Evaluating the potential of iron-based interventions in methane reduction and climate mitigation

Keeping global surface temperatures below international climate targets will require substantial measures to control atmospheric CO2 and CH4 concentrations. Recent studies have focused on interventions to decrease CH4 through enhanced atmospheric oxidation. Here for the first time using a set of models, we evaluate the effect of adding iron aerosols to the atmosphere to enhance molecular chlorine production, and thus enhance the atmospheric oxidation of methane and reduce its concentration. Using different iron emission sensitivity scenarios, we examine the potential role and impact of enhanced iron emissions on direct interactions with solar radiation, and on the chemical and radiative response of methane. Our results show that the impact of iron emissions on CH4 depends sensitively on the location of the iron emissions. In all emission regions there is a threshold in the amount of iron that must be added to remove methane. Below this threshold CH4 increases. Even once that threshold is reached, the iron-aerosol driven chlorine-enhanced impacts on climate are complex. The radiative forcing of both methane and ozone are decreased in the most efficient regions but the direct effect due to the addition of absorbing iron aerosols tends to warm the planet. Adding any anthropogenic aerosol may also cool the planet due to aerosol cloud interactions, although these are very uncertain, and here we focus on the unique properties of adding iron aerosols. If the added emissions have a similar distribution as current shipping emissions, our study shows that the amount of iron aerosols that must be added before methane decreases is 2.5 times the current shipping emissions of iron aerosols, or 6 Tg Fe yr−1 in the most ideal case examined here. Our study suggests that the photoactive fraction of iron aerosols is a key variable controlling the impact of iron additions and poorly understood. More studies of the sensitivity of when, where and how iron aerosols are added should be conducted. Before seriously considering this method, additional impacts on the atmospheric chemistry, climate, environmental impacts and air pollution should be carefully assessed in future studies since they are likely to be important.


Introduction
Controlling atmospheric concentrations of methane (CH 4 ) is a critical and complex challenge to address climate change (IPCC Core writing team 2023).Methane is a potent greenhouse gas that contributes significantly to global warming due to its high radiative forcing (RF) and heat-trapping capabilities (Intergovernmental Panel on Climate Change (IPCC) 2023).Both natural and human activities contribute to methane emissions, and its atmospheric growth rate has been increasing since 2007 (IPCC 2021).Chemical reactions of methane with hydroxyl radicals (OH) and, to a lesser extent, atomic chlorine (Cl), are its main sinks in the atmospheric (Saunois et al 2020).It is largely the abundance of OH that determines the rate of methane oxidation and ultimately its chemical lifetime in the atmosphere.In addition to its role in global warming, methane also contributes to the formation of ozone, which also has direct implications for global warming, air pollution and human health, highlighting the idea that addressing methane can meet both air quality and climate goals (O'Grady 2021, United Nations Environment Programme and Climate and Clean Air Coalition 2021, Staniaszek et al 2022).
Recent research by (van Herpen et al 2023) has argued that there is link between North African mineral dust and chemical methane removal through a photocatalytic cycle for chlorine atom production, and that the proposed mineral dust-sea spray aerosols (MDSA) mechanism can explain the measurements of isotopes of carbon monoxide (CO) over the North Atlantic.More specifically, iron aerosols in the atmosphere catalyze the production of chlorine gas from sea salt aerosols through a complex photochemical process involving the release of Fe (III) chlorides, sunlight activation, and subsequent reoxidation, ultimately contributing to atmospheric halogen release (Wittmer et al 2015, Wittmer andZetzsch 2017).This opens up the possibility of decreasing atmospheric methane by increasing the atmospheric production of chlorine (Cl) by adding iron aerosols (Dietrich Oeste et al 2017).Laboratory studies show that when sea salt and iron aerosols interact, iron chloride species are formed and can generate chlorine molecules (Cl 2 ) through photochemical reactions (Wittmer et al 2015, Wittmer and Zetzsch 2017, Mikkelsen et al 2024).Chlorine is highly reactive in the atmosphere, and will react with both ozone and methane, as well as other chemical species (Saiz-Lopez and von Glasow 2012, Wang et al 2019, Li et al 2023).The emissions of molecular chlorine (Cl 2 ) above the global ocean surface over a certain threshold (∼90 Tg Cl yr −1 ) have been shown to remove methane (Li et al 2023).However, below this threshold adding chlorine results in an unwanted increase of methane, attributed to the reaction of chlorine with ozone and the resulting decreased production of OH, the main chemical sink for methane (Horowitz et al 2020, Li et al 2023).However, if sufficient chlorine is added, this approach offers a potential pathway to decrease atmospheric methane levels.
Adding iron-containing aerosols to the atmosphere has significant environmental implications.Although the net radiative effect of iron-containing aerosols remains uncertain, they absorb short-wave solar radiation (Matsui et al 2018), contributing to climate-warming aerosols in the atmosphere.Aerosol cloud interactions (ACI) from added anthropogenic aerosols are well studied, but still highly uncertain (e.g.Kikstra et al 2022), and may also result from the addition of iron aerosols.In this paper we emphasize the iron aerosol specific impacts such as impact the radiative balance of the atmosphere through modifying cloud properties, leading to cooling through enhanced cloud brightening or warming through cloud burn-off (not quantified in this work) (Fan et al 2016).Aerosols such as iron are also air quality hazards (Lim et al 2012).Deposition of iron into the open ocean can stimulate phytoplankton productivity, impacting on ocean biogeochemistry (Moore et al 2006, 2013, Mahowald et al 2018, Ito et al 2021, Hamilton et al 2022).Thus, implementation of interventions such as adding iron to the atmosphere requires careful consideration, as the implications and potential side and indirect effects need to be thoroughly investigated.
In the atmosphere, iron primarily originates from desert dust aerosols (95%) which is approximately 3.5% iron, with about 5% of iron aerosols deriving from industrial, maritime shipping and biomass burning combustion sources (Mahowald et al 2008, Hamilton et al 2023).The current-day iron global emission from industrial and combustion sources is estimated to be 1.5-7.4Tg Fe yr −1 (Ito and Miyakawa 2023).Iron in aerosols can have very different chemical states, from highly insoluble iron oxides, to more soluble iron in clays, to highly soluble iron in nanoparticles or other forms such as ferric sulphates (Shi et al 2009, Hamilton et al 2022).Soluble iron is thought to be the most relevant for ocean biogeochemistry, as it is more bioaccessible (Jickells et al 2005).In some regions of the open ocean, new primary production is limited by the low supply of bioavailable iron from the combination of atmospheric deposition, recycling, and ocean upwelling and mixing (Moore et al 2013, Tagliabue et al 2017).Thus, there is the potential that additional ocean bioavailable iron input from an iron aerosol-CH 4 intervention could enhance regional ocean productivity and potentially the uptake of carbon dioxide (Jickells et al 2005, Hamilton et al 2022) because of the importance of atmospheric iron deposition to ocean biogeochemistry there has been substantial work on modeling and observations of soluble and total iron (e.g.Mahowald et al 2008, Myriokefalitakis et al 2018, Ito et al 2019).Photochemically active iron is more rarely studied, but is a fraction of soluble iron (Zhu et al 1997, Chen andSiefert 2003).
This study, for the first time, uses a set of global models to explore the complex relationship between iron emissions, molecular chlorine production, and their impact on methane abundance, with special emphasis on the potential effectiveness of iron sea salt aerosols as a methane removal mitigation strategy.Using different iron emission sensitivity scenarios, we examine the potential role and impact of iron emissions on direct interactions with solar radiation, and on the chemical and radiative response of methane.First, we explore the spatial effectiveness of different scenarios for iron emissions on atmospheric methane levels.We analyze the regional variations in methane removal efficiency from iron aerosols, classifying different regions based on the corresponding methane response.Second, we examine the relationship between iron emissions and molecular chlorine production, investigating the factors and regimes that control this important step of the process.Finally, we assess the potential changes in future RF and surface temperature due to additional iron emissions, in order to determine whether the anticipated benefits of iron emissions are substantial.Additionally, we highlight in the supplementary information the main uncertainties that should be considered when interpreting our findings.

Community earth system model description
In order to consider iron-chlorine-methane interactions we need: (i) a model of soluble atmospheric iron, (ii) a parameterization for the chlorine production from atmospheric soluble iron, and (iii) a model of atmospheric chlorine chemistry.MIMI is an improved atmospheric iron cycle module embedded within the Modal Aerosol Model version 4 (MAM4) (Liu et al 2016) that simulates the emission and atmospheric (acidic and organic) processing of iron in aerosols from both mineral dust and combustion sources.In particular, it simulates the time-evolving fraction of atmospheric iron that is soluble.The model includes dust and combustion iron emissions, which are transported in the model, and removed by wet and dry deposition (Hamilton et al 2019).The iron is divided into four types (slow reacting iron in dust, medium reacting iron in dust, industrial combustion iron, and fire iron) and each type is split into soluble and insoluble.For the optical properties of combustion soluble and insoluble iron we use the same optics as insoluble iron from mineral dust.Additional information on the iron optical properties can be found in the SI.More details on the iron species and their processing in the model can be found in (Hamilton et al 2019) (2007)(2008)(2009)(2010) and discard the first year as spin up, using the last 3 years for this study.The iron is added from combustion only above the ocean (i.e.iron from shipping), for both the soluble and insoluble iron species.Note that adding combustion iron as we do here is different from geoengineering proposals to spray FeCl 3 (Meyer-Oeste 2010, Dietrich Oeste et al 2017, Sturtz et al 2022, Mikkelsen et al 2024), a material that when photolyzed will produce photoactive iron (Mikkelsen et al 2024) and will have different evolution and effects than evaluated here.
Second, we calculate the Cl 2 production using the soluble iron global distribution as determined in MIMI and assuming instant mixing between the iron and sea salt aerosols (wherever the sea salt aerosol concentration is above 10 −9 kg kg −1 ).Note that this assumption will lead to an underestimate of the mixing and coagulation time scale (Carslaw 2022) and therefore an overestimation of the production.Additionally, for the Cl 2 production calculation we assume that the global percentage of photoactive iron to soluble iron is 31.6%, as was measured in Barbados (Zhu et al 1997).Laboratory studies show that when sea salt and iron aerosols combine in the presence of light, iron chloride species are formed and can generate chlorine molecules (Cl 2 ) (Wittmer et al 2015, Wittmer and Zetzsch 2017, Mikkelsen et al 2024).We argue that it is only the photoactive iron that is able to produce chlorine (van Herpen et al 2023).While the limited observations suggest that photoactive iron is a subset of soluble iron, we do not know from field or laboratory studies the sources or attributes of photoactive iron.Here we simply assume the measured percentage at Barbados (Zhu et al 1997) holds everywhere.This approach was used in (van Herpen et al 2023) to simulate the impact of iron aerosols from dust on atmospheric methane, although (van Herpen et al 2023) assumed that 1.8% of iron is soluble and photoactive, while in the current study the soluble fraction as simulated in MIMI produces an average of about 2% photoactive iron (figure S2soluble iron).Since the current suggested approach for decreasing atmospheric methane from iron additions is to add more iron to the exhaust from shipping (Dietrich Oeste et al 2017), we assume that the added iron has similar properties to the shipping iron currently included in the model.This allows the added iron to start out more soluble than dust (6% versus 1%) and for the insoluble portion to be more quickly converted to soluble with a faster reaction rate, as described in (Hamilton et al 2019, 2020, Rathod et al 2020).
The amount of Cl 2 produced from iron aerosols each timestep in a given grid box is a function of the incoming solar radiation at that time and place, divided by the solar radiation at Barbados from the observations of (Zhu et al 1997), multiplied by 11.4 photochemical cycles per hour, as measured by (Zhu et al 1997).While simple, this approach uses all available observations as much as possible and allows analysis to be extrapolated to other latitudes and seasons.Our approach gives a similar production of Cl 2 from mineral aerosols over the North Atlantic as (van Herpen et al 2023) which allowed the model to simulate measurements of CO isotopes (Mak 2003).
Third, we added the calculated molecular chlorine produced into the surface layer of the climatechemistry version of CESM1-CAM4 (CAM4-chem).CAM4-chem includes a comprehensive representation of atmospheric chemistry including comprehensive halogen sources and chemistry (Li et al 2022).The simulations are run at a spatial resolution of 2.5 • × 1.9 • × 26 (longitude by latitude by vertical layers) using a prescribed SST and a standard land model.The boundary layer scheme is described in Hurrell et al (2013) and Park and Bretherton (2009).Additional chlorine sources are included in the model used in this study and are detailed in Li et al (2022).The model has an enhanced representations of natural short-lived halogens (Cl, Br, I) calculated online (Iglesias-Suarez et al 2020).Anthropogenic chlorine species and their processes are modeled based on (Keene William et al 1999, Hossaini et al 2016, Claxton et al 2020, Li et al 2022).The simulations are conducted within the CESM1 framework in a free running mode, using the prognostic weather and climate mode to capture atmospheric composition-climate feedbacks for 2020-2030, preceded by a 60 year spin-up .CH 4 is added by a transient emission inventory instead of lower boundary conditions, as described in Li et al (2022), to maintain free atmospheric chemical interactions of global methane.Li et al (2022) showed the global O 3 and tropospheric OH concentrations at the end of the spin-up period are in good agreement with observation and previous modeling studies.Molecular chlorine produced from iron is added as a surface emission with a diurnal cycle.We use other gas and aerosol emissions from 2020 to 2030 from the representative concentration pathway 8.5 (RCP8.5)as the main baseline scenario (henceforth designated as Base).

Scenario description
To understand the impact of additional molecular chlorine from iron aerosols in today's climate prior to any additional iron emissions (argued in (van Herpen et al 2023), and not included in current CESM models), we first simulate a baseline case using the present-day production of iron aerosols, as simulated in MIMI along with the parameterization of molecular chlorine production as discussed above.Present day iron emissions from maritime shipping are implemented in this default version of the model (Hamilton et al 2019, Rathod et al 2020).The resulting impact on methane gives similar results as (van Herpen et al 2023) (BASE case in table 1).From this base case, we introduce a series of scenarios to explore how interventions by adding additional iron aerosols impact atmospheric methane.In the first scenario, we assume additional iron emissions occur at a constant rate over all ocean surfaces (henceforth labeled as OC) (see table 1).We explore iron emissions in ship tracks (ST) in a second group of scenarios where we assume additional iron is emitted from maritime shipping activities.This would offer an efficient and readily available means of introducing additional iron into the atmosphere.We increase iron emissions in ship tracks, which are currently 2.2 Tg yr −1 in the base scenario by 100-, 500-, and 1,000 times henceforth labeled as ST100, ST500, and ST1000, respectively.Finally, in a group of box scenarios we restrict iron emissions to a regional box to better identify if there are particularly effective regions for emitting iron.In the box scenarios, we emit iron (kg m −2 s −1 ) inside 10 × 10 model grid-boxes (1.25 • × 0.9 • ) at one of three constant rates (henceforth labeled as low, mid, and high).The box emission regions are the North Pacific, North Atlantic, Southern Ocean, and equatorial Pacific (henceforth labeled as NP, NA, SO, and EP, respectively).We chose these regions in order to explore the importance of the efficiency of methane removal (discussed below), choosing some regions that are more efficient (NP, NA, EP) and some that are less efficient (SO).We summarize all twenty simulated cases in table 1.In each case we assess both the iron aerosol production of Cl 2 and then the impact of the Cl 2 on the methane chemistry.For all these cases we assume that the additional iron emitted has the same solubility and atmospheric processing attributes as shipping iron, which starts much more soluble and is more easily solubilized than dust (Hamilton et al 2019).The chlorine production from the resulting photoactive iron is calculated as discussed above.
In order to show that our results are robust to our modeling assumptions, we performed three sensitivity tests of our approach with the ST1000 iron emission scheme.These scenarios are described in the supplementary information.

MAGICC description
In this paper we include only a 10 year simulation of the impact of iron additions on CH 4 .However as this is not long enough to calculate the overall radiative effect, we extrapolate the 10 years of simulations to 2050 using the reduced-complexity Model for the Assessment of Greenhouse Gas Induced Climate Change version 6 (MAGICC6) (Meinshausen et al 2011a(Meinshausen et al , 2011b) ) to simulate the change in radiative effects and surface temperature relative to 1850-1900 resulting from the different scenarios.In particular, we extrapolated the global percentage change of O 3 and CH 4 burden compared to the base scenario as given in the CAM4-chem simulations.MAGICC6 comprises four boxes representing land and ocean in both the northern and southern hemispheres.We drive MAGICC6 with RCP8.5 initial concentrations for most species and timeseries (2020-2050) but use the extrapolated ratios to change the surface CH 4 and O 3 .To account for the iron RF, as calculated in MIMI, we use the percentage radiative change of iron (between the different scenarios and the Base scenario) and increase the black carbon (BC) aerosols in MAGICC6 to match the RF from added iron aerosols.Finally, we simulate the change in surface temperature and RF until 2050 across all scenarios.Our goal is to demonstrate the potential influence of these mitigation scenarios on surface temperature and RF, defining the change as the global surface temperature change relative to 1850 due to potential future iron emission additions.We use MAGICC because the lifetime of the aerosols is quite different than that of CH 4 , and MAGICC allows us to calculate the impact of a long term addition of iron aerosols onto climate and extrapolate the CESM simulations from 10 years to the more climate relevant 30 years.

Methane removal efficiency calculation
To evaluate how effective the iron addition will be at removing methane, we introduce the concept of methane removal efficiency.We can define the efficiency either globally or locally.The global methane removal efficiency is given as: where the change in total CH 4 loss rate (Tg/yr) is the change in the global CH 4 loss rate over 10 years compared to the Base case, and the change in iron is the total change in iron added from emissions in 10 years, or equivalently at steady state, the total amount of iron removed by deposition.Note that the change in methane burden from the base case includes changes from all chemical loss pathways due to iron addition (i.e.OH, Cl, and, photochemical reactions).The global efficiency gives an overall metric of the efficiency of the iron additions in destroying methane.
To identify where the iron additions would be most effective, we consider the efficiency calculated on a spatial basis.The spatial efficiency is calculated for each grid box as the change in column methane loss, divided by the change in iron flux in that grid box, Locally the iron flux could be calculated as the emission or the deposition of iron, and we explore the utility of using both metrics.While in the global simulation, at steady-state, the iron emissions and deposition are the same this is not true locally.

Efficiency analysis
One immediate question that arises when considering iron addition is if there are some regions in which adding iron destroys more methane than other regions.We use constant emissions over the entire ocean (OC) scenario to calculate the spatial efficiency since the results are easier to interpret than other emission scenarios.The spatial efficiency reveals distinctly different spatial patterns when considered based on iron emission or deposition (figure 1(a) versus figure 1(c): equation ( 2).The efficiency based on emissions is solely determined by the methane loss since the emissions are constant (E2).This results in maximum (minimum) efficiency in regions of large methane losses (small methane losses).For example, the efficiency based on emissions (figure 1(a)) is comparatively high in the Pacific Inter Tropical Convergence Zone (ITCZ) where the EP box scenario is located (figure 1(a)), indicating large methane losses in this region.On the other hand, calculating the efficiency by dividing by the iron deposition rate instead (figure 1(d)) is dependent both on the methane loss and on the deposition rate.These results show that higher (lower) efficiency occurs in the regions with smaller (larger) deposition rates, or longer (shorter) iron lifetimes.In this case the high precipitation rates in the Pacific ITCZ (i.e.EP scenario) with its high rates of wet removal (figure 1(d)) result in a low efficiency.Since these two measures of efficiency give different results, we examine the methane removed by explicitly considering the different box scenarios, in the next section, where we show that using the deposition to define the efficiency locally better reflects the regional box results.

Variations in methane removal efficiency
Our analysis reveals significant regional variations in global methane removal efficiency (figure 2) depending on where the iron is emitted.The NP and NA scenarios have the most methane removed with additional iron emissions, indicating that specific regional dynamics, such as favorable locations for iron emissions and associated molecular chlorine production, can contribute the most to increased methane removal in these regions (figure 3).In contrast, the SO and EP scenarios have lower global methane removal (figure 2) with less chlorine produced per iron added (figure 3).
The removal efficiency is also impacted by the amount of iron added regionally.The nonlinear relationship between the addition of iron and methane is similar to that between chlorine and methane from other studies (Horowitz et al 2020, Li et al 2023) (figure 2).In all regions methane increases when additional iron emissions are low.The nonlinearity is caused by the additional produced Cl 2 reacting with both ozone and methane, where the reaction with ozone decreases the production of OH and thus increases CH 4 , while the reaction with CH 4 decreases its concentration.When enough Cl2 has been produced, ozone becomes sufficiently decreased and no longer limits the reaction with methane, thereby effectively decreasing methane concentrations.Notably, even in the ST100 ship track scenario, where additional iron emissions are more than 16 times the current global combustion iron emissions, no visible decrease in methane burden was observed.
Also note, that at sufficient iron addition rates (>5-9 Tg −1 yr) the box scenarios all result in larger methane losses than the low ship-track (ST100) scenario, suggesting that focusing on one or two regions could be more efficient.

Regional efficiency variations
Next, we will discuss the reasons for the regional differences, focusing on the regional variations in chlorine production and its reactivity towards ozone and methane.
Chlorine production per emitted iron differs by over a factor of two between different regions (figure 3).Fundamentally the chlorine production depends on how photoactive the iron is, how long it lasts in the atmosphere, and the incoming solar radiation.The production ratio of Cl 2 to iron is 5:1 g Cl 2 /g iron in the ship track scenario (figure 3).The scenario where iron is emitted constantly above the ocean (OC scenario-green circle in figure 3) gives  similar results as the ST scenario (black diamonds in figure 3).However, emissions in the more efficient regions, such as the boxes in the North Atlantic (NA) and North Pacific (NP), results in a high Cl 2 production ratio that increases to 7:1 g Cl 2 /g iron (figure 3).These efficient scenarios (boxes NA and NP) are best predicted by those regions where the local efficiency is defined in terms of the iron loss rate (figure 1(c)).This again suggests the regions with small iron deposition loss are most efficient at removing methane, resulting in a longer iron lifetime and a greater production of Cl 2 from the iron.The fact that we do not see a strong dependence of Cl 2 production on incoming solar radiation between the NA and NP scenarios is likely because of the similar latitudes chosen.1).Dotted lines represent slopes deduced from the different cases, where the upper lines (7:1 gCl2/gFe) come from the more efficient NP and NA regions, while the lowest dotted lines (3:1 gCl2/gFe) come from the less efficient SO region.The same scenarios and color scale are used as in figure 2. Figures S5 and S11 show the Cl2 produced and iron emission for all the scenarios.
The Cl 2 is assumed to be produced from a globally constant fraction of the soluble iron (as described in the methods) multiplied by the amount of insolation at a given location.Consequently, even globally constant rates of oceanic iron emissions (in the OC scenario), will not result in constant Cl 2 emissions (as was assumed in (Li et al 2023)), since the iron is removed at unequal rates, there are differences in soluble iron composition, and there are differences in insolation (figure 1(d)).
The efficiency of methane removal from iron addition between different regions is also impacted by the extent to which chlorine reacts with methane or with ozone as demonstrated previously in (Li et al 2023).Therefore, the global emission scenarios (OC and ST) are less efficient at removing methane per added iron, consistent with the fact that global ozone should be removed before the chlorine effectively reacts with methane.
We define the chlorine reactivity sensitivity (CRS) as the rate chlorine reacts with ozone divided by the rate at which it reacts with methane: Chlorine reactivity towards ozone is higher where the ozone concentrations are higher (figure S6), i.e. in the northern hemisphere (figure 4 S7).However, as shown above the reactivity towards methane is not the prime determinant in the location of the most efficient regions, which is dominated by iron lifetime.

Future RF and surface temperature change
Here we consider the net impact of adding iron aerosols to the climate, using the output from the CESM models as well as the MAGICC model.When comparing the different scenarios to the Base scenario, the ozone RF decreases in all cases (figure 5), similar to what was seen in (Li et al 2022).In some cases, the methane burden and thus its RF also decreases (in the scenarios with higher iron emissions), while in other cases the methane burden and RF increases (figure 2).On the other hand, the presence of additional iron in the atmosphere, particularly in the form of iron oxides, has a warming effect on climate, although the strength of this warming is uncertain (Zhang et al 2015, Matsui et al 2018, Li et al 2021) (figure 5(b)).Notice that depending on where and how much iron is added there is a difference in the effects of iron aerosols on direct radiative effect (DRE) (figure 5(c)).For example, the difference between NP and EP in the iron direct effect is a result of the overall albedo change produced by the aerosols, which is very sensitive to the background albedo (including cloud distribution and land surface).
The addition of iron decreases the overall RF in the NP scenario the most due to comparatively large decreases in the RF due to O 3 and CH 4 removal and comparatively small increases in the direct effect from the iron additions.However, adding iron does not always result in a net cooling because of increased iron aerosol solar absorption (figure 5).In all EP scenarios (i.e.low, mid, and high), the increase in DRE attributable to iron is higher than the decrease in RF due to the combined effects of ozone and methane (figure 5(b)).Adding iron in the HighEP scenarios increases the RF the most of all the scenarios examined compared to the base case due to the warming from the iron aerosols (figure 5(b)).In all low scenarios there is an increase in the methane-related RF but this is often compensated for by the decrease in the ozone RF.
Overall, in most scenarios, the increase in RF directly attributable to iron partially or wholly masks the beneficial reduction resulting from ozone and methane loss (figure 5(b)).In all cases, the resulting changes in surface temperatures by 2050 are very small across all scenarios (figure S8).Even in the highest scenarios analyzed, ST1000 and OC, in which almost 200 Tg of iron are added, the surface temperature reduction is, at most, approximately a tenth of a degree (figure S8).The uncertainties in MAGICC for surface temperature in the 2050 are large and increase in 2100 (Beusch et al 2022).The projected temperature uncertainties are as high as a degree, therefore, the highest modeled value of 0.12 • C decrease in the ST1000 scenario is not a significant change.
Although well studied, there is a very large uncertainty in ACIs for anthropogenic aerosols, as can also be seen in the latest IPCC report where the uncertainty in the effect of ACI is as large as the effect itself (Kikstra et al 2022).In this study we focus on the unique response to adding iron aerosols.Quantifying the effect of ACI on the global radiation budget carries a significantly larger uncertainty compared to the DRE of methane, ozone or iron aerosols, although the magnitude of ACI effect may well swamp the impact of adding iron specifically (figure S9).Thus, we do not include the impact of ACI of the additional anthropogenic aerosols into our MAGIC simulations, although this should be explored in the future.Additionally, absorbing aerosols like iron aerosols might also result in cloud burn-off (Fan et al 2016), but this effect is not included in our simulation.

Scenario overview
The CH 4 removed per Cl 2 produced maximizes at approximately 0.12 (Tg/Tg) with large iron additions in most of the scenarios examined (table 2).However, the maximum amount of CH 4 removed per iron added is very sensitive to the particular scenario, with the most efficient scenarios the boxes in the North Pacific and North Atlantic regions.This is due to iron losses as discussed in the regional efficiency variations section.These are also the most efficient scenarios for the Cl 2 produced per iron added and for the net climate effect of iron additions.However, despite  somewhat lower efficiencies in CH 4 removed per added iron the ST1000 scenario is almost as effective in the net climate effect of iron additions.However, despite their efficiency, the net change in RF (table S2) and surface temperature by 2050 remain relatively low (figure S8).

Summary and conclusions
In this paper we focus on the challenge of mitigating atmospheric methane to combat climate change by investigating enhanced oxidation of methane interventions by artificially adding iron aerosols to the marine boundary layer.This intervention has the potential to decrease methane levels through enhanced molecular chlorine production.However, the potential trade-offs associated with such intervention are large (figure 5, table 2) as the added iron aerosol also increases the direct effect.We utilized a set of well-characterized CESM models to assess the efficacy and climate implications of such an approach.Spatial analyses revealed that adding iron aerosols in some regions such as the boxes in the North Atlantic and North Pacific are more likely to remove methane the most effectively while adding iron aerosols to a box in the Southern Ocean and Equatorial Pacific are likely to have very low methane removal efficiencies.
Note that the regional boxes may not be representative of the full ocean basins, as the iron loss shown in figure 1(d).
We show that the reactivity of chlorine towards methane and ozone influences the efficiency of methane removal in different regions, but the lifetime of the iron in the atmosphere is the most important control on efficiency.Although iron emissions contribute to decreasing methane and ozone concentrations, the overall impact on RF and surface temperatures by 2050 remains relatively modest (a reduction of 0.12 • C in the highest ship track scenario) across all scenarios, even though up to 200 Tg yr −1 of iron aerosols are added (60 times current combustion iron emissions).Part of this modest result is because the added iron aerosols absorb solar radiation, thus partially offsetting the benefits of removing methane and ozone.
Further research is needed on many topics (see supplement for a discussion of main uncertainties), especially the sensitivity of our result to the photoactive portion of iron that is added.The impact of the iron deposition on ocean fertilization is briefly discussed in the SI but is also an important outcome of such iron-based strategy.The stratospheric chlorine injection due to the different scenarios will also affect the stratospheric ozone and the ozone hole.This will be better quantified with a fully aerosolgas coupled model.In addition, there is likely to be a strong dependence of the efficiency on the details of the location and timing of the iron addition which needs more exploration.Overall, a better understanding of the positive and negative implications of adding iron aerosols and the resulting chlorine gas is vital to advance our understanding of effective iron-based strategies to mitigate atmospheric methane.
Our findings indicate that while there may be some localized enhancements in methane reduction due to very large iron additions (>60 Tg yr −1 of iron aerosols or at least 10-17 times global current combustion iron emission depending on the emission scenario), the scalability, global effectiveness, and environmental impacts of this method as a methane mitigation strategy remain uncertain.

Figure 1 .
Figure 1.(a) Methane removal efficiency calculated by iron emission from the OC scenario following equation (2).(b) Change in total methane loss between OC scenario and Base scenario.(c) Methane removal efficiency calculated by iron loss from the OC scenario following equation (2).(d) Change in iron loss between OC scenario and Base scenario.The black boxes show the box scenarios NP, NA, SO, and EP, described in table 1, and labeled in figure (a).Figures S3, S4, and S10 show the absolute methane loss, methane burden, and iron loss (deposition) for all the scenarios.

Figure 2 .
Figure 2. The relationship between the global methane burden after 10 years of intervention (Tg on vertical axis) and the additional iron emission (Tg/yr) for all the scenarios compared to Base scenario.The scenarios are the three ST scenarios (black diamond), the OC scenario (green circle), and the colored squares which represent different regions (shown in the insert maps): EP (cyan), NA (maroon), NP (yellow), SO (purple).The zero point is inserted to the plot on a linear scale and the dashed lines are to emphasize the initial increase in methane burden.

Figure 3 .
Figure3.The relationship between additional Cl2 production (vertical axis) and the additional iron emission (horizontal) for all the scenarios compared to Base scenario except for the scenarios with the nonlinear response, i.e. the low box scenarios (table1).Dotted lines represent slopes deduced from the different cases, where the upper lines (7:1 gCl2/gFe) come from the more efficient NP and NA regions, while the lowest dotted lines (3:1 gCl2/gFe) come from the less efficient SO region.The same scenarios and color scale are used as in figure2.FiguresS5 and S11show the Cl2 produced and iron emission for all the scenarios.
(a)).The global averaged CRS shifts from Cl + O 3 dominated in the Base scenario (figure 4(a)) to a regime where the reaction of chlorine with ozone and methane are comparable in the ST1000 (figure 4(c)), HighNP (figure 4(d)), and HighNA (figure S7) scenarios.In the cases where iron is added locally in a box the CRS shifts towards a regime where Cl + CH 4 are highly favored locally (figure

Figure 4 .
Figure 4. Chlorine reactivity towards ozone over reactivity towards methane (E3) in the troposphere for: (a) Base scenario, (b) OC scenario, (c) ST1000 scenario, and (d) HighNP scenario.Green colors indicate more Cl reactivity towards methane, and white colors indicate more Cl reactivity towards ozone.The black box represents the emission box in the NP scenario.Figure S7 shows the chlorine reactivity towards ozone over reactivity towards methane in the troposphere for all the scenarios.

Figure 5 .
Figure 5. (a), and (b) change in DRE compared to the base scenario using MAGICC in the year 2050 for iron (red), ozone (light blue), methane (dark blue), and the net radiative change (asterisk).(c) Change in spatial iron DRE compared to the base scenario using CESM in the highest additions of the shown scenarios.Note the difference in y-axis scale.

Table 1 .
Scenario name designation, description and amount of iron added (Tg/yr).

Table 2 .
Result summary showing the methane removed over iron added (equation (1)), global and annual mean iron addition, change in net RF over iron emission, Cl2 production over iron emission, methane removed over Cl2 added.
a Methane removed refers to the methane removed after 10 years, total iron added and total Cl2 produced refers to the total amount added/produced in 10 years.