be consistent with past developmental spans, such start-dates would require a funded launch decision by a financially capable actor such as a major state approximately this year, which is not known to be under consideration anywhere and therefore seems highly unlikely.A review of the developmental timescales of eight large air transport category aircraft programs initiated since 1990 suggests an average span of more than 11 years.This includes major programs by Boeing, Airbus, Comac, Bombardier, and Embraer.It should be noted that the end date for such developmental programs is not 'first flight', which is merely an early milestone in the flight testing program, but rather entryinto-service, which requires certification by the relevant governmental authority and initial deliveries of lineready aircraft to customers.
Although the certifying authorities and respective procedures are different, the large rocket programs developed by the private space flight industry provide another basis for comparison.SpaceX's two operative large platforms required seven and eight developmental years, respectively.For the SpaceX Starship and Blue Origin New Glenn rockets, 2024 marks the 8th and 12th years of their development programs-and counting.If one assumes that New Glenn enters service in 2024 and Starship the year after (both possible but not assured), that would put the average for these four rocket programs at over 9 years, and the average for the airliners-plusrockets at over 10 years.
One could reasonably argue that airliners and private sector rockets are distinct categories that should be considered separately, but for our current purpose, they share one important feature: the key funding and configuration decisions are made by private companies.This is distinct from military aircraft, where the key decisions are made by governmental officials and are subject to much wider scrutiny and review beyond that required for airworthiness certification.With respect to governmental procurement projects, there is a much longer process that must precede the final 'Authority to Proceed' decision.First, a service branch must articulate a statement of need, after which a political process must determine whether the government wishes to fund a responsive program.This decision is based in part on the cogency of the statement of need, but also on such bases as in whose political jurisdiction the aircraft will be built and by which contractor.Typically, several prime Original Equipment Manufacturers (OEMs) bid against each other, with the outcome subject to second guessing, compromise, or litigation.After a period of contract negotiation and public review, a funding decision may or may not be made.To be sure, large private defense contractors have their own onerous bureaucracies and review procedures, but adding another layer of political considerations atop these only impedes the pace of progress.This is all the more true if there is not merely a single government making funding decisions, but several allied national governments seeking relative advantage in intergovernmental procurement projects.All of the above conspires to cause the developmental span for government-procured platforms to be longer than those noted above, even though the same prime contractors are often involved.
The developmental span for the two large Western military transports introduced in the last 30 years averaged 23 years.For the three modern bombers, the average was 17.Eight modern fighter programs averaged over 22 years, assuming that the two currently in development (the Future Combat Air System and the Tempest) maintain their projected schedules.Large unmanned aerial vehicles seem to blossom much more quickly, but even the six programs noted here averaged nine years (two of thesethe X-45A and X-47-never actually entered service, but did complete extensive flight test programs).
Among the 34 large aircraft and rocket programs reviewed in tables 1 and 2, the developmental spans for commercial and government-procured platforms are roughly one and nearly two decades, respectively.For a high-altitude SAI deployment platform, the developmental budget alone would run in the billions of dollars (Smith 2020), leading to a platform for which there are no currently conceived alternative uses, in that a plane optimized to deliver chemicals to the lower stratosphere would be very inefficient operating at lower altitudes.Subsequent annual operating costs would likely run in the tens of billions (Smith 2020), which would quickly exhaust the fortunes of most potential philanthropists.While Make Sunsets has attempted to fund their solar geoengineering venture via the sale of voluntary cooling credits, the amount of sulfur they claim to be lofting would have no discernable impact on the climate, and their operation is less than one billionth the magnitude of the global deployment scenario considered here.If SAI were to be undertaken on a large scale, the most plausible funding source would be governments or perhaps a coalition of them, placing this platform on the slowest developmental track.
Yet another set of relevant data points are those related to the development of major new jet engine programs.A purpose-built jet for SAI deployment at high altitudes could conceivably be powered by preexisting engine designs, but many new aircraft designs require newly developed engines, and the very unusual requirements here could call for that.Developing a new engine would be done in parallel with the aircraft design and therefore need not necessarily add to the total span, but the 16 modern programs listed in table 3 averaged spans of 8.7 years, further reinforcing the roughly decadal expectation.
I am often challenged by colleagues as to why we could not go faster in this particular instance given how high the stakes may be when confronting a climate emergency.However, this timeline is mostly beyond the control of either the OEM or the ultimate customer, and is instead driven by the certification process of the surveilling authority such as the FAA or EASA.Those agencies have evolved a spectacular safety culture that has rendered travel by air much safer than travel by car, but they have done so by meticulously honing and following their processes and procedures.Every customer hopes to evade these speed limits, but for planning purposes, it would seem more reasonable to assume that an aircraft developed for SAI will follow the development timeframes typical of other large contemporary aircraft.

Fleet build span
Another misconception often embedded in the 'break glass' framing of solar geoengineering is that should a transgressed tipping point or other 'climate emergency' call for it, we might flip a binary switch and 'turn down the sun' (Meyer 2019).First, as noted above, unless the aircraft fleet was developed and manufactured in advance on a precautionary basis, 'flipping the switch' will merely initiate a ∼10-15-year development cycle rather than commence geoengineering.But even at the end of that, we would merely have a single certified test article.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.At the outset, the OEM will not shift from zero to a mature annual production rate in a single year.Thousands of suppliers stand behind a production line for large aircraft, and each of them must ramp into their activity level, build their tools, and train their workers.Hence, even an urgent program might take three to five years to ramp into a mature annual production rate, as is illustrated in table 4 by annual deliveries for three modern airliner programs.The OEM will then seek to maintain that production rate for a decade or more, and will press its customer to accept a steady number of annual deliveries rather than flex up and down each year per the dictates of a climate model.Thus, a fleet of 400 aircraft might take a few years to ramp into a mature annual production rate and then another ∼decade to churn out ship number 400.Only then would the fleet reach its intended size, such that the cooling program could achieve its intended temperature target.Differently put, even if 'time to launch' (the developmental time span to produce the first line-ready aircraft) could be compressed to the ten years typical for commercial aircraft (unlikely), the subsequent 'time to fleet target' might be a similar or longer span.Deployment could begin in Year 11 of that theoretical two-decade timeframe, but only in small magnitudes that would ramp gradually to the intended target.
To clarify as to potential fleet sizes, table 5 shows the number of aircraft that might be required to reach global average surface temperature reductions of 0.5, 1, and 1.5 °C using the payload and altitude assumptions incorporated in Bingaman 2020.It further shows the annual production rates required to produce these fleets in a 15 year span, assuming a three-year ramp to a mature production level.Efficacy is sourced from (Zhang et al 2023) as being 0.0847 degrees C per Tg-SO 2 /yr.
As a check on the reasonableness of the annual production rates in tables 5, table 6 presents the current production rates of various large aircraft programs.While the 737 and A320 demonstrate that faster annual production rates than those in table 5 are possible, it should be noted that ramping into these production rates took decades and that there are no modern peacetime precedents for cramming a fleet's worth of production into say five years and then shutting the line.That of course does not imply that it could not be done, but rather that doing so would be far more costly.Industrial logic will call for manufacturing the fleet over a 15-20 timeframe that ramps to and then maintains a steady annual production rate.
The effect of these more industrially realistic aircraft production assumptions on the deployment program presented in (Zhang et al 2023) is illustrated in figures 1 and 2. Zhang's program assumes that a world which had warmed by 1.5 °C above the preindustrial baseline in 2035 would be cooled over the subsequent 15 years to a 1 °C temperature anomaly.Once the average surface temperatures have returned to a 1 °C anomaly by 2050, SAI is used to maintain that temperature through 2070 (the end of the model horizon).However, since this cooling is  being implemented against a backdrop of still-rising global greenhouse gas concentrations, such a program requires approximately 1 °C of cooling by 2050 and more than 1.5 °C by 2070 relative to the temperatures that would have obtained without SAI.Such a program calls for a surge in annually deployed aerosol masses and therefore in aircraft deliveries in the early net-negative years, followed by a more moderate delivery rate in the subsequent temperature stabilization years.However, the aircraft are assumed here to have a useful economic life of 25 years, such that deliveries in year 26 consist of the new aircraft required for further fleet growth in that year plus replacement aircraft for the Year 1 deliveries that retire in year 26.This causes production to surge again in year 26 and thereafter.This aircraft delivery profile is illustrated in figure 1 as the 'Requested Deliveries' and has a distinct 'barbell' shape, with high early-and late-year production rates bracketing a fallow middle period.The OEM will abhor this profile, as it would entail overbuilding tooling and facilities in the early years, laying off most of the workforce after the initial surge, and rehiring/retraining them a decade later.Therefore, it would likely counter-propose a 'Level Deliveries' plan optimized from its standpoint, showing a three-year build to a mature production rate and then level annual production for the succeeding 32 years.This would lead to the same fleet count in 2070, but with substantial under-deliveries in the early years of production relative to the Requested Deliveries plan and far more in the fallow period.After negotiation, something like the 'Dampened Barbell Deliveries' production plan might arise, entailing a compromise between the two-not as erratic as the Requested plan, but more so than the Level plan.
One could object that the revised SAI program presented above shows the tail wagging the dog and that it must surely be the SAI program that will inform the aircraft delivery rate and not the reverse, but the truth is likely somewhere in between.The fate of the global climate must indeed be the primary consideration, but anything that is done will be mediated through the realities of the aeronautical supply chain, and that will introduce constraints as to the pace of what should be expected.

Conclusion
The temporal constraints I note here on the pace at which an SAI program might unfold routinely lead to the 'Manhattan Project/Moonshot' conjecture, wherein these two historical analogies are unearthed to demonstrate how quickly we might do something if we simply put our minds to it.However, the more recent of these commenced more than 60 years ago and took a decade to achieve.The former was forged in the most searing crucible of the last century.There is no modern aeronautical precedent for throwing out the rules and concentrating all our resources in the manner implied by these analogies.That does not prove that a faster developmental span would be impossible, but it does strongly suggest that it would be an incautious way to plan.The default assumption should be that the aeronautical infrastructure development required for SAI will be implemented by one or more of the world's major OEMs using the processes they have honed over decades, and will be surveilled by one or more of the world's major certifying authorities, using the rules and regulations by which they operate.Failing miracles, that is the way this industry works.
However, as noted at the outset, all of these constraints apply specifically to the most common SAI deployment scenario as illustrated in (MacMartin et al 2022) and are not necessarily binding on other notions of SAI deployment or other modes of solar radiation management.In particular, deployments of more limited spatial extent, temporal scope, or cooling ambition may be able to find expedited sub-scale or sub-optimal paths to deployment.However, these would not fulfill the same climate ambition as is generally imagined for SAI.
Should we someday wish to respond to a climate emergency by breaking the glass and employing SAI, the bad news is that we would need the foresight to start developing the tool behind the glass well in advance.On the other hand, the good news is that the same speed limits that constrain would-be climate rescuers also constrain potential rogues, so humanity need not worry that unilateral globally effective SAI will be sprung upon it in the near future.
It may also be the case that developing a sufficient scientific knowledge base to facilitate informed decisions about SAI as well as a governance structure that could confer legitimacy upon it could both prove to be 'longer poles in the tent' than the span required to build the aircraft.Nonetheless, the fleet development interval defines a speed limit that would be hard to exceed, and that speed limit suggests that planetary scale SAI is at least two decades away from being realizable.

Table 1 .
Developmental time spans-Commercial aircraft a .
a Program start refers to the date on which the Original Equipment Manufacturer (OEM) publicly announced the airplane program alongside the announcement of a purchase order from a major airline or airplane buyer (such as the U.S. Department of Defense).Entry into service refers to the date on which the airplane began performing its role in a non-testing context.For military aircraft, this implies that the aircraft was integrated into the existing fleet and began being flown by pilots on missions and training.For passenger and cargo aircraft, entry into service implies the aircraft performed its first flight with ticketed passengers or a commercially designated load of cargo.

Table 3 .
Engine program developmental time spans.

Table 4 .
Aircraft production ramp rates.

Table 5 .
Fleet requirements for cooling scenarios.