Oceanic emissions of dimethyl sulfide and methanethiol and their contribution to sulfur dioxide production in the marine atmosphere

. Oceanic emissions of dimethyl sulfide (CH 3 SCH 3 , DMS) have long been recognized to impact aerosol particle 10 composition and size, the concentration of cloud condensation nuclei (CCN), and Earth’s radiation balance. The impact of oceanic emissions of methanethiol (CH 3 SH, MeSH), which is produced by the same oceanic precursor as DMS, on the volatile sulfur budget of the marine atmosphere is largely unconstrained. Here we present direct flux measurements of MeSH oceanic emissions using the eddy covariance (EC) method with a high-resolution proton-transfer-reaction time-of-flight mass spectrometer (PTR-ToFMS) detector and compare them to simultaneous flux measurements of DMS emissions from a coastal 15 ocean site. Campaign mean mixing ratios of DMS and MeSH were 72 ppt (28-90 ppt interquartile range) and 19.1 ppt (7.6-24.5 ppt interquartile range) respectively. Campaign mean (and interquartile range) emission fluxes of DMS ( F DMS ) and MeSH ( F MeSH ) were 1.13 (0.53-1.61) and 0.21 (0.10-0.31) ppt m s -1 respectively. Linear least squares regression of observed MeSH and DMS flux indicates the emissions are highly correlated with each other (R 2 = 0.65) over the course of the campaign, consistent with a shared oceanic source. The campaign mean DMS to MeSH flux ratio ( F DMS : F MeSH ) was 5.5 ± 3.0 calculated 20 from the ratio of 304 individual coincident measurements of F DMS and F MeSH . Measured F DMS : F MeSH was weakly correlated (R 2 = 0.15) with ocean chlorophyll concentrations, with F DMS : F MeSH reaching a maximum of 10.8 ± 4.4 during a phytoplankton bloom period. No other volatile sulfur compounds were observed by PTR-ToFMS to have a resolvable emission flux above their flux limit of detection or to have a gas phase mixing ratio consistently above their limit of detection during the study period, suggesting DMS and MeSH are the dominant volatile organic sulfur compounds emitted from the ocean at this site. 25 The impact of this MeSH emission source on atmospheric budgets of sulfur dioxide (SO 2 ) was evaluated by implementing observed emissions into a coupled ocean-atmosphere chemical box model using a newly compiled MeSH oxidation mechanism. Model results suggest that MeSH emissions lead to afternoon instantaneous SO 2 production of 2.5 ppt hr -1 , which accounts for 30% of the instantaneous SO 2 production in the marine boundary layer at the mean measured F DMS and F MeSH . This contribution of MeSH to SO 2 production is driven by a higher effective yield of SO 2 from MeSH oxidation and the shorter 30 oxidation lifetime of MeSH compared to DMS. This large additional source of marine SO 2 has not been previously considered in global models of marine sulfur cycling. The field measurements and modeling results presented here demonstrate that MeSH Setser, 2021). Recent ab initio/RKKM calculations determined the CH 3 S· yield from MeSH + OH at 298 K and 760 torr to be 0.98 (Mai et al., 2020). 115 Given the slight inconsistency between the directly measured CH 3 S· yield of 1.1 ± 0.2 and the MeSH + OH branching fractions, we take these experiments to provide an upper and lower bound on the CH 3 S· yield of 0.87 and 1.1, and assume a CH 3 S· yield of 1 for R1 throughout this analysis. of and F correlated (R (mean F MeSH = 0.21 ppt m s -1 , compared to mean F DMS = 1.13 ppt m s -1 ). The average ratio of individual DMS to MeSH flux DMS : F MeSH ) measurements for the full campaign was 5.5 ± 3.0. We assess the impact of the observed large MeSH emission flux on production of SO 2 in the marine atmosphere through a coupled ocean flux -atmospheric chemistry 0-D box model with 145 a newly compiled MeSH oxidation mechanism. Modeling results suggest that MeSH contributes approximately 30% of instantaneous afternoon SO 2 production ( P SO2 ). Together these results show that MeSH emissions are an important contributor to sulfur budgets in the marine atmosphere and further field studies and laboratory oxidation mechanism investigations are warranted. MeSH mixing ratios was also tested. Measured emission fluxes of MeSH and DMS are coupled into the model to set the source term for those molecules. Meteorological inputs were acquired from the NOAA National Data Buoy Center (Stations LJPC1 and LJAC1) as well as from an onsite temperature and relative humidity data logger (OM-62, Omega Engineering). Chemical constraints included coincident measurements of O 3 and nitryl chloride (ClNO 2 ) measured at the same site in August 2018 245 (Vermeuel et al., 2020), an assumed OH profile that followed the solar cycle with a peak at 4.0 x 10 6 molecules cm -3 , and constant concentrations of other major trace gases as listed in Table S2 . A constant first-order dilution loss term was used with a one-day lifetime to approximate mixing out of the boundary layer. A static boundary layer height (BLH) of 500 m was assumed (Faloona et al., 2005; Stull, 1988; Wei et al., 2018). Clear sky conditions were also assumed (i.e. no heterogeneous loss from reactive uptake on cloud droplets (Vermeuel et al., 2020)). An updated oxidation mechanism for DMS and MeSH 250 was implemented, expanding upon the default DMS oxidation scheme in the MCM v3.3.1 to include oxidation of MeSH to form CH 3 S· (R1) and to include HPMTF chemistry, detailed in Table S1 . The model was allowed to spin up for two days to allow reactive intermediates to reach equilibrium, with all reported values taken from day three of the model run. The slope of the correlation was 0.19 with a linear least squares regression coefficient (R 2 ) of 295 0.61. The ratio of DMS:MeSH mixing ratios reaches a minimum near hours 05-07 following the buildup of MeSH overnight. Both DMS and MeSH were observed to reach a maximum in concentration at night, and minimum concentrations in the early afternoon as shown in the diel profiles in Fig. 3 . The observed diel profile is consistent with expectations due to the significantly lower oxidative loss rate at night and has been observed in other studies (Lawson et al., 2020). MeSH varies by approximately a factor of five between its diel average maximum and minimum concentration, compared to DMS which varies by 300 approximately less than a factor of three. The larger diel variability in MeSH is due to its approximately five times faster bimolecular rate constant with OH ( k OH+MeSH = 3.3 × 10 -11 cm 3 molecule -1 s -1 at 293 K) compared to DMS ( k OH+DMS = 7.8 × 10 12 cm 3 molecule -1 s -1 at 293 K) resulting in a lifetime of MeSH to oxidation by OH during the afternoon on the order of 3 hours compared to 16 hours for DMS (for [OH] = 2 × 10 6 molecule cm -3 ). In the remote south-west Pacific Ocean, Lawson et al. (2020) measured mean DMS and MeSH mixing ratios of 208 and 18 ppt respectively, with maximum concentrations observed 305 at night. They also found that DMS and MeSH were correlated, with a slope of 0.07 and an R 2 of 0.3 over the full campaign. We observe similar mean MeSH (19.1 ppt), lower DMS (72.6 ppt), and a larger slope for the correlation of DMS and MeSH (slope = 0.19) in this study compared to the Lawson et al. (2020) observations. Still our results show general qualitative agreement with Lawson et al (2020), with both showing that atmospheric MeSH is present at a significant ratio relative to DMS. 310 observational constraint on BrO during our study, we elect to exclude BrO chemistry from the model base case used in subsequent calculations. Inclusion of BrO chemistry would have minimal impact on model MeSH as described, and would serve to reduce DMS lifetime, increase the yield of DMSO and MSA from DMS oxidation, and reduce the yield of SO 2 from DMS oxidation. The demonstrated model ability to reproduce observed DMS and MeSH mixing ratios during daytime when 385 we have robust constraints on the emission flux suggests that the DMS and MeSH oxidation mechanism added to the MCM v3.3.1 in this work is suitably accurate to provide meaningful information on the oxidative fate of DMS and MeSH.

= 0.15) with ocean chlorophyll concentrations, with FDMS:FMeSH reaching a maximum of 10.8 ± 4.4 during a phytoplankton bloom period. No other volatile sulfur compounds were observed by PTR-ToFMS to have a resolvable emission flux above their flux limit of detection or to have a gas phase mixing ratio consistently above their limit of detection during the study period, suggesting DMS and MeSH are the dominant volatile organic sulfur compounds emitted from the ocean at this site. 25 The impact of this MeSH emission source on atmospheric budgets of sulfur dioxide (SO2) was evaluated by implementing observed emissions into a coupled ocean-atmosphere chemical box model using a newly compiled MeSH oxidation mechanism. Model results suggest that MeSH emissions lead to afternoon instantaneous SO2 production of 2.5 ppt hr -1 , which accounts for 30% of the instantaneous SO2 production in the marine boundary layer at the mean measured FDMS and FMeSH.
This contribution of MeSH to SO2 production is driven by a higher effective yield of SO2 from MeSH oxidation and the shorter 30 oxidation lifetime of MeSH compared to DMS. This large additional source of marine SO2 has not been previously considered in global models of marine sulfur cycling. The field measurements and modeling results presented here demonstrate that MeSH https://doi.org/10.5194/acp-2021-891 Preprint. Discussion started: 3 November 2021 c Author(s) 2021. CC BY 4.0 License.
is an important contributor to volatile sulfur budgets in the marine atmosphere, and must be measured along with DMS in order to constrain marine sulfur budgets. This large additional source of marine reduced sulfur from MeSH will contribute to particle formation and growth and CCN abundance in the marine atmosphere, with subsequent impacts on climate. 35

Introduction
Dimethyl sulfide (CH3SCH3, DMS) emissions from the ocean are the most abundant source of reduced sulfur to the marine atmosphere (Andreae, 1990;Bates et al., 1987bBates et al., , 1992Carpenter et al., 2012). The role of DMS as a driver of cloud condensation nuclei (CCN) production, which ultimately impacts Earth's radiative budget in the marine atmosphere, has been studied extensively (Bates et al., 1987a;Carslaw et al., 2013;Charlson et al., 1987;Quinn and Bates, 2011). The oxidation of 40 DMS in the atmosphere ultimately leads to the production of methane sulfonic acid (CH3SO3H, MSA) and sulfur dioxide (SO2) which can be further oxidized to sulfuric acid (H2SO4), contributing to particle formation and growth (Clarke et al., 1998;Hoffmann et al., 2016;Schobesberger et al., 2013;Sipila et al., 2010). Direct observations and mechanistic understanding of the intermediate products in the oxidation of DMS are limited, leading to large variability in estimates of SO2 yields (31 -98%), where SO2 is a precursor to sulfuric acid (H2SO4) and non-sea salt sulfate aerosol (nss-SO ) (Faloona, 2009;Hoffmann 45 et al., 2016). Previous efforts to constrain the total budget of SO2 in the marine boundary layer (MBL) have required assigning a near 100% yield of SO2 from DMS oxidation . A 100% yield of SO2 appears inconsistent with known production of MSA from DMS oxidation and with results from multiple laboratory studies (Faloona, 2009). The high yield of SO2 from DMS necessary for closure of the SO2 budget in that study prompted speculation on the existence of other unknown marine sulfur species which could contribute to SO2 production (Gray et al., 2011). Existence of other marine contributors to 50 SO2 production would serve to reduce the implied SO2 yield from DMS and would potentially bring that yield closer to the 40-80% range typically determined in laboratory and modeling studies (Gray et al., 2011). Implementation of oceanic MeSH emissions and oxidation to SO2 in chemical transport models would result in an increase in sulfuric acid production with subsequent impacts on new particle formation and growth and CCN abundance.

DMS and MeSH oceanic production 55
Both DMS and the volatile reduced sulfur molecule methanethiol (CH3SH, MeSH) are produced in seawater from the same precursor metabolite, dimethylsulfoniopropionate (DMSP) (Kiene and Linn, 2000a). Bacterial cleavage of dissolved DMSP (DMSPd) primarily produces dissolved DMS (DMSd), and DMSP demethylation or demethiolation produces dissolved MeSH (MeSHd) (Yoch, 2002). MeSHd is the dominant product of DMSPd consumption with a total yield on the order of 75%, compared to an approximately 10% yield of DMSd (Kettle et al., 2001;Kiene andLinn, 2000b, 2000a). While yields of MeSHd 60 are higher compared to DMSd, MeSHd is also more rapidly consumed by heterotrophic bacteria and phytoplankton, resulting in significantly faster turnover times (hours for MeSHd, days for DMSd) and lower steady state dissolved concentrations (Kiene, 1996). Both MeSHd and DMSd are persistently supersaturated in the dissolved phase resulting in ventilation to the atmosphere https://doi.org/10.5194/acp-2021-891 Preprint. Discussion started: 3 November 2021 c Author(s) 2021. CC BY 4.0 License. (Kettle et al., 2001;Lee and Brimblecombe, 2016). Surface ocean concentrations of DMSd have been measured extensively and are on the order of 1-7 nM with higher values in the summer (Lana et al., 2011). This extensive collection of measurements 65 has permitted the development of global climatologies of DMSd and emission fluxes for implementation into global chemical transport models (Galí et al., 2018;Lana et al., 2011). In contrast, measurements of MeSHd are sparse. Underway measurements from a transect of the Atlantic in September and October of 1998 showed a mean MeSHd of 0.39 ± 0.34 nM and a maximum of 1.7 nM (Kettle et al., 2001). Mean MeSHd during that study was approximately 20% of DMSd. In the Baltic Sea, mean MeSHd was 0.16 ± 0.12 nM compared to 2.6 ± 1.6 nM for DMSd (Leck and Rodhe, 1991). MeSH emission fluxes 70 were estimated to be 10% of DMS in that study. Significant variability in DMSd to MeSHd ratios were observed in those studies, emphasizing the need for more detailed study of the biogeochemical factors that control relative consumption and production of DMSd and MeSHd.

Atmospheric Fate of DMS
Once emitted to the atmosphere, DMS is oxidized by hydroxyl (OH), nitrate (NO3), chlorine (Cl), and bromine oxide (BrO) 75 radicals to produce lower volatility oxidized products which can contribute to aerosol particle formation and growth (Bates et al., 1987a;Charlson et al., 1987;Quinn and Bates, 2011). Gas phase mixing ratios of DMS in the MBL are typically on the order of 50 -600 ppt, with higher concentrations generally associated with regions of high phytoplankton abundance, and with diel maxima at night when oxidative loss is at a minimum (Kettle et al., 2001;Kettle and Andreae, 2000;Kim et al., 2017;Lana et al., 2011;Lawson et al., 2020;Sciare et al., 2000). Oxidation by OH, thought to be the largest loss pathway, proceeds 80 either through OH addition or hydrogen abstraction. OH addition produces MSA, methane sulfinic acid (MSIA), dimethyl sulfoxide (DMSO), and SO2, while H-abstraction is traditionally thought to primarily produce SO2 (Barnes et al., 2006;Conley et al., 2009;Hoffmann et al., 2016). The branching fraction of the OH oxidation channels is highly temperature dependent with H-abstraction favored at higher temperatures (~70% at 298 K). The DMS H-abstraction product rapidly produces the methylthiomethyl peroxy radical (MTMP, CH3SCH2OO) following recombination with atmospheric oxygen. Until recently it 85 was thought that MTMP primarily participates in further bimolecular reactions with the hydroperoxyl radical (HO2), nitric oxide (NO), or other peroxy radicals (RO2) which efficiently produce the methyl thiyl radical (CH3S·) and ultimately SO2 (Barnes et al., 2006;Hoffmann et al., 2016). Theoretical and laboratory studies have shown that MTMP can also undergo a series of intramolecular hydrogen shift rearrangements and additions of O2 to form the stable product hydroperoxymethyl thioformate (HPMTF; HOOCH2SCHO) at a rate that is competitive with bimolecular reactions (Berndt et al., 2019;Wu et al., 90 2015). HPMTF has been shown to be globally ubiquitous in the marine boundary layer from airborne observations (Veres et al., 2020) and at a coastal ocean ground site . Global chemical transport modelling shows that 46% of emitted DMS goes on to form HPMTF (Novak et al., 2021). The atmospheric fate of HPMTF is an active topic of research but ambient observations show that dry deposition to the ocean surface is a significant loss term (lifetime ~30 hours, ) and that HPMTF is efficiently lost to clouds (Novak et al., 2021;Veres et al., 2020;Vermeuel et al., 2020), 95 resulting in a 35% decrease in global SO2 production from DMS oxidation (Novak et al., 2021). These previously unconsidered https://doi.org/10.5194/acp-2021-891 Preprint. Discussion started: 3 November 2021 c Author(s) 2021. CC BY 4.0 License. loss processes for DMS derived sulfur may necessitate reevaluation of marine SO2 budgets (Bandy et al., 2011;Faloona et al., 2009).

Atmospheric Fate of MeSH
Comparatively little is known about the atmospheric abundance of MeSH. To date, only two ambient observations of MeSH 100 atmospheric mixing ratios have been presented in the literature. Measurements in the remote south-west Pacific showed MeSH mixing ratios from <10 ppt to 65 ppt which were 3-36% of coincident DMS mixing ratios (Lawson et al., 2020). DMS and MeSH emission fluxes were inferred in that study from the rate of accumulation at night when oxidative loss was assumed to be zero, which showed that MeSH emission fluxes were 14-24% of the DMS flux. MeSH mixing ratios in the coastal and inshore waters west of the Antarctic peninsula were up to 3.6 ppt, which was 3% of coincident DMS (Berresheim, 1987). To 105 our knowledge there have been no previous direct eddy covariance measurements of MeSH emission flux from the ocean.
While the bimolecular rate constants of MeSH with the primary atmospheric oxidants (OH, BrO, NO3, Cl) are known, there has been limited study on the reactive intermediates or yields of stable products from MeSH oxidation (Butkovskaya and Setser, 1999;Tyndall and Ravishankara, 1991). However, oxidation of MeSH by OH (R1) has been shown to produce the methyl thiyl radical (CH3S·) at a yield of 1.1 ± 0.2 (Tyndall and Ravishankara, 1989), providing a mechanistic link to known 110 reactions in the DMS H-abstraction pathway (R1) (Barnes et al., 2006). CH3S· production from DMS H-abstraction is the major pathway for SO2 production from DMS, and the reactions of CH3S· are therefore well studied (Barnes et al., 2006).
Other studies have shown that the reaction of MeSH with OH occurs primarily at the S-H group (0.87 ± 0.03) forming CH3S·, with a minor channel of H-abstraction from the methyl group (0.13 ± 0.03) (Butkovskaya and Setser, 2021). Recent ab initio/RKKM calculations determined the CH3S· yield from MeSH + OH at 298 K and 760 torr to be 0.98 (Mai et al., 2020). 115 Given the slight inconsistency between the directly measured CH3S· yield of 1.1 ± 0.2 and the MeSH + OH branching fractions, we take these experiments to provide an upper and lower bound on the CH3S· yield of 0.87 and 1.1, and assume a CH3S· yield of 1 for R1 throughout this analysis.
The atmospheric yield of SO2 from the oxidation of MeSH by OH under low NOx was recently reported as 0.98 based on 135 modeling results constrained by a laboratory oxidation study, which is in good agreement with the efficient production of SO2 from MeSH proposed in our reaction scheme (Chen et al., 2021). We exploit this link between MeSH oxidation by OH and known DMS oxidation chemistry to develop a MeSH oxidation mechanism for implementation into a 0-D chemical box model as described further in the subsequent text. The full set of reactions with rate equations are provided in Table S1.
Here we present eddy covariance (EC) flux measurements of DMS and MeSH emissions at a coastal ocean site using a high 140 sensitivity Vocus PTR-TOF for detection (Krechmer et al., 2018). Results from this study show that emission fluxes of DMS (FDMS) and MeSH (FMeSH) were well correlated (R 2 =0.65) and that FMeSH is a significant contributor to marine sulfur emissions (mean FMeSH = 0.21 ppt m s -1 , compared to mean FDMS = 1.13 ppt m s -1 ). The average ratio of individual DMS to MeSH flux (FDMS:FMeSH) measurements for the full campaign was 5.5 ± 3.0. We assess the impact of the observed large MeSH emission flux on production of SO2 in the marine atmosphere through a coupled ocean flux -atmospheric chemistry 0-D box model with 145 a newly compiled MeSH oxidation mechanism. Modeling results suggest that MeSH contributes approximately 30% of instantaneous afternoon SO2 production (PSO2). Together these results show that MeSH emissions are an important contributor to sulfur budgets in the marine atmosphere and further field studies and laboratory oxidation mechanism investigations are warranted.

Scripps Pier Flux Experiment Overview
Measurements of DMS and MeSH gas phase mixing ratios and EC flux were made continuously from the end of the Ellen Browning Scripps Memorial Pier (hereon SIO Pier) at the Scripps Institution of Oceanography in La Jolla, CA, USA during September 2019. The SIO Pier is 330 m long and extends over 100 m beyond the wave breaking zone. The SIO Pier site has been used regularly for EC studies of ocean-atmosphere trace gas exchange (Kim et al., 2014;Porter et al., 155 2018;Vermeuel et al., 2020). The SIO Pier experiences a characteristic sea-breeze circulation pattern during summer where winds are from the ocean at moderate windspeeds (0-6 m s -1 ) during daytime and are from land at night, which limits nighttime flux determinations. DMS and MeSH were detected with a latest generation Vocus PTR-TOF (Vocus) instrument (TOFWERK, Aerodyne), with an HTOF mass analyzer (resolution ca 5000 m/Δm for DMS and MeSH) (Krechmer et al., 2018). The Vocus was housed in a temperature-controlled trailer at the end of the pier and sampled through a 19 m long PFA 160 inlet (0.625 cm i.d.) enclosed in an opaque housing to prevent photochemistry in the sampling line. The inlet was pumped at https://doi.org/10.5194/acp-2021-891 Preprint. Discussion started: 3 November 2021 c Author(s) 2021. CC BY 4.0 License. 22 slpm which maintained turbulent flow and short residence times in the sampling line (Reynolds number 4280, calculated volumetric evacuation time 1.7 s). The full inlet line was held at 40°C which was always above ambient temperatures to prevent condensation of water vapor on inlet surfaces. A bypass line through the Vocus front end subsampled from the main inlet at 5 slpm through a PFA tee located at the instrument interface. The Vocus subsampled from this bypass at a flow rate of 100 sccm 165 immediately in front of the Vocus capillary inlet into the instrument drift tube. This sampling configuration was used to reduce residence times in the sampling lines as much as possible (total estimated inlet and instrument residence time ~1.9 s). In addition to the main inlet line, all surfaces in contact with the ambient sample flow, including unions and valves, were composed of PFA or PEEK in order to minimize known surface artifacts for MeSH sampling except for one stainless-steel union at the Vocus subsampling point (Perraud et al., 2016). The ambient inlet sampling point was collocated with a sonic 170 anemometer recording three-dimensional winds at 10 Hz (Gil HS-50). The sonic anemometer and Vocus inlet were mounted on a 6.1 m long boom extended beyond the end of the pier to minimize flow distortions from the pier. The inlet was mounted on the boom at a height of 13 m above the mean lower low tide level. Ocean depth below the pier sampling point was ca. 6 m.
The Vocus was operated at a drift tube pressure of 1.5 mbar and an axial electric field gradient of 41.5 V cm -1 resulting in a high overall effective field strength (E/N) of 143 Td. Mass spectra were recorded at 10 Hz for the full mass range of 19-500 175 m/Q. The second RF-only focusing quadrupole in the instrument was operated at an amplitude of 275 V at the start of the campaign before being reduced to 235 and ultimately 215 V later in the deployment. The reduced amplitude on the quadrupole increased transmission of low mass (<50 m/Q) ions as described by Krechmer et al., (2018). MeSH (m/Q 49) transmission was increased by 10% at the reduced amplitude of 215 V, compared to 275 V, which was accounted for in the data processing.
Transmission efficiency of DMS was independent of the quadrupole amplitude as its nominal mass (m/Q 63) is larger than the 180 mass discrimination window of the quadrupole. High resolution peak fitting and integration of the mass spectra was performed in the Tofware 3.2.0 software developed by the instrument manufacturer (TOFWERK).
Additional ancillary measurements made continuously from the pier included O3 mixing ratios, temperature, relative humidity, and incoming solar irradiance. O3 mixing ratios were measured at 1-minute time resolution (POM, 2B Technologies) in line with the Vocus with a subsampling point immediately downstream of the Vocus subsampling point. Temperature and RH 185 (Vaisala HMP110) were also measured inline downstream of the Vocus subsampling point at 1 Hz time resolution. Incoming total solar irradiance at 1 Hz time resolution (Licor LI-200R) was measured via a sensor mounted on top of the trailer housing the Vocus. Measurements of sea-surface temperature (SST), salinity, and chlorophyll are continuously collected at a 1-minute time resolution from the end of the pier from an automated shore station operated by the Southern California Coastal Ocean Observing System (Wright, 2016). 190

DMS and MeSH calibrations and limit of detections
Instrument calibration factors for DMS were determined during ambient sampling by a two-point standard addition of a DMS gas standard (Praxair, 5.08 ppm ± 5%) to the full sampling inlet every 2.5 to 4 hours. Campaign mean sensitivity to DMS was 3.9 cps ppt -1 . Sensitivity to DMS was independent of RH. MeSH was not directly calibrated for during field sampling. Post https://doi.org/10.5194/acp-2021-891 Preprint. Discussion started: 3 November 2021 c Author(s) 2021. CC BY 4.0 License. campaign calibrations for MeSH were performed in the laboratory using a MeSH compressed gas standard (Airgas, 6.11 ± 5% 195 ppm) yielding a calibration factor of 1.3 cps ppt -1 for dry conditions. MeSH sensitivity was slightly humidity dependent resulting in the sensitivity at 80% RH being 11% lower than at 0% RH. Calibrations for MeSH were performed with the same 19 m sampling line and flow conditions used for ambient sampling. We expect that this reduced sensitivity for MeSH at high RH is at least partially due to reactions on inlet surfaces which is a known complication in MeSH sampling (Perraud et al., 2016). Ambient MeSH calibration factors were determined by scaling the measured in-field DMS calibration factors by the 200 RH dependent DMS:MeSH sensitivity ratio determined from the laboratory calibrations. We note that the observed sensitivity ratio of DMS:MeSH is different from what would be predicted based on measured proton transfer rate coefficients which are approximately equal for DMS and MeSH (Williams et al., 1998), emphasizing the need for direct calibration of MeSH. Instrument backgrounds were determined by overflowing the full inlet line with dry UHP N2 at the tip of the ambient sampling point. Ambient sampling periods were subdivided into 30-minute blocks and were matched to the nearest temporal calibration 205 and background determination point. The campaign mean, median, and interquartile range for DMS were 72.6, 49.2, and 28.0-89.8 ppt respectively. MeSH mean, median, and interquartile ranges mixing ratios were 19.1, 13.4, and 7.6-24.5 ppt. Limits of detection (LOD) at a signal-to-noise (S/N) ratio of three were 2.6 ppt for DMS and 3.6 ppt for MeSH for a 10 s averaging time following the calculation method of Bertram et al., (2011).

Eddy Covariance Calculation Overview
The flux (F) of trace gas across the interface is described by Eq. 1, as a function of both the gas-phase (Cg) and liquid phase (Cl) concentrations and the dimensionless gas over liquid Henry's law constant (H), where Kt, the total transfer velocity for the gas, encompasses all of the chemical and physical processes that govern air−sea gas exchange (Liss and Slater, 1974). Overbars are means and primes are the instantaneous variance from the mean. N is the total number data points during the flux 220 averaging period. Ambient data was subdivided into approximately 30-minute flux averaging periods for the EC flux calculation.

EC Flux Data Processing and Quality Control
Several standard EC data processing steps, data filters, and quality control checks were applied during flux analysis including: 225 1) filtering by wind direction for periods of onshore winds (true wind direction 200-360°), 2) coordinate rotation of threedimensional wind components by the planar fit method to remove unintentional tilts in the sonic mounting and account for local flow distortions (Wilczak et al., 2001), 3) application of a friction velocity ( * ) threshold of 0.05 cm s -1 to reject periods of low shear driven turbulence, 4) despiking of DMS and MeSH data using a mean absolute deviation filter before the EC flux calculation (Mauder et al., 2013), 5) linear detrending of the scalar timeseries, and 6) flux stationarity filtering with flux periods 230 rejected if they were non-stationary at a 30% threshold (Foken and Wichura, 1996) (http://mcm.york.ac.uk, Jenkin et al., 1997;Saunders et al., 2003) in the Framework for 0-D Atmospheric Modeling (F0AM, Wolfe et al., 2016) with added chlorine chemistry (Riedel et al., 2014). The model was used to assess the impact of observed 240 MeSH emissions on production of secondary marine SO2. Model ability to reproduce observed diel profiles of DMS and MeSH mixing ratios was also tested. Measured emission fluxes of MeSH and DMS are coupled into the model to set the source term for those molecules. Meteorological inputs were acquired from the NOAA National Data Buoy Center (Stations LJPC1 and LJAC1) as well as from an onsite temperature and relative humidity data logger (OM-62, Omega Engineering). Chemical constraints included coincident measurements of O3 and nitryl chloride (ClNO2) measured at the same site in August 2018 245 , an assumed OH profile that followed the solar cycle with a peak at 4.0 x 10 6 molecules cm -3 , and constant concentrations of other major trace gases as listed in Table S2. A constant first-order dilution loss term was used with a one-day lifetime to approximate mixing out of the boundary layer. A static boundary layer height (BLH) of 500 m was assumed (Faloona et al., 2005;Stull, 1988;Wei et al., 2018). Clear sky conditions were also assumed (i.e. no heterogeneous loss from reactive uptake on cloud droplets ). An updated oxidation mechanism for DMS and MeSH 250 was implemented, expanding upon the default DMS oxidation scheme in the MCM v3.3.1 to include oxidation of MeSH to form CH3S· (R1) and to include HPMTF chemistry, detailed in Table S1. The model was allowed to spin up for two days to allow reactive intermediates to reach equilibrium, with all reported values taken from day three of the model run. Two distinct model cases were developed which differ in how they treat the diel profile of FDMS and FMeSH. The first (termed the Pier Model Case) used the observed diel profile of FDMS and FMeSH at Scripps Pier to test model ability to reproduce observed diel profile of DMS and MeSH gas phase mixing ratios. Nighttime flux measurements from 23:00 to 09:00 were limited during this study due to persistent off-shore wind conditions. Instead, we apply a constant nighttime emission flux taken as the average of the 9:00-10:00 and 21:00-22:00 flux observations for DMS and MeSH. A three-hour moving mean was also applied to the 260 observed diel flux profiles to reduce the influence of experimental variability on the model. This Pier Model Case is used in the analysis of diel profiles presented in Section 3.4.

Open Ocean Case
A second case (termed Open Ocean Model Case) was developed to provide a general assessment of the relative contribution of DMS and MeSH emissions to SO2 production as described in

Meteorology Overview 275
Observed meteorology and ocean physical and biogeochemical parameters showed minimal variance over the sampling period.
Mean sea-surface temperature during the campaign was 23.3°C (21.6 to 24.7°C interquartile range). Observed mean and https://doi.org/10.5194/acp-2021-891 Preprint. Discussion started: 3 November 2021 c Author(s) 2021. CC BY 4.0 License. interquartile ranges of air temperatures and relative humidity were 22°C (19.5 to 23.6°C) and 79.9% (72.3 to 88.3%) respectively. Chlorophyll concentrations suggest moderate biological productivity with an observed campaign mean of 1.86 μg L -1 (1.5 to 2.0 μg L -1 interquartile range). O3 mixing ratios showed a clear diel pattern peaking in mid-afternoon with a 280 campaign mean of 32.6 ppbv (27.6 to 38.9 ppbv interquartile range). Wind speeds during onshore wind periods were from 0 to 6 m s -1 , typically peaking in late afternoon with a campaign mean of 2.8 m s -1 . Clear sky conditions were observed for most afternoons during the study period, with total solar irradiance peaking near noon. The period from day of year (DOY) 268-271 saw occasional overcast skies during the afternoon. Light rainfall during the early morning of the DOY 271 was the only precipitation during the campaign. Morning and late evening periods showed occasional presence of marine stratocumulus 285 clouds which drove day-to-day variability in solar irradiance during those times. The campaign timeseries of wind speed, air temperature, sea surface temperature (SST), O3 mixing ratios, solar irradiance, and DMS and MeSH mixing ratios are presented in Fig. 1. Gaps in the DMS and MeSH timeseries are from instrument maintenance periods, power outages at the site, and periods where the instrument was operated in an alternative sampling mode.

DMS and MeSH Gas Phase Mixing Ratios 290
The campaign mean, median, and interquartile range of DMS mixing ratios were 72.6, 49.2, and 28.0-89.8 ppt respectively. Both DMS and MeSH were observed to reach a maximum in concentration at night, and minimum concentrations in the early afternoon as shown in the diel profiles in Fig. 3. The observed diel profile is consistent with expectations due to the significantly lower oxidative loss rate at night and has been observed in other studies (Lawson et al., 2020). MeSH varies by approximately a factor of five between its diel average maximum and minimum concentration, compared to DMS which varies by 300 approximately less than a factor of three. The larger diel variability in MeSH is due to its approximately five times faster bimolecular rate constant with OH (kOH+MeSH = 3.3 × 10 -11 cm 3 molecule -1 s -1 at 293 K) compared to DMS (kOH+DMS = 7.8 × 10 -12 cm 3 molecule -1 s -1 at 293 K) resulting in a lifetime of MeSH to oxidation by OH during the afternoon on the order of 3 hours compared to 16 hours for DMS (for [OH] = 2 × 10 6 molecule cm -3 ). In the remote south-west Pacific Ocean, Lawson et al.
(2020) measured mean DMS and MeSH mixing ratios of 208 and 18 ppt respectively, with maximum concentrations observed 305 at night. They also found that DMS and MeSH were correlated, with a slope of 0.07 and an R 2 of 0.3 over the full campaign.
We observe similar mean MeSH (19.1 ppt), lower DMS (72.6 ppt), and a larger slope for the correlation of DMS and MeSH

DMS and MeSH Emission Flux
A total of 304 valid quality-controlled flux averaging periods were measured during the campaign. Campaign mean emission 320 fluxes of FDMS and FMeSH were 1.13 ppt m s -1 (0.53 -1.61 ppt m s -1 interquartile range) and 0.21 ppt m s -1 (0.10 -0.31 ppt m s -1 interquartile range) respectively. Both FDMS and FMeSH reached a steady maximum between hour of day 10 to 17 as shown in Fig. 3 which corresponds to the typical period of sustained peak wind speed. The magnitude of both FDMS and FMeSH were found to increase with wind-speed as shown in Fig 4, following expectations for supersaturated species at moderate wind speeds where flux magnitude is controlled by physical transfer terms (Carpenter et al., 2012;Huebert et al., 2004;Kim et al., 325 2017;Marandino et al., 2007). Measurement of FDMS and FMeSH during nighttime was limited due to persistent winds from the land at night throughout the campaign. Less than 15% of the valid flux observations were between the hours 22 -07. Further, those nighttime flux measurements were smaller and showed high variability compared to daytime measurements. DMS and MeSH fluxes were highly correlated with each other (R 2 =0.65) as shown in Fig. 5a. Campaign mean FDMS:FMeSH calculated as the simple mean of the ratio of individual FDMS and FMeSH observations was 5.5 ± 3.0. Lawson et al. (2020) calculated the 330 average emission flux of MeSH compared to DMS (FMeSH / (FMeSH + FDMS)) to be between 14% -24%, where fluxes were calculated from the nighttime accumulation of DMS and MeSH when oxidative loss was assumed to be negligible. In this study using direct eddy covariance flux measurements, we calculate the mean FMeSH / (FMeSH + FDMS) to be 16%, which compares well to the Lawson et al. (2020) result. As shown in Fig. 5b. FDMS:FMeSH is partially correlated (R 2 = 0.15) with ocean chlorophyll concentrations. The timeseries of chlorophyll concentrations shown in Fig. 1b. shows that chlorophyll 335 peaked at the immediate start of gas phase sampling from DOY 245-246 at ca. 3.5 to 4 μg L -1 before declining over the course of several days to roughly constant concentrations from 1 to 2.5 μg L -1 over the remainder of the campaign. The profile of chlorophyll suggests a phytoplankton bloom peak and decay was sampled in the first period of the campaign which transitioned into a roughly constant moderately biologically productive state for the remainder of the campaign. FDMS:FMeSH during the period of peak chlorophyll concentrations over the first three days of the campaign (DOY 245 -247) was 10.8 ± 4.4 compared 340 to the mean ratio from the full campaign of 5.5 ± 3.0. The relative production and consumption of DMS and MeSH in seawater is known to be a complex function of the speciation and abundance of phytoplankton and bacteria as well as available organic sulfur and other biogeochemical parameters (Kiene et al., 2000;Kiene and Linn, 2000b). No other measured meteorological parameters including wind speed, SST, and solar irradiance showed a significant correlation with FDMS:FMeSH. The underlying cause for the correlation between FDMS:FMeSH and chlorophyll in our dataset is not clear without additional constraints on the 345 ocean biochemistry. However, this result highlights that biological activity can drive variations in dissolved ratios of DMS and MeSH resulting in variability in ambient FDMS:FMeSH emission ratios, and that further study is needed to elucidate this mechanism and its spatiotemporal variability. Measurements during an induced mesocosm phytoplankton bloom experiment using seawater collected as SIO pier immediately before this study showed that the ratio of gas phase DMS to MeSH varied by more than a factor of five over the course of a phytoplankton bloom and decay (Kilgour et al., 2021). DMS to MeSH ratios 350 in that study were strongly linked to changes in bacterial sulfur demand and changes in the available pool of dissolved sulfur https://doi.org/10.5194/acp-2021-891 Preprint. Discussion started: 3 November 2021 c Author(s) 2021. CC BY 4.0 License. across the phytoplankton bloom and decay cycle. While induced mesocosm blooms are not fully representative of the ambient ocean, those results demonstrate the controlling role of ocean biology on FDMS and FMeSH and ultimately on marine SO2 production which must be better constrained through further ambient observations. 355

Chemical Box Model Comparison to Pier Observations 365
We assessed the ability of the coupled ocean-atmosphere chemical box model described in Section 2.4.2 using the Pier Model Case to replicate the observed mean diel profiles of DMS and MeSH mixing ratios from the SIO pier. The model and measurement diel profiles of DMS and MeSH are shown in Fig. 6. For MeSH the model agrees with measurements to within

Impact of MeSH on Marine Sulfur Dioxide Production 390
Production of SO2 as a function of FDMS and FMeSH was assessed using the Open Ocean case of the coupled ocean-atmosphere 0-D box model described in Section 2.4.2.2. Chemical box modeling of MeSH emission and gas phase oxidation suggest that MeSH contributes significantly to marine boundary layer SO2 concentration. For a model case where FMeSH is taken to be 0, and FDMS is taken to be the campaign mean of 1.1 ppt m s -1 , modeled afternoon (hours 12 to 16) mean SO2 mixing ratio is 29.5 ppt, and the instantaneous SO2 production rate (PSO2) is 5.8 ppt h -1 . When FMeSH is added to the model at the observed campaign 395 mean of 0.21 ppt m s -1 , model afternoon mean SO2 increases to 46.5 ppt and PSO2 increases to 8.3 ppt h -1 . Model diel profiles of SO2 mixing ratios and PSO2 with and without MeSH emissions are shown in Fig 7. In the campaign mean case MeSH emissions contribute 30% of the overall SO2 production (or a 43% increase in total SO2 production compared to the FDMS only case). The model yield of SO2 from MeSH oxidation was 0.99 which is comparable to an experimentally constrained model determination of the atmospheric yield of 0.98 (Chen et al., 2021). We include the full MeSH oxidation mechanism for 400 completeness due to its overlap with known DMS chemistry and ease of implementation in the box model. But given our https://doi.org/10.5194/acp-2021-891 Preprint. Discussion started: 3 November 2021 c Author(s) 2021. CC BY 4.0 License. determined model yield of SO2 from MeSH of 0.99, and a recently determined yield of 0.98 constrained by laboratory oxidation studies (Chen et al., 2021), future modeling efforts may be justified in simplifying this mechanism by including only a direct MeSH + OH → SO2 reaction at a yield of 1. Prior efforts to constrain the total SO2 budget in the marine boundary layer required a unit yield of SO2 from DMS  which stands in contrast to the 40-80% range typically determined in 405 laboratory and modeling studies (Faloona, 2009;Gray et al., 2011). This discrepancy has prompted consideration that other potential SO2 precursors might be present which might reduce the yield of SO2 from DMS needed to close the SO2 budget (Bandy et al., 2011). Oceanic MeSH emissions as observed in our study are likely one such additional contributor to secondary marine SO2 which has not been previously considered. Eddy covariance flux measurements of ocean-atmosphere trace gas exchange have generally been limited to a small set of molecules (e.g. DMS, acetone, methanol, acetaldehyde) (Novak and 410 Bertram, 2020). As demonstrated with the MeSH measurements presented here, marine EC flux observations of new molecules are critical to constrain marine sulfur and volatile organic compound (VOC) budgets. Global spatiotemporal variability of MeSH emission flux magnitude and the ratio of FDMS:FMeSH are both highly uncertain due to the sparsity of ambient observations, which will need to be better constrained through future studies.
Heterogeneous chemistry of the DMS oxidation product HPMTF is not included in our base model case. HPMTF 415 heterogeneous chemistry has been proposed to be a potentially large sink for HPMTF which would reduce SO2 production from DMS (Novak et al., 2021;Veres et al., 2020;Vermeuel et al., 2020). These details of HPMTF heterogeneous chemistry do not impact the yield of SO2 from MeSH described previously but do impact the calculated relative production of SO2 from MeSH compared to DMS. Inclusion of HPMTF heterogeneous chemistry (at γ=0.01 and aerosol surface area of 48 μm 2 cm -3 ) reduces model SO2 production from DMS to 2.7 ppt h -1 compared to 5.8 ppt h -1 in the model base case. In the HPMTF 420 heterogeneous chemistry case MeSH oxidation accounts for 48% of marine SO2 production. Further details on HPMTF chemistry are given in SI S3.
We also note that the yield of SO2 production from MeSH does not have a temperature dependence, unlike DMS, which may result in MeSH being an especially important source of SO2 in colder high latitude regions. At lower temperatures, the DMS OH-addition reaction pathway becomes more favored, resulting in less efficient production of SO2 from DMS oxidation as 425 production of highly soluble intermediates begins to dominate compared to at higher temperatures. Model calculations presented here used the measured diel temperature profile at the SIO pier during this study which had a mean of 293 K.
Modeled daytime PSO2 as a function of DMS and MeSH emission flux magnitude are shown in Fig. 8 with the interquartile range of FMeSH and FDMS measured SIO pier emission fluxes overlaid as a constraint. These results highlight the potential variability in PSO2 in varying regimes of FDMS:FMeSH. Model results were from the base Open Ocean Case described in Section 430 2.4.2.2. Given the potential biological control on FDMS:FMeSH, the temperature dependence of SO2 yield from DMS, and the impact of HPMTF heterogeneous chemistry on SO2 yield from DMS, we expect there may be significant temporal and regional variability in the relative contribution of DMS and MeSH to marine PSO2 across the global oceans. This additional SO2 production from MeSH will likely contribute to new particle formation and growth through enhanced production of sulfuric acid (H2SO4) with subsequent impacts on CCN abundance in the MBL. Given the newly determined significance of both 435 https://doi.org/10.5194/acp-2021-891 Preprint. Discussion started: 3 November 2021 c Author(s) 2021. CC BY 4.0 License.
MeSH emissions and HPMTF heterogeneous chemistry on marine SO2 production, a critical reevaluation of the global marine SO2 budget is likely warranted.  In addition to DMS and MeSH, several other volatile organic sulfur compounds (VOSCs) have been reported to be significant in the marine atmosphere, either from direct oceanic emissions or gas phase oxidation of precursor species. Recent shipborne observations in the Arabian Sea reported high mixing ratios of dimethyl sulfone (DMSO2, 40-120 ppt) and methane sulfonamide (MSAM, 20-50 ppt) downwind of a biologically productive upwelling region (Edtbauer et al., 2020). Reported DMS in the same region was from 100-300 ppt. The authors in that study propose a direct oceanic emission source of MSAM 450 as there is no known gas phase oxidation pathway to produce MSAM from DMS. For our observations at Scripps Pier, both MSAM and DMSO2 were found to have no observable emission flux above the flux limit of detection at either the standard 3σ LOD threshold or a more relaxed 1σ LOD threshold. Gas phase mixing ratios of MSAM, DMSO, and DMSO2 were also not found to be significant, with none of those species consistently observed above the 10 s averaging time LOD of 2.4, 7.0, and 9.2 ppt respectively. All determinations of MSAM, DMSO, and DMSO2 mixing ratio and LOD assume they have the same 455 detection sensitivity as DMS. Our measurements of DMSO and DMSO2 are consistent with box model calculated mixing ratios which show hourly maxima of 4.9 ppt and <0.1 ppt for DMSO and DMSO2, respectively which are both below the instrument LOD. The box model conditions used were the Pier Model Case with the addition of [BrO]max of 1 ppt. Oxidation of DMS by BrO favors DMSO and DMSO2 production and the inclusion of [BrO]max of 1 ppt was done as a test of the reasonable upper limit case for DMSO and DMSO2 production for conditions during this study. Dimethyl disulfide (DMDS, 460 CH3S2CH3) was also not consistently observed above the LOD in our measurements.
While a direct oceanic emission source of MSAM to the atmosphere has been proposed (Edtbauer et al., 2020), we did not measure a detectable MSAM emission flux in this study. As noted by Edtbauer et. al. (2020), given the Henry's law constant of MSAM (KH = 3.3 × 10 4 M atm -1 ), the waterside concentration of MSAM necessary to drive a net emission flux of MSAM to the atmosphere is on the order of 1700 nM. Given that maximum total dissolved organic sulfur (DOS) in ocean 465 surface waters are on the order of 200 nM (Ksionzek et al., 2016), even if all surface ocean DOS was in the form of MSAM, we still would not expect an emission flux of MSAM to the atmosphere. This exercise suggests MSAM may instead be formed through an unknown reaction pathway in the atmosphere which was active in the Arabian Sea but not in our observations at SIO Pier in coastal southern California. Taken together our observations indicate that MSAM, DMSO, and DMSO2 emission fluxes and mixing ratios were small at the SIO Pier site consistent with model calculations. 470 This discussion on the lack of observations of other VOSCs only extends to molecules that are detectable with the Vocus PTR instrument, such as DMSO, DMSO2, MSAM, and DMDS. For example, it does not include inorganic sulfur containing molecules such as carbonyl sulfide (OCS) and carbon disulfide (CS2) which also have a known oceanic source but are not readily detectable by PTR. In particular our measurements do not include the recently observed DMS oxidation product HPMTF which was found to be a globally ubiquitous sulfur reservoir from airborne observations in the marine atmosphere 475 (Veres et al., 2020) and in prior surface observations at the SIO Pier . HPMTF is not detectable with the Vocus PTR ion chemistry used in this study. HPMTF was previously observed in the summer of 2018 at the SIO Pier site https://doi.org/10.5194/acp-2021-891 Preprint. Discussion started: 3 November 2021 c Author(s) 2021. CC BY 4.0 License.
to have a strong diel profile, peaking in the early afternoon with an average daytime mixing ratio of 12.1 pptv . While observations of HPMTF were made one year prior to the DMS and MeSH observations reported here, it suggests that HPMTF may comprise up to 10-20% of the total daytime VOSC concentration at this site during summer. 480

Conclusions
We present the first direct eddy covariance flux measurement of MeSH emissions from the ocean which show that MeSH emissions account for a mean of 16% (FMeSH/(FMeSH + FDMS)) of emitted volatile organic sulfur measurable with the Vocus PTR-ToF during this study. DMS and MeSH emission fluxes were correlated with each other (R 2 = 0.65), consistent with their shared oceanic source from the degradation of DMSP. Measured FDMS:FMeSH was found to have a weak correlation (R 2 = 0.15) 485 with chlorophyll concentrations, which highlights the need for further study of biogeochemical cycling in the ocean surface which may drive significant spatiotemporal variability in FMeSH. The atmospheric implications of ocean MeSH emissions were assessed by development of a MeSH oxidation mechanism and incorporation into a coupled ocean-atmosphere 0-dimensional chemical box model. Modeling results show that oxidation of MeSH by OH produces SO2 at a high yield (~99%) and is an important contributor to SO2 production in the marine atmosphere, driving an increase in afternoon PSO2 of 2.5 ppt hr -1 , 490 corresponding to 30% of total afternoon SO2 production rates. In a model case including HPMTF heterogeneous uptake to aerosols, SO2 yield from DMS is reduced and MeSH becomes even more important, accounting for 48% of marine SO2 production. Taken together these results demonstrate that MeSH is an important contributor to volatile sulfur budgets in the marine atmosphere and that further studies are needed to constrain spatiotemporal trends of MeSH emission and oxidation relative to DMS. 495 Code and Data Availability. DMS and MeSH flux and concentration data and associated meteorological data from this study are archived at http://digital.library.wisc.edu/1793/82383.
Author Contributions. GAN, DBK, and THB designed research. GAN carried out the ambient sampling campaign and analyzed the data. MPV assisted with the ambient deployment. DBK conducted laboratory calibrations and developed sampling methods. CMJ and MPV contributed to model development. THB supervised the project. GAN and THB wrote the 500 paper. All authors reviewed and edited the paper.
Competing Interests. Timothy H. Bertram is a member of the editorial board of Atmospheric Chemistry and Physics. The authors have no other competing interests to declare.