Exploring dimethyl sulfide (DMS) oxidation and implications for global aerosol radiative forcing

Abstract. Aerosol indirect radiative forcing (IRF), which characterizes how aerosols alter cloud formation and properties, is very sensitive to the preindustrial (PI) aerosol burden. Dimethyl sulfide (DMS), emitted from the ocean, is a dominant natural precursor of non-sea-salt sulfate in the PI and pristine present-day (PD) atmospheres. Here we revisit the atmospheric oxidation chemistry of DMS, particularly under pristine conditions, and its impact on aerosol IRF. Based on previous laboratory studies, we expand the simplified DMS oxidation scheme used in the Community Atmospheric Model version 6 with chemistry (CAM6-chem) to capture the OH-addition pathway and the H-abstraction pathway and the associated isomerization branch. These additional oxidation channels of DMS produce several stable intermediate compounds, e.g., methanesulfonic acid (MSA) and hydroperoxymethyl thioformate (HPMTF), delay the formation of sulfate, and,
hence, alter the spatial distribution of sulfate aerosol and radiative
impacts. The expanded scheme improves the agreement between modeled and
observed concentrations of DMS, MSA, HPMTF, and sulfate over most marine
regions, based on the NASA Atmospheric Tomography (ATom), the Aerosol and
Cloud Experiments in the Eastern North Atlantic (ACE-ENA), and the
Variability of the American Monsoon Systems (VAMOS)
Ocean-Cloud-Atmosphere-Land Study Regional Experiment (VOCALS-REx)
measurements. We find that the global HPMTF burden and the burden of sulfate produced from DMS oxidation are relatively insensitive to the
assumed isomerization rate, but the burden of HPMTF is very sensitive to a
potential additional cloud loss. We find that global sulfate burden under PI and PD emissions increase to 412 Gg S (+29 %) and 582 Gg S (+8.8 %), respectively, compared to the standard simplified DMS oxidation scheme. The resulting annual mean global PD direct radiative effect of DMS-derived sulfate alone is −0.11 W m−2. The enhanced PI sulfate produced via the gas-phase chemistry updates alone dampens the aerosol IRF as anticipated (−2.2 W m−2 in standard versus −1.7 W m−2, with updated gas-phase chemistry). However, high clouds in the tropics and low clouds in the Southern Ocean appear particularly sensitive to the additional aqueous-phase pathways, counteracting this change (−2.3 W m−2). This study confirms
the sensitivity of aerosol IRF to the PI aerosol loading and the
need to better understand the processes controlling aerosol formation in the PI atmosphere and the cloud response to these changes.



Cloud-borne Aerosols
In-cloud aqueous-phase reactions are handled separately in CAM6-chem Rasch et al., 2000). After the chemical production and loss in the gas-phase reactions and the aqueousphase within the interstitial aerosols, the amounts of SO2, MSA, and H2SO4 collected by cloud are computed based on their concentrations, effective Henry's Law constants, and cloud fraction in each 30 grid cell. For each cloudy grid cell, the gas-aqueous equilibrium is controlled by the equilibrium pH of local bulk cloud water. Such equilibrium pH is evaluated by iteratively solving an electro-neutrality equation that balances the charges of all dissolved species, including H + , NH4 + , Na + , Ca 2+ , OH -, Cl -, NO3 -, HCO3 -, HSO3 -, SO3 2-, and SO4 2-. Based on this equilibrium pH, a portion of available SO2 is dissolved in cloud water and oxidized by H2O2 and O3, forming cloud-borne sulfate aerosol. Another 35 source of cloud-borne sulfate is cloud uptake of MSA and H2SO4. Gains in sulfate mass from these incloud processes are contributed to each mode in proportion to its relative aerosol number abundance, and it is assumed that no new particles are formed.

Aerosol Formation and Growth 40
Aerosol formation and growth in CAM6-chem are treated by MAM4 (Liu et al., 2016), we do not modify these schemes in our work, but provide a short description of relevant processes here. New particle formation initiated by clusters of H2SO4 vapor is modeled through three regimebased parameterizations (He and Zhang, 2014), namely, 1) H2SO4-NH3-H2O ternary nucleation when NH3 > 0.1 ppt (Merikanto et al., 2007); 2) H2SO4-H2O binary nucleation when NH3 is low or absent 45 (Vehkamäki, 2002), and; 3) empirical formulas depending solely on the concentration of H2SO4 to reduce the model low-bias of boundary layer nucleation rate (Wang and Penner, 2009). All nucleation processes are assumed to occur in the vapor phase homogeneously. Loss due to coagulation as new particles grow from critical clusters to Aitken mode is accounted for following the treatment in Kerminen and Kulmala (2002). The remaining new particles are added to both number and mass of 50 Aitken mode sulfate and ammonium aerosols.
Aerosol uptake of NH3 ceases when the local mass ratio of NH4 + /SO4 2reaches two. This dynamic process contributes to aerosol size growth. Aerosols in Aitken and accumulation modes that grow larger than the model-defined size-bin boundary will trigger reassignment of number and mass from their 55 original modes to accumulation and coarse modes correspondingly.
Coagulation rates of Aitken and accumulation aerosols are calculated using the fast/approximate algorithms of community Multiscale Air Quality (CMAQ) model 4.6 while the much slower coagulation processes involving the coarse mode aerosol are neglected (Binkowski and Roselle, 2003).
Intramodal coagulation results in a reduction in particle number in corresponding modes but their totals 60 remain unchanged. Intermodal coagulation reduces the number of Aitken mode aerosols and transfers masses from the Aitken mode to the accumulation mode.
Water uptake by aerosol is handled by MOSAIC (Zaveri et al., 2008(Zaveri et al., , 2021. The bulk hygroscopicity of each mode is determined as the volume-weighted mean hygroscopicity of all components in that mode. Values of the hygroscopicity for sulfate (0.507) and other aerosols are from 65 Petters and Kreidenweis (2007) and various references detailed in Liu et al. (2012). Due to its high hygroscopicity, higher sulfate concentration results in enhanced water uptake capability of airborne aerosols.
Interstitial aerosols are activated and become cloud-borne under sub-grid vertical mixing with the presence of cloud. The activation is parameterized as a function of updraft velocity and the averaged 70 properties of all aerosol modes (Abdul-Razzak and Ghan, 2000). During activation, the model transfers both mass and number of aerosols from the interstitial attachment state to the cloud-borne state. This process can trigger new cloud formation or increase the aerosol loading of existing cloud. CAM6-chem assumes a three-hour in-cloud residence time for air parcels (Lelieveld and Crutzen, 1990), i.e., 1/6 of the cloud-borne aerosol is dissipated and re-suspended as air-borne particles after each model timestep 75 of 30 minutes.

Dry & Wet Deposition
Gravity-and turbulence-driven dry deposition processes are considered in CAM6-chem. All species above the surface layer are subject to gravitational settling at velocities determined following 80 Seinfeld et al., (1998). Turbulent dry deposition is handled based on the resistance approach with phasedependent adjustments (Emmons et al., 2010;Wesely, 1989). For gas-phase species, the model calculates the aerodynamic and the boundary resistances based on online atmosphere dynamics, while the surface resistance over land is determined according to online CLM5 surface variables, e.g., canopy height and leaf area index (LAI), as well as species-dependent reactivity factor for oxidation and 85 effective Henry's Law constants. For aerosols, the aerodynamic resistance is the same as that of gases, but the boundary and surface resistances are replaced by a single resistance term which depends on the surface friction velocity. Deposition velocities are evaluated as the reciprocal of the sum of their corresponding resistance terms, and deposition rates are the product of their deposition velocities and concentrations. 90 Wet deposition of gaseous chemicals in CAM6-chem is handled by the Neu and Prather (2012) scheme which assumes a first-order loss due to in-cloud and below-cloud scavenging processes. The wet removal rate of a depositing gas is the product of its concentration, loss frequency (depending on its effective Henry's Law coefficients), and the fraction of the grid box that is undergoing scavenging events, such as the presence of cloud or precipitation. Aerosol wet removal is handled using a separate 95 routine Liu et al., 2012). In-cloud aerosol removal rates depend on the mass mixing ratio of activated aerosols and precipitation rates. Below-cloud scavenging of interstitial aerosols is assumed a first-order removal process and the removal rate is determined by the product of scavenging coefficient and precipitation rate. 100

Cloud
CAM6-chem simulates the fate of cloud liquid drops/ice crystals using the two-moment microphysics (MG2) for prognostic evolution of mass and number mixing ratios of cloud (Gettelman and Morrison, 2015). This scheme is coupled to MAM4 for droplet activation (Abdul-Razzak and Ghan, 2000) and ice nucleation (Liu et al., 2007). Briefly, the fraction of activation is controlled by critical 105 superstation of particles, which depends on, e.g., bulk hygroscopicity and size distribution of aerosols.
Occurrence of stratiform, shallow convective, and deep convective cloud are predicted by the Cloud Layers Unified by Binormals (CLUBB) scheme (Bogenschutz et al., 2012(Bogenschutz et al., , 2018Golaz and Larson, 2002). This scheme also enables the simulation of prognostic precipitation, including the autoconversion of CCN and accretion by rain or snow (Gettelman, 2015;Gettelman and Morrison, 110 2015).

Radiative Transfer
Longwave (LW) and shortwave (SW) radiative transfer within CAM6-Chem are represented by the broadband k-distribution Rapid Radiative Transfer Model for Global Circulation Models (RRTMG) 115 (Iacono et al., 2008) with sixteen LW and fourteen SW spectral intervals. RRTMG is coupled with the aerosol and cloud schemes in CAM6-chem for spatial distribution and optical properties of the condensed-phase particles, drops, and ice crystals. Absorption by aerosols and clouds is included for LW while extinction by aerosols and cloud drops is considered for SW. Mass-specific aerosol optical properties are parameterized as functions of wet refractive index and mode wet surface radius for each 6 band of SW (extinction, single-scattering albedo, and asymmetry parameter) and LW (mass-specific absorption) (Ghan and Zaveri, 2007). Sulfate refractive indices at visible wavelengths are 1.43 + 0.00i (Hess et al., 1998). Liquid cloud optics is parameterized following the gamma size distribution implemented by Morrison and Gettelman (2008). Ice cloud optics is also determined using a similar look-up approach produced by the modified anomalous diffraction approximation (MADA) (Mitchell 125 et al., 2006). Sub-grid variabilities of cloudiness and cloud overlapping are handled using the Monte-Carlo Independent Column Approximation (McICA) (Pincus et al., 2003).

Uncertainty in DMS emission
We employed the OASISS flux scheme and calculated the DMS ocean-to-air fluxes based on sea-surface DMS concentration climatology generated by Lana et al. (2011). Their climatology map 155 was based on a series of site measurements and extrapolation of the data for months when data is unavailable. Data are scarce in remote regions such as the Southern Ocean, resulting in larger uncertainty in surface DMS concentration. Such uncertainty propagates to our DMS flux estimates, particularly over the Southern. We hence performed a sensitivity test on [MOD_2000] by reducing the sea surface DMS concentration in regions south to 30ºS by 50% (aliased as [MOD_lessSDMS_2000]). 160 This test resulted in lowering the global-total DMS emission to 19 Gg-S yr -1 . It also led to decreases in the DMS mixing ratios in the lower troposphere (<5 km) by ~60% from [MOD_2000], further reducing the model-ATom difference (Figure S4). It is noteworthy that the parameterization methods of the seaand air-side resistance as well as meteorological variabilities, e.g., near-surface air temperature and wind speed, may also contribute to the uncertainties in DMS flux estimation. 165

Sensitivity tests on the production and loss processes of HPMTF
We explore the impact of using a different kiso (isomerization rate of MSP) determined by our 175 recent laboratory experiment and two cloud uptake rates (kHPMTF+cloud) on reducing the modelobservation deviations. All simulations for the sensitivity tests are performed with our modified chemistry and PD-level emissions, i.e., based on [MOD_2000]. Table S2 summarizes the model settings and key results of these tests.
180 Table S2. Summary of the results in the sensitivity tests on the production and loss processes of HPMTF.  Introducing MSA into the model via our DMS chemistry imposes a substantial impact on radiative balance over the high latitudes. Since the standard CAM6-chem does not include MSA when calculating radiative transfer, we assume that the radiative properties of MSA are identical to sulfate.
Then, we perform an extra simulation for each of [MOD_1850] and [MOD_2000] with rapid MSA-tosulfate conversion to capture the joint radiative effect of MSA and sulfate. We denote them as 220 burden (though lower than the combined burden of sulfate and MSA) but reduces its lifetime. Thus, the results represent a lower bound of the radiative effects and forcing of sulfate and MSA.
The PD MSA and sulfate burdens are 7.5 Gg-S and 582 Gg-S, respectively, from [MOD_2000] while the new sulfate burden is 588 Gg-S in [MOD_RE_2000]. The slightly lower total sulfur aerosol 225 burden is likely due to faster deposition rates of sulfate, but sulfate lifetime maintains at 4.4 days as [MOD_2000]. Similarly, the rapid conversion also induces an increase in PI sulfate burden from 411 Tg-S in [MOD_1850] to 418 Tg-S in [MOD_RE_1850]. Hence, the PD-PI sulfate burden difference of [MOD] is +170 Tg-S (similar to [MOD_RE]) and weaker IRF (+0.76% with respect to [MOD_RE]).
230 Figure S10. Sensitivity of PI-to-PD percentage changes of cloud condensation nuclei (CCN), liquid water path (LWP), ice water path (IWP), cloud droplet number concentration (CDNC), cloud droplet radius (rd), frequency of precipitation (fprep), cloud coverage (fcloud), SW CRE, and LW CRE against 235 each unit of increase in sulfate burden, on five spatial scales: global, 60ºS to 60ºN, marine only, marine with low clouds, and marine with high clouds. CCN is estimated using global/regional total value while other variables are global/regional means. 240 Figure S11. Contrasting the zonal-means of difference in PI-to-PD changes of (a) sulfate column concentration, (b) liquid water path (LWP), (c) ice water path (IWP), and (d) SW DCRE, overlapping with marine low and high clouds, modeled by simulations with different chemistry settings.