A novel approach to sulfate geoengineering with surface emissions of carbonyl sulfide

Sulfate geoengineering (SG) methods based on lower stratospheric tropical injection of sulfur dioxide (SO2) have been widely discussed in recent years, focusing on the direct and indirect effects they would have on the climate system. Here a potential alternative method is discussed, where sulfur emissions are located at the surface in the form of carbonyl sulfide (COS) gas. A time-dependent chemistry-climate model experiment is designed from year 2021 to 2055, assuming a 40 Tg-S/yr artificial global flux of COS, geographically distributed following the present day anthropogenic COS surface emissions. The 5 budget of COS and sulfur species is discussed, as well as the effects of SG-COS on the stratospheric sulfate aerosol optical depth (∆τ=0.080 in years 2046-2055), aerosol effective radius (0.46 μm), surface SOx deposition (+8.7 %) and tropopause radiative forcing (RF) (-2.0 W/m for clear sky conditions and -1.5 W/m including the cloud adjustment). Indirect effects on ozone, methane and stratospheric water vapor are also considered, along with the COS direct contribution (with an overall gas phase global radiative forcing of +0.23 W/m). According to our model results, the resulting net RF of this SG-COS experiment 10 is -1.3 W/m for the year 2050, and it is comparable to the corresponding RF of -1.7 W/m obtained with a sustained injection of 4 Tg-S/yr in the tropical lower stratosphere in the form of SO2 (SG-SO2, able to produce a comparable increase of the sulfate aerosol optical depth). Significant changes of the stratospheric ozone response are found in SG-COS with respect to SG-SO2 (+4.9 DU versus +1.5 DU, globally). According to the model results, the resulting UVB perturbation at the surface accounts to -4.3% as a global-annual average (versus –2.4% in the SG-SO2 case), with a springtime Antarctic decrease of -2.7% (versus 15 a +5.8% increase in the SG-SO2 experiment). Overall, we find that an increase in COS surface emission may be feasible, and produce a more latitudinally-uniform forcing without the need for the deployment of stratospheric aircrafts.


COS increased emission fluxes in SG-COS
The annual upward flux is averaged over the period 2046-2055. 3 Results

Sulfate burden
COS is the most abundant sulfur-containing species in the atmosphere under quiescent conditions (i.e. not considering explosive volcanic eruptions). It is efficiently lost at the surface via dry deposition on soils and vegetation: taking this sink into account, the net global lifetime (atmospheric chemistry plus surface deposition) is approximately 4 years, depending on the assumed 95 magnitude of the soil and vegetation sink (Sandoval-Soto et al., 2005;Van Diest and Kesselmeier, 2008). In the troposphere the COS chemical reactivity (mostly with the hydroxyl radical) is rather slow: COS is thus well mixed and is easily transported in the stratosphere through the tropical tropopause layer (TTL). In the mid-stratosphere COS becomes efficiently photolyzed by solar UV radiation, becoming an important source for stratospheric SO 2 and finally for sulphuric acid aerosols.
When increasing the emission fluxes in SG-COS, it takes ∼ 15 years before the concentration reaches a new equilibrium, 100 from 0.5 to 35.5 ppbv (Fig. 2a). In the same timespan, the global AOD increases, reaching a value of 0.08 by 2035, similar to the global value that is reached by the direct injection of SO 2 in the equatorial stratosphere in SG-SO2; in that case, however,  In the SG-COS experiment, surface COS fluxes are adjusted so as to have the same global aerosol optical depth (AOD) ≈ 0.08 (see table 1). This is done in order to more easily compare the latitudinal distribution of the aerosols, and to better quantify the differences in the radiative forcing from both direct and indirect (ozone, methane and water vapor) changes in atmospheric 115 composition.
There is a large difference in the latitudinal distribution of stratospheric sulfate optical depth, as shown in figure 3 (a). SG-COS produces an AOD more uniformly distributed over all latitudes with respect to the SG-SO2 case, where the increase of optical depth is most prominent in the tropics; this is due to the efficient tropospheric mixing of COS before it reaches the stratosphere.

120
The differences in the latitudinal distribution of AOD are also observable in the differences in the particle sizes and in the surface area density (SAD). Figure 3 (b) shows that the stratospheric effective radius is smaller in the SG-COS experiment and uniform for all latitudes, with a global value of 0.46 µm. In SG-SO2, the effective radius is higher in the tropics (0.63 µm); AOD is also larger in the tropics in that case, due to a larger concentration of particles there, even if larger particles are less effective at scattering incoming solar radiation (English et al., 2012).
125 Figure 4 shows a comparison of the effective radius (a) and SAD (b) between the BG, SG-COS and SG-SO2 cases, separating the tropics, mid-latitudes and polar regions. As SO 2 is injected at the equator, all oxidation and nucleation happens in the tropics in SG-SO2. This is reflected in the vertical distribution which has a maximum in the lowermost stratosphere. On the other hand in SG-COS, the effective radius increase is reached at higher altitudes, between 18-30 km, which is consistent with COS reaching higher altitudes through deep tropical convection before it is photochemically destroyed (Barkley et al., 2008). The As the size of the particles is determined by nucleation in the tropical region, where SO 2 oxidation occurs, mid-latitude and polar behaviour of the aerosols depends on the polewards transport by the Brewer-Dobson Circulation (BDC).
In SG-SO2, aerosols grow rapidly in the tropical region due to the high concentration of SO 2 , and their larger size affects sedimentation rates, thus decreasing their lifetime. Consequently, the amount of aerosols transported to higher latitudes is 135 lower; in SG-COS, smaller particles with a higher lifetime are either easily transported towards the poles or directly formed there. Smaller particles at a higher concentration, and larger particles at a lower concentration may then result in a SAD which looks similar at mid-latitudes and polar region, but for different reasons.
The vertical distribution of particles and their optical properties are shown in Figure 5  As discussed before, the formation of larger particles in SG-SO2 in the tropical region reduces the amount of aerosol transported to the poles compared to the SG-COS case, where a larger number of smaller particles produces a positive change in SAD and, consequently, in extinction.    Table 3. Globally-annually averaged wet deposition rates of sulfur species (Tg-S/yr) [years [2046][2047][2048][2049][2050][2051][2052][2053][2054][2055]]. The last column shows the net balance of total sulfur sources and sinks (Tg-S/yr).

Deposition
The enhancement of the stratospheric sulfate burden would produce an increase in sulfur deposition, in dry form through acid gas deposition and in wet form through rain, fog and aerosol particles. Here we analyse how dry and wet deposition of sulfur species are distributed globally as a result of the two SG interventions. Table 2 and 3 summarize wet and dry deposition rates for the SG-COS and BG experiments, respectively, and they include the contribution of each species to the total deposition. In particular, in SG-COS the increase in COS fluxes produces both an increase in sulfuric deposition, after its photolysis and oxidation to sulfuric acid, and in dry deposition of COS itself, as it is 160 removed to the ground through uptake by vegetation and soils (Kettle et al., 2002). Asia (Kettle et al., 2002). Dry deposition of COS doesn't contribute to acid deposition and, currently, there is no information available on how different soils or ecosystems would be affected by higher local COS concentrations; therefore, we assumed that their uptake efficiency does not change. The robustness of this assumption will need to be studied.
The global distribution of SO x deposition is also shown in figure 6. Panels (c) and (d) show dry and wet deposition, respec-170 tively, for the background case. Dry deposition maxima are localized in urban areas close to the source where the emitted sulfur dioxide is immediately oxidized, while wet deposition distribution depends both on sulfate concentration and precipitation.
Panels (e) and (f) show the total SO x deposition change in SG-COS with respect to the baseline case, in absolute terms and as a percentage of the baseline case, and most of its increase is due to wet deposition (see tables 2 and 3, and see Tables S1-4 for a breakdown of global sources and sinks of sulfur species). In both figures, the distribution of deposition is more uniform 175 over the globe with respect to the tropical injection of SO 2 , except for the polar regions, because of the reduced precipitation rates. Consequently, figure 6 (f) shows a large increase in percent deposition in the polar region (17% in the Arctic, 8% in Antarctic) because of very low values in the baseline case. On the other hand, deposition change is close to zero in polluted regions.
Globally, the annual differences in deposition fluxes for all species compared to the background case amount to 8.3 ± 0.2 180 Tg-S/yr for SG-COS and 3.9 ± 0.2 Tg-S/yr for SG-SO2, which equates to an increase of 8.9 ± 0.3 % and 4.2 ± 0.3 %, respectively.   UVA 11.43±0.01 11.20±0.01 11.25±0.01 -2.0±0.1 -1.6±0.1 The simulated enhancement in the stratospheric aerosol layer would produce two main effects: an increased scattering of solar radiation, that in turn would reduce surface temperatures, and the local absorption of more near-infrared solar and terrestrial 185 radiation, that would warm the stratospheric layer where the aerosols reside (as observed for volcanic eruptions, see Lacis et al., 1992;Labitzke and McCormick, 1992). Furthermore, the increase in the surface area density of the aerosols would affect the heterogeneous chemistry of ClO x and NO x with implications for ozone concentration and UV radiation at the surface (Tilmes et al., 2009(Tilmes et al., , 2018b(Tilmes et al., , 2021. For SO 2 , it has been shown that the combination of surface cooling, perturbation of stratospheric temperatures and changes 190 in tropospheric ozone and in UV at the surface also affects methane lifetime (Visioni et al., 2017a). In this section we analyse the differences in these changes also for the SG-COS experiment. column, mostly due to a reduction in tropospheric ozone, as visible in panels (c) and (d), as a direct consequence of the surface cooling (Nowack et al., 2016). On the other hand, at higher latitudes an overall increase in the total column is observable due to an increase in stratospheric ozone. This is particularly evident closer to the poles.
During springtime months, there is some Antarctic ozone depletion, while in the Arctic a recovery of ozone is observable.
In the Antarctic spring, the polar vortex is strengthened by the stratospheric heating in the tropics that affects the equator-topole thermal wind balance (Visioni et al., 2020), resulting in greater confinement of cold air, that, in turn, enhances the ozone where the cycles of chlorine (ClO x ) and bromine (BrO x ) are dominant, there is an increase in ozone loss since reduction of 220 NO x , that normally bounds chlorine (ClONO 2 ), allows ClO to destroy more ozone (Tilmes et al., 2018b;Grant et al., 1992).
At low latitudes, stratospheric ozone concentration is also driven by change of tropical upwelling : the reduction of tropical upwelling of ozone-poor air coming from the lowermost stratosphere leads to higher ozone concentration at altitudes of about 20-22 km (Tilmes et al., 2018b). In SG-SO2, the highest concentration of absorbing aerosols leads to positive w * above 20 km due to the local warming but this doesn't affect the transport of ozone-poor air from the lower layers.
Above the discussed altitudes, there is a net ozone production in both SG experiments, with an higher increase of ozone mixing ration in SG-COS experiment with respect to SG-SO2, especially in the extra-tropical region. Ozone depletion at these 230 altitudes is mainly controlled by the catalytic cycle of NO x , that is inhibited by the denitrification process due to heterogeneous reactions on aerosols.
Globally, the annually-averaged ozone column increases of 4.9 and 1.5 DU for SG-COS and SG-SO2, respectively (table 4).
The stratospheric ozone increase affects UVB at the surface due to its photodissociation, while aerosol could affect UVA radiation by scattering processes: the projected changes are shown in figure 8 for both UVA and UVB. We estimated these 235 changes using TUV (from https://www2.acom.ucar.edu/modeling/tropospheric-ultraviolet-and-visible-tuv-radiation-model), using in input our model latitudinal and monthly values for the period (2046-2055) for aerosol optical depth, total ozone column, climatological cloud cover and surface albedo.
In both SG experiments, the negative changes of UVB radiation at surface, except in the Antarctic region, are related to the variation in stratospheric ozone. as well as the interannual variation that increase towards the poles, due to the seasonal variation 240 of ozone, as discussed before. UVA decrease is everywhere negative in both SG experiments. In particular, the correlation between UVA change and particles scattering is evident if we compare this latitudinal distribution with the stratospheric AOD of figure 3(a). The globally averaged UVB and UVA changes at surface are summarized in table 5.
Methane is an indirect source of tropospheric ozone (West and Fiore, 2005), and it is also a greenhouse gas. Knowing its variation is fundamental to understand the final contribution to the radiative forcing that one would wish to achieve with this 245 geoengineering method. From table 4, we measure a global increase in methane lifetime of 12.7% in SG-COS and 12.2% in SG-SO2, which we can identify in the increase in methane itself. The reason for the increase in methane is to be found in the behaviour of the hydroxyl radical (OH), as the main sink of methane is the oxidation reaction with OH: decrease of OH means an increase of methane lifetime. As discussed by Visioni et al. (2017b), mechanisms that cause an increase in OH are as follows: (a) surface cooling lessens the amount of tropospheric water vapor and inhibits the temperature-dependent reaction   The ULAQ-CCM radiative transfer module calculates online the radiative forcing due to aerosols, greenhouse gases (GHGs), In SG-COS, obviously, the increase of COS concentration, which is a GHG, must be taken into account. We estimated its contribution to RF based on the definition of global warming potential (GWP) on a mass/mass basis as in Bruhl et al. (2012) for a time horizons of 30 years . GWP can be approximated as follows by the expression of Roehl et al. (1995), 270 assuming that the perturbation of the radiation balance of the Earth by greenhouse gases COS and CO 2 decays exponentially after a pulse emission for a time horizon ∆T.
We assumed an overall lifetime of τ COS =3.8 yr and τ CO2 =75 yr, and the radiative forcing of 1 kg of COS relative to 1 kg of CO 2 added to the present atmosphere (RF COS /RF CO2 ) is 724 (Brühl and Crutzen, 1988). This results in a GWP of 111. For 275 our time period, the mass of COS and CO 2 added to the atmosphere (∆m) is 1.97 10 12 kg of COS and 1.23 10 15 kg of CO 2 .
Therefore, the COS radiative forcing can be calculated as: where RF CO2 in RCP6.0 is estimated to be 0.83 W/m 2 considering an increase of 68.5 ppm from a baseline of 409.2 ppm.
Overall, this results in a radiative forcing from the COS increase of 0.17 W/m 2 .

280
The main contributions of sulfate aerosols and clouds are summarised in tables S5 and S6 for SG-COS and SG-SO2, respectively. Globally, the estimated values are similar for the Clear-Sky SW and LW forcing from the sulfate aerosols: in terms of the latitudinal distribution, however, SG-SO2 presents a peak in the tropics whereas the forcing from SG-COS is much more latitudinally even.
The reduction in optical depth from cirrus clouds (see table 1  We briefly discuss here the technical feasibility of the approach described in this paper, as related to the increase of surface COS emissions (for SO 2 injections, see for instance Smith and Wagner, 2018  and may yield over 90% CO 2 (Wang et al., 2013). CS 2 is produced via numerous means, perhaps the easiest being from coke (carbon) and molten sulfur: Approximately 1 million tons of CS 2 is produced per year (Madon and Strickland-Constable, 1958), with China consuming approximately half of the global production of CS 2 for rayon manufacturing. CS 2 is highly unstable and is flammable in air. It is also toxic at low concentrations (10 ppm).

305
Given the reactions above, about 0.5 Tg of S will produce 0.94 Tg of COS: this amounts to 0.16 Tg of C (coke) and 0.55 Tg of molten sulfur. In the last decade, approximately 70 Tg of sulfur were produced worldwide, so this would constitute an increase in S production of 0.8%. The price varied between $50 and $200 per ton, leading to an annual cost of approximately $25-100 million. Worldwide production of coke was around 640 Tg, so this increase in production is negligible. The price of coke varies between $50 and $100 per ton, leading to an annual cost of approximately $8-16 million.

310
To this we would have to add the cost of CO 2 , as well as the production and energy costs. Considering an estimate of $400 million per year for each Tg of S between CO 2 and production and energy cost, and assuming an effort shared between 1000 locations, this would add up to $400.000 per location per year per each Tg of S. The overall cost is roughly of the same order of magnitude as that in Smith and Wagner (2018) for a stratospheric aerosol deployment at ∼ 20 km of injection (so different from the injection set-up in our study for SG-SO2), but without the need to develop a new aircraft-based delivery system.

Conclusions
We have presented here the results of a modeling experiment with the aim of producing an optically thick cloud of sulfate aerosols in the stratosphere without the injection of sulfate precursors directly in the stratosphere, but rather using increased surface emissions of carbonyl sulfide (COS). The low reactivity of COS in the troposphere, where it is not reactive and where it is predominantly absorbed by some soils and by plants, allows for a large portion of its emissions to reach the stratosphere, 320 where it is turned into sulfate aerosols by photo-dissociation and oxidation.
We compare the results obtained with an increased emission of 40 Tg-S/yr (roughly 400 times more than the background emissions) with those from a 4 Tg-S/yr injection of SO 2 in the stratosphere as prescribed in previous experiments (Kravitz 20 https://doi.org/10.5194/acp-2021-813 Preprint. Discussion started: 11 October 2021 c Author(s) 2021. CC BY 4.0 License. et al., 2011;Visioni et al., 2017a). Both experiments result in a similar global optical depth from the produced stratospheric aerosols (∼ 0.07), but with different latitudinal distributions: for SO 2 , as previously observed in various modeling experiments, 325 equatorial injections result in an increased concentration of aerosols in the tropical stratosphere that tends to overcool the tropics and undercool the high latitudes (Kravitz et al., 2018;Jiang et al., 2019), while also reducing the efficacy of the back-scattering from the aerosols due to the increased size of the particles (Visioni et al., 2018c). On the other hand, with COS emissions the uniform mixing of the gas allows for a more uniform distribution of the produced aerosols in the stratosphere, resulting in increased optical depth also at very high latitude.

330
The differences in distribution and size of the particles result in different changes to the composition of the atmosphere: smaller particles absorb and heat the stratosphere less, thus resulting in fewer dynamical changes. From a chemical perspective, stratospheric ozone would be impacted differently from the two geoengineering schemes. For SO 2 injections, previous studies have shown that the overall effect is the result of a combination of various dynamical and chemical factors that behave differently depending on the latitude and altitude of the aerosols.

335
At low latitudes the increase in lower stratospheric water vapor produced by the warming of the tropopause layer enhances the halogen-driven destruction of ozone in the lower stratosphere (Tilmes et al., 2018b) due to NO x depletion. This effect is balanced by reduced ozone destruction in the middle stratosphere due to the slowing down of the NO x cycle produced by enhanced heterogeneous chemistry (Pitari et al., 2014;Richter et al., 2017;Franke et al., 2020).
Overall, in the case of COS emissions the further increase in SAD produced by smaller particles increases the inhibition of 340 the ozone cycles in the middle stratosphere, resulting in a net increase in stratospheric ozone and thus in a larger decrease of UV radiation at the surface. Similarly, the larger sulfate burden at high latitudes produces further ozone recovery and thus less UV radiation also at the poles for the COS case.
Our results point to the feasibility of surface emissions of COS as a possible substitute to stratospheric SO 2 (or other sulfate precursors) injections to produce stratospheric sulfate aerosols. This would sidestep the problem of deploying methods not 345 already available to bring the sulfate at those altitudes, including development of novel aircraft (Bingaman et al., 2020).
Since COS is already a byproduct of human activities, it might be possible to devise methods of mass-production of the required quantities that may be cheaper than the known proposed methods .
Overall, there are some weak points in a geoengineering strategy using COS compared to SO 2 that need to be addressed.
First, it necessitates a larger amount of emissions to achieve the same global stratospheric AOD, resulting in larger amounts of 350 deposition. It would be less easily scalable, and both deployment and phase-out, as we have shown, would require a much longer time-frame compared to the almost instantaneous effect produced by SO 2 injections. Considering the dangers to ecosystems presented by a too fast deployment or termination of sulfate geoengineering (Trisos et al., 2018), this might not actually be a large drawback, but it does remove the possibility of rapidly "regulating" the necessary amount of stratospheric sulfate in case of changes in strategy or external conditions (such as a Pinatubo-like volcanic eruption; Laakso et al., 2016). Furthermore, the 355 mixing happening in the troposphere would not allow any control in the latitudinal or seasonal distribution of the resulting aerosols, as proposed elsewhere Dai et al., 2018;Visioni et al., 2019).