Atmospheric Management

This brief is about deliberately managing what the air is made of, other than carbon dioxide. That subject has exactly two entries on the historical ledger and they sit at opposite ends of it. One is finished and worked: the Montreal Protocol removed a class of manufactured gases from the atmosphere and the stratosphere is measurably responding. The other is proposed and unvalidated: removing methane, which is not manufactured by anyone in particular, is already dispersed through the whole troposphere, and is destroyed by natural chemistry at a rate nobody can pin down. Carbon Capture at Scale owns CO₂; Climate Engineering and Planetary Cooling Concepts own the energy budget; Geoengineering Governance owns the instruments. What is left here is composition, and composition is a chemistry problem wearing an engineering problem's clothes.

One number does most of the work. Atmospheric methane sits near 1,860 parts per billion against a preindustrial 750 ppb. Direct air capture of CO₂ is already expensive at 420 parts per million; methane is the same class of separation problem at roughly one two-hundredth of the concentration. That factor is not a cost multiplier, it is a regime change, and it is why the credible end of the field has abandoned capturing anything. What is left is to enhance the atmosphere's own destruction chemistry — the hydroxyl radical, or chlorine — and let the oxidation happen in place at zero air-handling cost.

Which lands the brief on its central difficulty. In 2022 a paper in Nature attributed roughly half of the record 2020 methane growth to a weakening of the hydroxyl sink. In 2024 a paper in PNAS, reading carbon isotopes over the same surge, explicitly rejected that explanation. Neither has been withdrawn; no reconciliation has been published. The field proposes to move deliberately, and detectably, a quantity it cannot agree moved by itself. Everything else here — the roadmaps, the reactor chemistry, the governance question — is downstream of that unresolved fact.

Frontier Start with the disagreement, because it is the brief. Peng, Lin, Thompson, Xi, Liu and colleagues, in Nature 612:477–482 (2022), attribute the 2020 methane growth to a reduced atmospheric sink, 53 ± 10%, with higher natural emissions supplying 47 ± 16%, mostly from wetlands, on a global net emission increase of 6.9 ± 2.1 Tg CH₄ per year; the companion conference record puts 2020 growth at 15.1 ± 0.4 ppb, the highest annual increase from 1984 to 2020. Frontier Michel, Lan, Miller, Tans and colleagues, in PNAS 121 (2024), report the growth rate reaching a record 15.4 ppb per year across 2020–2022 and, from the shift in methane's carbon isotopes, conclude the driver was microbial — explicitly rejecting hydroxyl-sink reduction, along with fossil-fuel emissions, as the dominant cause. Frontier Feng, Palmer, Parker, Lunt and Bösch, in Atmospheric Chemistry and Physics 23:4863–4880 (2023), inverting GOSAT satellite retrievals, require emission increases of 27.0 ± 11.3 Tg in 2020 and 20.8 ± 11.4 Tg in 2021, largest in Eastern Africa at 13 Tg. Established Three independent methods, two conclusions, no published reconciliation — and this brief declares the tie rather than averaging it. Half of a record atmospheric excursion is either a real natural sink collapse or a modelling artefact, and the field cannot say which.

Established The hydroxyl radical is the quantity everything in this brief depends on, and it is the worst-constrained number in it. OH destroys roughly nine-tenths of atmospheric methane; methane's lifetime is not a property of methane but an output of a model of OH. Established He, Naik and Horowitz, in Geophysical Research Letters 48 (2021), report 0.24 years of difference in estimated methane lifetime arising purely from which meteorological reanalysis is used to drive the model — same chemistry, same emissions, different weather fields. Frontier The multi-model record does not settle it either: AerChemMIP, reported by Stevenson and thirteen co-authors in ACP 20:12905–12920 (2020), finds an approximately 9% increase in tropospheric OH over 1980–2014, while ACCMIP, reported by Naik and colleagues in ACP 13:5277 (2013), found global mean OH roughly constant across 150 years because large regional changes offset one another. Established A 0.24-year structural spread is the measurement floor, and the community's best historical proxy for OH, methyl chloroform, has decayed out of the atmosphere as a direct consequence of the Montreal Protocol succeeding. Frontier The most successful atmospheric intervention in history destroyed the instrument used to monitor the second one.

Established The concentration problem is not a difficulty within methane removal; it is the shape of the whole field. Jackson, Solomon, Canadell, Cargnello and Field, in Nature Sustainability 2:436–438 (2019), set the target: restoring preindustrial methane means removing roughly 1,110 ppb from the entire atmosphere. Established Separating a gas from air costs at minimum the entropy of mixing, and both that minimum work and the volume of air handled scale adversely with dilution. Frontier So any scheme passing ambient air over a sorbent, through a membrane or into a reactor is dominated by the cost of moving air, not by chemistry — which is why the serious literature converged on enhancing atmospheric oxidation in place rather than capturing anything. Established Jackson, Abernethy, Canadell and colleagues, in Phil. Trans. R. Soc. A 379:20200454 (2021), state the field's position plainly: carbon dioxide removal has an established research framework and “no similar framework exists for methane removal”. They proposed a Methane Removal Model Intercomparison Project as the remedy. Established It has not reported. Established The US National Academies published A Research Agenda Toward Atmospheric Methane Removal in 2024; this brief states that the report exists and states nothing from it, because it was unreachable to the research pass behind this page.

Frontier Iron salt aerosol is the most developed removal route, and its own roadmap describes it as a hypothesis. The mechanism: iron(III) chloride on aerosol surfaces photolyses to release atomic chlorine, which attacks methane roughly an order of magnitude faster per molecule than OH does but is present at far lower abundance. The proposal is to enlarge a natural sink, not to build a machine. Established The laboratory basis is real. Wittmer, Bleicher, Ofner and Zetzsch, in Environmental Chemistry 12:461–475 (2015), measured in a simulation chamber that the initial rate of atomic chlorine production per unit aerosol surface rises with Fe(III), modulated by NO₂, O₃ and SO₂. Speculative The origin paper is far larger in ambition: Oeste, de Richter, Ming and Caillol, in Earth System Dynamics 8:1–54 (2017), propose that doubling current natural iron emissions into the troposphere “would enable the prevention or even reversal of global warming” — fifty-four pages making a total claim, and the field's founding document rather than a result. Frontier Gorham, Abernethy, Jones, Hess, Mahowald and colleagues, in ACP 24:5659–5670 (2024), publish instead a five-year roadmap to explore whether the approach is viable and socially acceptable at all, stating it must complement rather than replace emission reductions. Frontier And the sign may not be fixed. Pennacchio, van Herpen, Meidan, Saiz-Lopez and Johnson, in a 2024 ChemRxiv preprint, report that methane removal per iron atom is always net positive at NO_x mixing ratios below 50 ppt or above 390 ppt. The corollary is the most consequential unstated sentence in this literature: between roughly 50 and 390 ppt, it may not be. A technology whose sign depends on ambient NO_x behaves differently over open ocean, over farmland, and downwind of a city.

Established The catalytic-oxidation literature is excellent and it is solving a different problem. Thermo-catalytic oxidation on CuO_x/zeolite, fluidised catalytic oxidation at pilot scale, low-temperature ozone-assisted oxidation over cobalt zeolites: all published since 2024, all using the words “low concentration” and “ultralow concentration”. Established In that literature those words mean coal-mine ventilation air at roughly 0.1–1% methane by volume — between 500 and 5,000 times atmospheric. Established None of it is a demonstration at 1.9 ppm, and the air-handling energy penalty at ambient concentration is why that is a difference in kind rather than in degree.

Established Now the case that worked, with the numbers. The Montreal Protocol is the only completed deliberate change to global atmospheric composition, and its result was detected late and statistically rather than obviously. Kuttippurath and Nair, in Scientific Reports 7 (2017), report “robust evidence that Antarctic ozone has started to recover” — a publishable claim thirty years after the treaty was signed, for the best-monitored trace-gas problem in the atmosphere. Established The completion date has also been receding. The CCMVal-2 multi-model assessment reported by Eyring and colleagues in ACP 10:9451–9472 (2010) projected Arctic total column ozone returning to 1980 levels around 2025–2030 and Antarctic around 2045–2060. Frontier Later return dates now circulate widely, sourced to the WMO/UNEP 2022 assessment; that assessment was unreachable to this brief's research pass and its figures are therefore not printed here. What is safe, and is itself the finding, is the direction: the projected completion of the one successful atmospheric intervention has moved later between successive assessments, and an honest account of “the treaty that worked” has to show that.

Established The stress test is CFC-11, and it is the part nobody invokes. Montzka and colleagues, in Nature 557 (2018), found that the atmospheric decline of CFC-11 had been constant from 2002 to 2012 and then slowed by about 50% after 2012, implying emissions higher by 13 ± 5 gigagrams per year, a 25 ± 13% increase, with the interhemispheric concentration difference widening by roughly 50% — the evidence placing the source in the Northern Hemisphere. Reported production had been near zero since 2006, so this was unreported new production against a phase-out completed in 2010. Established Rigby, Park, Saito, Western, Redington, Fang and colleagues, in Nature 569:546–550 (2019), localised it: eastern mainland China 7.0 ± 3.0 Gg per year higher in 2014–2017 than in 2008–2012, concentrated in Shandong and Hebei, at least 40–60% of the global rise, with no significant increase elsewhere in monitored eastern Asia. Established Montzka and colleagues, in Nature 590:428–432 (2021), recorded the resolution: emissions fell 18 ± 6 Gg per year, 26 ± 9%, from 2018 to 2019, to a 2019 value of 52 ± 10 Gg per year. Established And the caveat that is the actual lesson: the CFC-11 bank — produced but not yet emitted — had grown by 90 to 725 gigagrams by the beginning of 2020. The illicit production stopped; the material it made sits in foam and equipment and will emit for decades, and the estimate of how much spans a factor of eight.

Established The machinery that caught it was smaller than the story suggests. The localisation in Rigby et al. was possible because AGAGE-affiliated stations at Gosan in South Korea and Hateruma in Japan happen to sit downwind of the source region. Morton, in Eos 100 (2019), summarised the episode's lesson in one line: “better atmospheric monitoring networks are needed to enforce the Montreal Protocol”. Established The compliance mechanism of the world's most successful environmental treaty is, in practice, a modest number of atmospheric measurement stations run by research groups on competitive grants, with large parts of the world uncovered. Frontier And the attribution may not be finished: Niu, Kong, Yan and colleagues, in Earth System Science Data 17:6731–6746 (2025), report substantial evidence of CFC-11 emission from coal combustion in China, a source class that had been overlooked entirely. Single study, 2025, essentially no policy uptake. Frontier If a meaningful share of the anomaly was combustion rather than production, the enforcement narrative changes and the monitoring problem gets harder, not easier: an unintentional distributed source is far worse to police than a factory.

Established The record that verifies all of this has a documented step change at more than a third of its stations. Stauffer, Thompson, Kollonige, Witte, Tarasick, Davies, Vömel, Morris, Van Malderen, Johnson, Querel, Selkirk, Stübi and Smit, in GRL 47 (2020), report that between 2014 and 2016, ozonesonde total column ozone at 14 of 37 stations exhibited a sudden dropoff, traced not to the atmosphere but to electrochemical concentration cell instrument artefacts — the sonde hardware and its solutions. Established The network's own quality-assurance literature is candid about it. Tarasick and colleagues, in Earth and Space Science 8 (2021), report precision improved by a factor of 2 over 20 years through strict standard operating procedures rather than new hardware; Stauffer and colleagues, in the same journal in 2022, find agreement with Aura OMI stable to about ±2% across 60 stations over 18 years; Thompson and colleagues, in BAMS 100:155–171 (2019), report SHADOZ stations following WMO protocols within 3% of a reference instrument, and in 2026 that sondes remain irreplaceable at 100–150 m vertical resolution. Established Stated honestly: stability about ±2%, reference agreement about 3%, trends of comparable magnitude, and a hardware change producing a step at 38% of stations within living memory. Frontier This brief's research pass could not establish anything about the network's funding, and so asserts nothing about budget lines, station closures or programme terminations. The technical fragility is documented; the fiscal question is open and is named as open.

Frontier The open question that dominates the others is whether the hydroxyl sink moved in 2020, and it is open in the strong sense: two methods, two answers, no reconciliation. It is not a matter of error bars overlapping. Peng et al. assign 53 ± 10% of the surge to the sink; Michel et al. read the isotopes and reject the sink explanation outright. Resolving it requires either an independent OH constraint the field does not currently have, or a demonstration that one of the two inversions is structurally biased.

Frontier The NO_x-regime sign dependence of iron salt aerosol is the second live item and it is a preprint from a single group. Pennacchio et al.'s statement that removal per iron atom is net positive below 50 ppt and above 390 ppt NO_x has not been independently reproduced. Speculative If it holds, deployment maps become NO_x maps, and the same intervention has different signs in different places in the same week — a governance object with no precedent. Frontier Horowitz and colleagues' 2024 EGUsphere preprint on the intended and unintended consequences of atmospheric oxidation enhancement is the first systematic treatment of the side-effect question, and it too is a preprint with full text unreachable to this pass. Frontier The obvious things to watch are exactly the things the ozone regime spent forty years learning to model: tropospheric ozone production from enhanced radical chemistry, altered lifetimes for other OH-controlled species including carbon monoxide and the HFC/HCFC replacements, and any chlorine reaching the tropopause. The sharpest internal tension in this brief is that the proposed methane fix runs on the chemistry the ozone treaty exists to police.

Established The absence of a methane removal intercomparison is itself a frontier finding. MR-MIP was proposed in 2021 and no published intercomparison was located. Frontier In a neighbouring field, inter-model disagreement is the headline result — Climate Engineering carries GeoMIP's regional sign disagreements as its central finding. Speculative Without the methane equivalent, nobody knows how much of the removal literature's optimism is structural to the models producing it, and that is not a question a single group can answer.

Frontier Coal combustion as an overlooked CFC-11 source is a genuine reinterpretation of the most famous enforcement episode in environmental treaty history, and it has had no uptake. One 2025 paper, unreplicated. It deserves attention precisely because the policy literature has treated the CFC-11 story as closed since 2021. Frontier Similarly live, and unglamorous: “is our instrument stable?” is a continuing publication stream into 2026 for the ozonesonde network, which tells you both that the community is honest and that the answer is not yet a flat yes.

Frontier Pilot-scale catalytic oxidation of ventilation-air methane is advancing quickly and the frontier is real — fluidised-bed pilot results were published in 2025. Speculative Whether that constitutes the bottom rung of a concentration ladder toward ambient removal, as direct air capture descended from flue-gas scrubbing, is contested: the steelman is that a technology proven at 5,000 times ambient is a plausible ancestor; the counter is that air-handling energy scales inversely with concentration, so the ladder has a cliff at the bottom rather than a rung.

Frontier One scheduled event will move this subject: the 2026 Scientific Assessment of Ozone Depletion. It is in preparation and, for the first time, will treat stratospheric aerosol injection prominently. Speculative That makes the ozone regime the body being asked to evaluate a technology from the neighbouring brief, using the observing system described above — ±2% stable, artefact-prone, and built to measure something else.

Speculative And one route is nearly unexplored. Boreal plants were shown to take up methane in 2012, and soil methanotrophs are the second-largest methane sink after OH. Handwave A land-management route to enhancing the biological sink would be slow, distributed, and almost certainly unattributable — but it carries none of the sign risks of atmospheric chemistry, and it is essentially absent from the removal literature. That absence looks more like a disciplinary boundary than a scientific judgement.

Established The binding constraint is the hydroxyl field, and it binds three ways at once. OH is the mechanism the leading interventions propose to use; it is the quantity that determines whether they work; and it is the noise floor against which their effect would have to be detected. Established You cannot verify a 20 ppb intervention against a sink whose lifetime estimate moves by 0.24 years depending on which weather reanalysis you feed the model. No amount of funding, political will or engineering shortcuts this link, which is what distinguishes it from every other item below.

Frontier The second bottleneck is a laboratory one and it is cheap. Chlorine yield per iron atom has been characterised across part of the relevant parameter space; the NO_x, humidity, ozone and SO₂ envelope where the net methane sink might change sign has not been mapped. Established This is chamber chemistry. It needs no permit, no outdoor release and no community consent. That the field's leading candidate technology has an unmapped sign boundary after a decade of advocacy is a funding failure, not a scientific obstacle.

Frontier Third, there is no closed side-effect budget for oxidation enhancement — tropospheric ozone production, HFC and HCFC lifetime shifts, secondary aerosol formation, and any stratospheric chlorine pathway. One preprint exists. Established Fourth, and institutional rather than scientific: there is no multi-model intercomparison, five years after the field's own agenda paper called for one, because no body owns the coordination.

Established Fifth, the measurement side has no specification. Nobody has written down what network density, which species, at what precision, would be required to detect a deliberate global methane perturbation and attribute it to the intervention rather than to wetlands or to the weather. Frontier The equivalent artefact is missing for solar geoengineering too, as Climate Engineering and Geoengineering Governance both record — but methane is the easier case and it has not been done either. Established Industrial capacity is not a bottleneck. Iron is cheap, chloride is cheap; the unpriced item is dispersal at a scale set by a natural cycle nobody has costed replicating. Frontier And there is a bottleneck on the ozone side that is easy to miss because it looks solved: the verification record for the one completed intervention is held together by recurring intercomparison campaigns rather than by hardware, which makes its reliability a function of operating budgets rather than of technology — and the funding position of that network could not be established by the research behind this page.

Established The chain starts with a replacement for methyl chloroform. The global OH field was historically constrained by measuring the decay of a manufactured tracer with a known source; the Montreal Protocol removed the tracer. A new proxy species, or an isotopic or satellite constraint of comparable power, is the enabling scientific result for everything else in this brief, and no programme is producing it.

Established The second dependency is reconciliation between inversion methods. Isotope-based, satellite-inversion and sink-inversion approaches gave incompatible answers about a single well-observed event. Frontier Whether that reflects a real natural excursion or structural bias in one class of method is answerable with existing data and existing groups; it requires someone to run the comparison as a designed exercise rather than as three separate papers.

Frontier Third, chamber kinetics. The iron-chlorine work depends on simulation-chamber facilities and on the analytical chemistry to measure atomic chlorine production rates at realistic aerosol loadings — capability that exists in a small number of laboratories. Established Fourth, the ozone side depends on sustained instrument intercomparison campaigns, which are the mechanism by which the sonde network's precision doubled over twenty years, and which are the first thing cut when a network's budget tightens. Speculative Fifth and least discussed: a detection theory for planetary-scale interventions. Both this brief and its neighbours need a formal answer to what constitutes attribution of an atmospheric change to a deliberate act, and no discipline currently owns that question. Established The statistical machinery has a long ancestry — trend detection from the ozonesonde network was already a published methodological problem in 1986 — but it was built to find a slow secular change in an undisturbed system, not to separate a deliberate perturbation from a natural one. Frontier Sixth, and a dependency by exclusion: nitrous oxide has no route at all, being effectively inert in the troposphere with its sink in the stratosphere, so there is nothing to enhance. The three non-CO₂ gases do not share a technology stack; they share only a name.

Established The target has to be stated precisely or the workback plan is meaningless. Not “a catalyst that oxidises methane”. The target is: a demonstrated, sign-verified, side-effect-bounded reduction in global mean methane concentration, attributable to a deliberate intervention, at a scale of at least tens of parts per billion, detected against natural sink variability. Nothing resembling this has been attempted.

Frontier Experiment one, and the cheapest high-value move available today: chamber chlorine yields across the full ambient envelope. Extend the Fe(III)/sea-salt chamber work to the NO_x, humidity, ozone and SO₂ parameter space implied by the sign-dependence preprint, and establish where net methane removal changes sign. Laboratory scale, no permit, no release, no consent question. Established It would settle whether the field's leading candidate has a stable sign, which is currently unknown.

Established Experiment two: a designed OH reconciliation. Run the isotope, satellite-inversion and sink-inversion methods over the 2020–2022 surge with common inputs and published per-method diagnostics rather than ensemble means. Frontier Experiment three: a completed side-effect budget, taking the oxidation-enhancement modelling from preprint to closed accounting across tropospheric ozone, halocarbon lifetimes, secondary aerosol and any stratospheric chlorine pathway. Established Experiment four: publish MR-MIP, with a common prescribed removal scenario and results reported per model, not averaged — averaging is how model-structural optimism survives.

Speculative Experiment five is the one that collides with governance: a small, instrumented outdoor iron-salt release, measuring atomic chlorine production, methane decay and ozone response within the plume — the methane analogue of the stratospheric experiment that was cancelled on consent grounds rather than scientific ones. Frontier Over ocean it may fall under marine-dumping machinery; over land it is an air-quality event. Neither has been tested. Established Experiment six is a document, not a field campaign: an observing-system specification stating what network density, which species and what precision would be needed to detect and attribute a 20 ppb global intervention. It costs a working group and it does not exist. Frontier Experiment seven is the one nobody has proposed: a deliberate search for a methyl chloroform replacement. The tracer that constrained the global hydroxyl field has decayed out of the atmosphere, and no successor proxy has been adopted; identifying one, or demonstrating that an isotopic or satellite constraint can substitute for it, is the measurement on which every other item in this list ultimately rests.

Established The engineering question in this subject is dispersal, not synthesis. Iron chloride is a commodity; sea salt is free; the reagents in the leading proposal are among the cheapest materials on Earth. Speculative What is unpriced is putting them into the marine boundary layer at a flux comparable to doubling the natural atmospheric iron cycle, sustained indefinitely, with a spatial pattern chosen to keep the intervention in a favourable chemical regime. No costed platform architecture for this exists in the peer-reviewed literature.

Established The engineering that has been done is at the wrong concentration, and it is good engineering. Fluidised catalytic beds, copper-oxide-on-zeolite systems and low-temperature ozone-assisted oxidation are working at pilot scale on 0.1–1% methane streams. Frontier Scaling any of them toward ambient runs into the fan: at 1.9 ppm the reactor is a rounding error and the air mover is the whole machine. Handwave A device that could economically process the required air volumes at ambient concentration would be a genuinely new class of machine, and no proposal on the table describes one.

Established On the measurement side the engineering requirement is unglamorous and specific: manufacturing consistency in a balloon-borne sensor. The step change at 14 of 37 ozonesonde stations was traced to the sonde and its solutions, not the atmosphere. Established A treaty's verification criterion depends on the batch-to-batch reproducibility of a consumable instrument, and the fix that has worked has been procedural standardisation rather than better hardware. Frontier The one tropospheric intervention already being operationalised by an industry — rerouting aircraft to avoid forming persistent contrails — has an engineering constraint of the same character: forecast skill for ice-supersaturated regions, not hardware. It requires nothing to be added to the atmosphere, which is why it is the most tractable item in this subject. Frontier Its efficacy is contested and its constraint is a forecasting problem rather than a manufacturing one — which makes it the only intervention here whose binding requirement is computational.

Established Direct air capture of CO₂ is the closest technical neighbour and the most misleading one. It shares the dilution problem and shares none of the chemistry: CO₂ has no atmospheric sink to enhance, methane has one that already destroys nine-tenths of it for free. Carbon Capture at Scale owns that comparison from the other side.

Established Mine-ventilation and landfill methane abatement is adjacent industrial technology with real value, mis-shelved. Every catalytic result cited in methane-removal contexts comes from it: thermo-catalytic oxidation over copper-oxide zeolites, fluidised-bed pilots, ozone-assisted low-temperature routes, and the iron-cage zeolite chemistry described in Science in 2021 that made the field visible. Established It is a genuine mitigation industry aimed at 0.1–1% methane streams, and its results migrate into removal advocacy with the concentration quietly dropped. Frontier Ocean iron fertilisation is adjacent in a way its advocates rarely acknowledge: an aerosol programme that doubles atmospheric iron deposits iron to the surface ocean, which is fertilisation by another route and sits under an existing marine-dumping regime. Speculative An iron-salt deployment could become a regulated marine geoengineering activity by accident.

Established The measurement neighbours matter more than the technology neighbours here. The AGAGE and NOAA flask and in-situ networks, isotope-ratio mass spectrometry, satellite column retrievals of methane, and the ozonesonde and Dobson/Brewer systems are the shared instrument base for this entire subject and for the verification of anything the neighbouring briefs propose. Frontier Atmospheric chemistry transport models are the other shared dependency, and the 0.24-year reanalysis spread is a property of that shared dependency rather than of any one group's work. Established Adjacent by argument rather than by hardware: Weather Modification is the same attribution problem one order of magnitude easier and still unsolved after eighty years, and Geoengineering Governance owns the instruments that would have to classify any of this.

Established The most striking institutional fact about atmospheric methane removal is that it has no institution. The field's own agenda paper says no research framework exists; it proposed an intercomparison project in 2021 and none has been published, because a multi-model intercomparison needs a coordinating body, a data protocol and a funding line, and methane removal has none of the three. Established The US National Academies produced a committee report in 2024, which marks the crossing from advocacy into formal science-policy assessment. This brief does not state what that report recommends, because the report was not obtained.

Established The ozone regime is the counter-example and the model, and its enforcement layer is thinner than its reputation. The CFC-11 detection worked because a research-funded observing network with fortunate geography existed, and the episode's own summary literature says plainly that better networks are needed for enforcement. Frontier A treaty whose compliance mechanism sits on competitive research grants is structurally exposed in a way its record does not reveal, and the technical evidence supports the concern independently: precision gains came from sustained intercomparison campaigns, which are exactly the recurring-cost activity that funding pressure removes first. Established This brief could not establish the network's funding position and does not assert one.

Frontier Third, the classification problem, which is concrete here rather than theoretical. A large iron-salt aerosol deployment is simultaneously an air-pollution event, a marine-dumping question, a climate intervention and an ocean-fertilisation activity. Speculative It would plausibly engage three regulatory framings at once and no instrument coordinates them; Geoengineering Governance names that arbitrage in general and this is the sharpest worked example available. Established Fourth: the assessment cycle itself is an institution and it is about to be stretched. The 2026 ozone assessment will evaluate stratospheric aerosol injection for the first time, using a body constituted to answer a different question, with an observing system built for a different measurement. Frontier Fifth, an institutional fact with no home: the Montreal regime's own success removed the tool used to monitor methane. Methyl chloroform was the tracer that constrained the global hydroxyl field; the treaty phased it out and it has decayed away. No body owns the replacement problem, because it falls between an ozone treaty that considers its work done on that substance and a methane community with no framework at all.

Established The moral-hazard argument is pre-empted here by the proponents themselves, unprompted, in their abstracts. The iron-salt roadmap states removal must complement rather than replace emission reductions; the field's agenda paper frames removal as an addition to a mitigation programme. Frontier That is a notable contrast with the solar-geoengineering literature, where moral hazard is contested empirically and the experimental results are largely null. It should be reported as a fact about this field's conduct rather than assumed away.

Speculative The distributional problem in this subject is unusual: the intervention's sign may vary by place. If removal per iron atom is net positive only outside a middle band of NO_x, then a deployment beneficial over the open ocean could be neutral or harmful downwind of industrial and agricultural regions — and the populations in those regions did not choose it. Frontier Chlorine chemistry also produces tropospheric ozone, which is a crop toxin and a respiratory hazard as well as a greenhouse gas, so the plausible local harm is a public-health harm, not an abstract climatic one.

Established Consent is the binding constraint on the one experiment that would settle the science. Two outdoor atmospheric-intervention experiments in the last five years were stopped for reasons of notification and community consent rather than for scientific reasons, as Climate Engineering records in detail. Frontier An instrumented iron-salt release would face the same question, and the field has no worked-out answer for who must agree when the plume is over international water. Established There is also an ethics of measurement here that gets no attention: the verification of any composition regime depends on instruments whose known artefacts are of the same magnitude as the trends they measure, and publishing a compliance judgement on that basis is a claim about people, not only about air.

Established The durable lesson from the one completed intervention is about stocks and flows, and it generalises to everything on this map. The Montreal Protocol controlled production. When illicit production stopped, the bank of CFC-11 already made and installed had grown by somewhere between 90 and 725 gigagrams, and it will emit for decades regardless of any treaty. An intervention controls flows; the atmosphere responds to stocks. Any future composition regime for a long-lived species inherits that gap between the act and the effect.

Established The second lesson is about time. Detection of the effect of the best-monitored, best-funded, most legible atmospheric intervention in history was still a publishable claim thirty years after the treaty was signed, and its projected completion date has been moving later between assessments. Speculative A civilization that intends to manage atmospheric composition is committing to measurement programmes that outlast the governments that fund them and the careers of everyone who starts them.

Frontier Third, and least comfortable: the tractable version of this subject is the one nobody calls geoengineering. Tropospheric ozone is a greenhouse gas, a crop toxin and a respiratory hazard, produced photochemically from precursors already regulated for air quality in most rich countries. Speculative It is the one atmospheric-composition problem where mitigation and management are the same action and the institutional machinery already exists — and it attracts almost no attention in the intervention literature, apparently because it is not exotic. Handwave That is a fact about what the field finds interesting, not about what would work.

Speculative Fourth, and the one structural advantage this subject has over its neighbours: methane management is reversible on a human timescale. Because methane's lifetime is a decade rather than a century, stopping an oxidation-enhancement programme returns the atmosphere to its baseline within a generation, with none of the termination dynamics that dominate the solar-geoengineering literature. Frontier That is the strongest argument available for taking this route seriously, and its own proponents rarely make it.

These horizons track measurement capability and assessment cycles rather than deployment, because nothing here is close to deployment:

• 10 yr: Established One date is fixed: the 2026 Scientific Assessment of Ozone Depletion reports, and will treat stratospheric aerosol injection prominently for the first time. Frontier Expect the iron-salt roadmap's five-year programme to produce laboratory and modelling results and to reach, at most, a small instrumented outdoor release; expect that release to be contested on notification grounds rather than on chemistry. Frontier Expect the 2020–2022 surge attribution to be revisited and possibly settled, because the data already exist and the disagreement is now visible enough to be embarrassing. Speculative Expect no demonstrated reduction in global methane concentration attributable to a deliberate intervention in this window, and expect the ozone completion date to move later again.
• 25 yr: Speculative If the sign question resolves favourably and a side-effect budget closes, the plausible state at this horizon is a regionally targeted oxidation-enhancement capability of contested efficacy, deployed at a scale below the detection threshold of the network watching it — which is the worst of both worlds and the most likely single outcome. Speculative The competing trajectory is that chamber chemistry finds the sign is unstable across realistic ambient regimes, the route is abandoned, and methane management collapses back into emission control, where the cheap abatement is. Frontier Either way the ozone recovery should be unambiguous in the record by then, which will be the first time a planetary intervention has been carried to a measured completion.
• 50 yr: Speculative At this horizon the interesting question is whether a permanent composition-monitoring institution exists — something with the standing of a weather service, mandated to measure the sinks rather than only the concentrations. Speculative Nothing currently proposes one. Handwave If enhanced-oxidation management ever becomes routine, its termination properties are benign compared with solar geoengineering, because methane's lifetime is a decade rather than a century: stopping returns the atmosphere to its baseline within a generation. That is the strongest structural argument in this subject's favour and it is rarely made.
• 100 / 250+ yr: Handwave Beyond forecasting. The only defensible statement is a structural one: the three non-CO₂ gases have permanently different intervention profiles — methane has an enhanceable sink, halocarbons had identifiable producers, and nitrous oxide has neither, being chemically inert in the troposphere with its only sink in the stratosphere. Handwave A civilization managing its atmosphere at this horizon would be managing three unrelated problems that share a name.

• Depends on This brief waits on one thing another brief could supply, and it is a measurement rather than a technology: Weather Modification holds the empirical base rate for atmospheric attribution — eighty years of a smaller, cheaper, more frequently repeated intervention that still cannot demonstrate its effect at the scale it is sold. The methane case is the same problem harder. Otherwise the dependencies here are scientific results nobody is producing, which is why the requires row carries the weight.
• Requires (not on this map) Neither constraint is a research result and neither is exotic; both are things somebody would have to decide to pay for and neither has been decided. The first is an owner for the science. The field's own agenda paper stated in 2021 that no research framework exists for methane removal and proposed a model intercomparison as the remedy; five years on, no intercomparison has been published, because such a project needs a coordinating body, a common scenario protocol and a standing budget, and methane removal has none of the three. The consequence is not slow progress but an unmeasured error: with no intercomparison, nobody can say how much of the removal literature's optimism is structural to the models generating it, and the same absence leaves the field with no home for the observing-system specification that would define what detection of a deliberate global perturbation even means. The second constraint is the one this brief most wanted to quantify and could not: the fiscal standing of the ozone observing network. The agency and treaty-body sources that would answer it were unreachable to this brief's research pass, so no budget figure, station count or programme termination is asserted here, and any such number encountered elsewhere should be traced to a primary source. What is documented is the technical case, and it argues the constraint on its own. Ozonesonde precision improved by a factor of two over twenty years through strict standard operating procedures rather than new hardware — that is, through sustained intercomparison campaigns and training, which are recurring operating costs rather than capital ones, and therefore the first line cut when a network tightens. Stability against satellite comparison is about ±2% and agreement with reference instruments about 3%, while a change in the sondes themselves produced a sudden drop in total column ozone at 14 of 37 stations between 2014 and 2016. A record that sensitive to consumable hardware survives on continuity, not on capital grants. And the structural point stands independently of any appropriation: the compliance mechanism of the Montreal Protocol is a modest set of measurement stations run by research groups on competitive funding, the CFC-11 source was localised only because two of them happened to sit downwind, and the episode's own summary literature concluded that better monitoring networks are needed for enforcement. An enforcement capability that depends on grant cycles and favourable geography is a market constraint on the whole subject.
• Enables Everything on this map that presumes atmospheric verification rests on the observing system this brief describes. Climate Engineering and Geoengineering Governance both identify detection and attribution as the unfunded precondition for any regime; this brief holds the instruments and can say what they can and cannot do — about ±2% network stability, a documented step artefact at 14 of 37 ozonesonde stations, and a 0.24-year structural spread in the methane lifetime. Planetary Cooling Concepts inherits the iron-dust question from the other direction, since a proposal to double atmospheric iron is also an ice-nucleating-particle intervention.
• Adjacent Atmospheric chemistry and photochemical kinetics; chemistry-climate and chemical transport modelling; isotope-ratio mass spectrometry; satellite trace-gas retrieval; aerosol simulation-chamber facilities; environmental treaty compliance monitoring; air-quality regulation, which already owns tropospheric ozone; and marine biogeochemistry, which owns what happens when the iron lands.

Established “The Montreal Protocol fixed the ozone layer.” Not yet, and the finish line has been moving away. The 2010 multi-model assessment projected return to 1980 levels around 2025–2030 for the Arctic and 2045–2060 for the Antarctic; the dates in circulation now are later still. Established Recovery detection was a publishable claim as late as 2017, thirty years after signature. It is the best environmental treaty there is and the layer is not fixed.

Established “The Montreal Protocol proves treaties can solve atmospheric problems.” True, and the usual three caveats — identifiable producers, drop-in substitutes, legible harm — are correct but incomplete. Established The stronger caveat is the bank. After illicit production stopped, 90 to 725 gigagrams of CFC-11 remained banked in installed foam and equipment and will emit for decades. A regime that successfully halts production still faces the stock it cannot reach, and the estimate of that stock spans a factor of eight. Controlling flows is not controlling stocks, and almost nobody making the ozone argument makes this point.

Established “The CFC-11 mystery was solved by satellites.” It was solved by a handful of ground stations that happened to be downwind. The localisation to Shandong and Hebei rested on AGAGE-affiliated observations from South Korea and Japan; the episode's summary in Eos states the lesson explicitly as needing better monitoring networks for enforcement. Frontier And the attribution may not be final: a 2025 paper argues coal combustion was an overlooked CFC-11 source, which if correct makes the policing problem worse rather than better.

Established “Catalysts can now remove methane from the air.” They can oxidise methane at 0.1–1% by volume, which is 500 to 5,000 times atmospheric concentration. Established The papers say “low concentration” and “ultralow concentration” and mean coal-mine ventilation air. Established Name the concentration whenever a catalytic result appears in a removal context.

Established “Methane removal is direct air capture for a shorter-lived gas.” This is the most consequential error in the subject. CO₂ is at about 420 ppm; methane is at about 1,860 ppb — a factor of roughly 220 lower. Established Because both the minimum work of separation and the air volume handled scale adversely with dilution, an approach that is expensive at 420 ppm is not merely more expensive at 1.9 ppm: it is in a different regime. Established That is precisely why the serious literature abandoned capture and moved to enhancing atmospheric oxidation in place.

Frontier “Iron salt aerosol removes methane; the only question is scale.” The sign is not established. The published statement is that removal per iron atom is net positive below 50 ppt or above 390 ppt of NO_x — a claim about where it works, and by implication where it may not. Established The field's own roadmap paper is a five-year programme to find out whether the approach is viable at all. Anyone presenting this as a scaling problem is skipping the science its proponents say has not been done.

Frontier “Enhancing the atmosphere's oxidising capacity is clean because it uses natural chemistry.” Natural chemistry is coupled chemistry. OH and chlorine set the lifetimes of carbon monoxide, of volatile organic compounds, and of the HFCs and HCFCs that replaced the CFCs; chlorine chemistry makes tropospheric ozone. Frontier The first systematic look at the side effects is a 2024 preprint. Established “It uses natural processes” is the argument that was made for ocean iron fertilisation.

Established “The ozone record is a clean, continuous series.” Between 2014 and 2016, total column ozone from ozonesondes dropped suddenly at 14 of 37 stations, and the cause was traced to electrochemical concentration cell instrument artefacts rather than to the atmosphere. Established Network stability against satellites is about ±2% and the trends being measured are of comparable size. This is not a reason to doubt recovery. It is a reason to be exact about what “measured” means, especially when a compliance judgement rests on it.

Established “Methane's lifetime is about nine years, so this is arithmetic.” The lifetime is a model output with a large structural spread: 0.24 years of difference from the choice of meteorological forcing alone, a roughly 9% modelled OH increase over 1980–2014 in one intercomparison, and roughly constant global mean OH over 150 years in another because regional changes offset. Frontier Quote a lifetime only with its uncertainty and its source.

Established “Methane removal would let us avoid cutting emissions.” Nobody in the field says this, and two of its central papers say the opposite unprompted in their abstracts. Frontier The case against removal as a priority is real: methane's short lifetime means emission cuts deliver temperature benefits within about a decade, faster than any other lever. Speculative The counter is also real — a substantial share of emissions comes from sources with no known abatement route at any price, including wetlands responding to warming, and for that residual a removal capability is the only conceivable instrument. This brief puts no percentage on that share, because the figures in circulation could not be traced to a primary source.

Established “CFC-11 emissions were measured in gigatons.” They were not, and this exact error was generated during the research behind this page by an automated summariser rendering “13 ± 5 Gg” as “13 ± 5 gigatons”. Established Global CFC-11 emission is tens of gigagrams per year — tens of thousands of tonnes — and the 2019 value was 52 ± 10 Gg per year. Any figure for it in gigatons is wrong by nine orders of magnitude, and this subject mixes gigagrams, teragrams, megatonnes and gigatonnes freely across its sources.

Frontier “Machine-assembled bibliographies are reliable enough for assessment work.” The research behind this brief caught its own tools substituting one document for another three times: a nine-order-of-magnitude unit error in a summarised abstract, a fabricated author list returned for a real paper, and a bibliographic query that returned an unrelated cryobiology result set. Established None produced an error message. Frontier This matters here specifically, because atmospheric composition is governed by large assessment reports assembled by exactly these methods, and because the failure mode is silent. Every author list on this page was taken from structured metadata rather than from summarised prose, for that reason.

Frontier “The 2020–2022 methane surge is understood.” It is not. One Nature paper assigns 53 ± 10% of the 2020 growth to a weakened hydroxyl sink; one PNAS paper reading carbon isotopes over the record 15.4 ppb-per-year surge explicitly rejects that explanation; one satellite-inversion study in ACP attributes it predominantly to emissions. Established The removal literature does not acknowledge the disagreement, and every hydroxyl- or chlorine-based removal proposal presumes an answer to it. That is the honest terminal position of this brief: the technology proposes to move, measurably and on purpose, the one quantity in the atmosphere the field cannot currently agree moved by itself.