Investigating the Tropospheric Chemistry of Acetic Acid Using the Global 3‐D Chemistry Transport Model, STOCHEM‐CRI

Acetic acid (CH3COOH) is one of the most abundant carboxylic acids in the troposphere. In the study, the tropospheric chemistry of CH3COOH is investigated using the 3‐D global chemistry transport model, STOCHEM‐CRI. The highest mixing ratios of surface CH3COOH are found in the tropics by as much as 1.6 ppb in South America. The model predicts the seasonality of CH3COOH reasonably well and correlates with some surface and flight measurement sites, but the model drastically underpredicts levels in urban and midlatitudinal regions. The possible reasons for the underprediction are discussed. The simulations show that the lifetime and global burden of CH3COOH are 1.6–1.8 days and 0.45–0.61 Tg, respectively. The reactions of the peroxyacetyl radical (CH3CO3) with the hydroperoxyl radical (HO2) and other organic peroxy radicals (RO2) are found to be the principal sources of tropospheric CH3COOH in the model, but the model‐measurement discrepancies suggest the possible unknown or underestimated sources which can contribute large fractions of the CH3COOH burden. The major sinks of CH3COOH in the troposphere are wet deposition, dry deposition, and OH loss. However, the reaction of CH3COOH with Criegee intermediates is proposed to be a potentially significant chemical loss process of tropospheric CH3COOH that has not been previously accounted for in global modeling studies. Inclusion of this loss process reduces the tropospheric CH3COOH level significantly which can give even larger discrepancies between model and measurement data, suggesting that the emissions inventory and the chemical production sources of CH3COOH are underpredicted even more so in current global models.


Introduction
Acetic acid (CH 3 COOH) plays a significant role in the acidity of precipitation and cloud water in the troposphere (Galloway et al., 1982;Keene et al., 1995;Khare et al., 1999;Vet et al., 2014;Zhang et al., 2011). Keene and Galloway (1986), and Andreae et al. (1988) proposed that carboxylic acids could account for as much as 80-90% of precipitation acidity in remote regions of the world. Because of the characteristic hygroscopic nature of the carboxylic acids and their ability to act as cloud condensation nuclei (Yu, 2000), they are responsible for absorbing and dispersing solar radiation, thus altering the global thermal balance.
The well-constrained sources of CH 3 COOH in the atmosphere include anthropogenic, vegetation, soil, and biomass burning which can contribute to the atmospheric burden of CH 3 COOH (Paulot et al., 2011). While direct anthropogenic emissions are a significant source of CH 3 COOH in industrialized areas, biogenic emissions are of greater importance in remote regions (Chebbi & Carlier, 1996). Soil emissions may be an important source of CH 3 COOH, especially in environments with low levels of terrestrial vegetation (Enders et al., 1992;Sanhueza & Andreae, 1991). Owing to the extensive array of factors that impact the emission of CH 3 COOH from soil, that is, soil composition, vegetation cover, temperature, humidity, and other meteorological parameters, local and global fluxes remain poorly catalogued. While direct vegetative emissions of CH 3 COOH do occur (Schäfer et al., 1992;Talbot et al., 1990Talbot et al., , 1995 and contribute to its atmospheric presence, it is indirect production following photochemical conversion of precursor volatile hydrocarbons (e.g., isoprene) that represents the main source of atmospheric CH 3 COOH and leads to the seasonal pattern of atmospheric concentrations (Paulot et al., 2011). The unconstrained sources include motor exhaust emissions (Grosjean, 1989(Grosjean, , 1992Kawamura et al., 1985;Zervas et al., 2001), atmospheric transformation of enols (Archibald et al., 2007), and aqueous phase oxidation of methylglyoxal (Carlton et al., 2006;Lim et al., suggests that the photochemical degradation of particulate organic matter associated with both marine aerosols and biomass burning may be a source of CH 3 COOH. The relatively small spatial variability in concentration and wet-deposition flux of carboxylic acids over broad geographic regions in their study suggests a fairly uniform marine source strength globally. The key chemical production channel is thought to be the oxidation of olefins producing CH 3 COOH via the Criegee intermediate (Calvert & Stockwell, 1983). The Criegee reaction that produces CH 3 COOH can be written as shown in Scheme 1.
Scheme 1 depicts the ozone-initiated oxidation of propene, although it is noted that trans-2-butenes and other alkenes that form the acetaldehyde oxide moiety will have analogous chemistry. There are several reactions in this scheme that reduce the efficiency of acetic acid formation through this mechanism, notably the following: 1. The formation of excited Criegee intermediates that do not survive long enough to participate in the reaction with water and its dimer 2. The primary ozonide may form the formaldehyde oxide Criegee intermediate, which will not lead to direct formation of acetic acid 3. The syn-acetaldehyde oxide and anti-acetaldehyde oxide can undergo isomerization, with the former leading to a vinyl hydroperoxide (a nonacetic acid-producing channel) and the latter producing excited acetic acid, which can decompose before it is collisionally stabilized.
The importance of processes 1 and 2 are defined by the alkene that is undergoing ozonolysis, with yields of excited versus stabilized and acetaldehyde versus formaldehyde oxide varying with the types of substitution about the olefinic bond. The importance of 3 depends upon the rates of the isomerization reactions and how they compete with the reaction with water and its dimers and are therefore dependent not only on the rate coefficients of these processes but also the relative humidity.
In polluted urban environments, the reactions of peroxyacetyl species (RCO 3 ) with NO 2 are dominant, producing peroxyacyl nitrates-one of the components of photochemical smog. However, in rural environments or areas with only moderate levels of pollution, the reactions of peroxyacetyl radicals with HO 2 and RO 2 can produce a significant amount of CH 3 COOH (Jacob & Wofsy, 1988;Madronich & Calvert, 1990): Reaction (2a) is very slow (~10%) compared with the reaction (2b) at room temperature. However, a significant amount of CH 3 CO 3 (reaches up to~4 ppt in the tropics) and RO 2 (consisting CH 3 O 2 and an additional 46 organic peroxy radicals reach up to 40 ppt in the tropics) can exist in the atmosphere (Khan, Cooke, Utembe, Archibald, Derwent, et al., 2015) and therefore reaction (2a) is anticipated to be a significant source of CH 3 COOH in clean tropospheric environments.
By virtue of the low reactivity of CH 3 COOH, its removal from the troposphere by chemical reactions occurs slowly and its major sinks in the atmosphere are dry and wet deposition (Seco et al., 2007). Precise estimates for dry and wet deposition rates of CH 3 COOH are of importance as its tropospheric lifetime is strongly controlled by the deposition loss (Chebbi & Carlier, 1996). However, recent studies Welz et al., 2014) suggested a very rapid chemical reaction of Criegee intermediates with carboxylic acids. Therefore, it is possible that the role of gas phase loss processes has been underestimated by previous studies.
There are a few modeling studies, for example, MOGUNTA by Baboukas et al. (2000), MATCH-MPIC by von Kuhlmann et al. (2003), IMPACT by Ito et al. (2007), and GEOS-Chem by Paulot et al. (2011) where the global budgets of CH 3 COOH have been explored. The most detailed global budget of CH 3 COOH is presented by Paulot et al. (2011), where they produced bottom-up estimates of the global sources of CH 3 COOH following the construction of an updated emissions inventory and chemical scheme and evaluated the model results against an extensive array of ground, ship, satellite, and flight campaign data. The GEOS-Chem model encapsulates the seasonality of CH 3 COOH concentrations well but generally underestimates their tropospheric levels indicating an undefined source of atmospheric CH 3 COOH (Paulot et al., 2011).
The global budget and distribution of CH 3 COOH are not well reproduced in global atmospheric models because of the poor inventory of sources and sinks (Rosado-Reyes & Francisco, 2006;Singh et al., 2000). This is a consequence of the lack of clarity in the relative importance of known sources or the possible importance of unknown sources and/or errors in the sink processes of CH 3 COOH at a regional and global scale and several studies highlight sizeable discrepancies between model and measurement data (Ito et al., 2007;Paulot et al., 2011;von Kuhlmann et al., 2003). Therefore, an investigation into the tropospheric chemistry of CH 3 COOH using the global 3-D chemistry transport model STOCHEM-CRI has been undertaken. The simulations are performed with updates to the chemical mechanism (e.g., water complexation of peroxy radicals and their subsequent reactions with CH 3 CO 3 , SOA oxidation, loss of CH 3 COOH by Criegee) and emissions inventory of the species (e.g., acetaldehyde and isoprene) of the model, and comparisons of the results made with other modeling studies as well as surface stations and flight campaigns measurement data. The relevance of the reaction of gas phase CH 3 COOH with Criegee intermediates as a potential loss process of CH 3 COOH has also been explored and is unaccounted for in the previous modeling studies.

Model Description
The 3-D global chemical transport model employed during this investigation was that of STOCHEM. An original version of the model was presented by Collins et al. (1997) with updates detailed in Derwent et al. (2008). STOCHEM is an "offline" model driven by archived meteorological data. The meteorological data were generated by the UK Met Office Unified Model at a climate resolution of 1.25°longitude × 0.83°latitude × 12 vertical levels up to a boundary of 100 hPa (Johns et al., 1997). Parameters incorporated in the data include temperature, wind, humidity, pressure, tropopause height, clouds, surface parameters, and boundary layer depth.
STOCHEM advects 50,000 constant mass air parcels using a Lagrangian approach such that the chemistry and transport processes are uncoupled. Within each of the parcels, chemical species are produced and lost in chemical reactions and photochemical dissociations. Physical processes acting on chemical species within the parcels include emissions, dry and wet deposition, convection, and subgrid scale mixing between air parcels (Derwent et al., 2003). The chemical mechanism used in STOCHEM is the common representative intermediates mechanism version 2 and reduction 5 (CRI v2-R5). The detail of the CRI v2-R5 mechanism is given by Jenkin et al. (2008), Watson et al. (2008), and Utembe et al. (2009). Further amendments to the chemical mechanism and a description of the addition of an organic aerosol module are reported by . The chemical scheme comprises 244 chemical species that take part in 564 chemical reactions and 98 photolysis reactions. The concentration of each chemical species is updated using a backward Euler integration with an advection chemical time step of Δt = 5 min (Collins et al., 1997). The photolysis rate for any given reaction is determined explicitly for each individual parcel with a time resolution of 1 hr. These rates are subsequently interpolated linearly to achieve the 5-min resolution values required for the chemical integration. More details about the photolysis scheme implemented in the model can be found in Khan, Cooke, Utembe, Archibald, Maxwell, et al. (2015).
Emissions within STOCHEM are characterized into three different types: surface, stratospheric, and threedimensional emissions. Surface emissions from ocean, soil, vegetation, and biomass burning are distributed using two-dimensional source maps at a resolution of 5°longitude × 5°latitude (Olivier et al., 1996), which varies by calendar month. Emissions totals for nitrogen oxides, carbon monoxide, and nonmethane volatile organic compounds for the year 1998 are taken from the Precursors of Ozone and their Effects in the Troposphere inventory (Granier et al., 2005). Further details of the STOCHEM emissions inventory for all species can be found in Khan et al. (2014).
Dry deposition is accounted for using a resistance approach in STOCHEM and the rate at which it occurs depends on whether it is over land or ocean with relevant species-dependent deposition velocities for both. The dry deposition velocities of CH 3 COOH over land and ocean were considered as 1.30 and 1.34 cm/s, respectively , which were used in all simulations. Wet deposition is represented in STOCHEM through the use of species-dependent scavenging coefficients. The dynamic scavenging of 2.0 cm À1 and convective scavenging of 4.0 cm À1 for CH 3 COOH are determined from a relationship plot of Henry's law solubility coefficients (taken from Sander, 2015) and scavenging coefficients (taken from Penner et al., 1994) for a suite of species. These scavenging coefficients are combined with scavenging profiles and precipitation rates to calculate the wet deposition loss rates for CH 3 COOH from an air parcel.
Four simulations were performed during the study: STOCH-Base, STOCH-AO, STOCH-WC, and STOCH-IM. The simulation, STOCH-Base, was based on the reference conditions described in  with additional amendments of the global emissions of CH 3 COOH (anthropogenic: 9.67 Tg/year, biomass burning: 11.23 Tg/year, soil: 3.48 Tg/year, and vegetation: 2.58 Tg/year) taken from Paulot et al. (2011). The rate coefficients and branching ratios for the production and loss reactions were extracted from the Master Chemical Mechanism website (http://mcm.leeds.ac.uk/MCM/). More details about the production and loss processes implemented in the model can be found in supporting information (Table S1; Butkovskaya et al., 2006;McGillen et al., 2017;Vereecken et al., 2017). Acetaldehyde is one of the major precursors of CH 3 COOH, which has a strong oceanic source; therefore, a simulation, STOCH-AO, was conducted by including the global emissions of acetaldehyde (oceanic: 57 Tg/year, biomass burning: 3 Tg/year, anthropogenic: 2 Tg/year, and biogenic: 23 Tg/year) into STOCH-Base based on the emissions inventory reported by Millet et al. (2010). An oceanic emission of isoprene (top-down estimate of 1.9 Tg/year by Arnold et al., 2009) was also included in the STOCH-AO simulation. The complexation of HO 2 and RO 2 radicals with water vapor molecules (Clark et al., 2008;Kanno et al., 2006) accelerate the reaction with CH 3 CO 3 producing higher CH 3 COOH. Thus, in the third simulation STOCH-WC, the reactions of water complexed HO 2 and RO 2 with CH 3 CO 3 were considered with a factor of 2 increment of the rate coefficients of the reactions (Khan, Cooke, Utembe, Archibald, Derwent, et al., 2015). Isoprene is another major precursor of CH 3 CO 3 ; a further simulation, STOCH-IM, was conducted with updating global vegetation emissions of isoprene (594 Tg/year) reported by Sindelarova et al. (2014). All simulations were conducted with meteorology from 1998 for a period of 24 months with the first 12 allowing the model to spin up. Analysis is performed on the subsequent 12 months of data. The Criegee chemistry was integrated in STOCH-IM to calculate the steady state concentration of stabilized Criegee Intermediates (sCIs) by including the production of the Criegee intermediate (CI) from the ozonolysis reactions of six alkenes (e.g., ethene, propene, t-but-2-ene, isoprene, α-pinene, and β-pinene) using the rate coefficients taken from the Master Chemical Mechanism (http://mcm.leeds.ac.uk/) and their individual loss through unimolecular, water, water dimer, and CH 3 COOH reactions. The rate coefficients of the loss of Criegee by unimolecular, water, and water dimer can be found in McGillen et al. (2017). The rate coefficient of the loss of Criegee by CH 3 COOH can be found in the supporting information (Table S2; Chhantyal-Pun et al., 2018;Khan et al., 2018;Welz et al., 2014). The CH 3 COOH lost by reaction with Criegee intermediates is determined as a percentage of the total theoretical amount of CH 3 COOH lost via the all chemical reactions and physical process (e.g., deposition). The CH 3 COOH hot spots in the tropics are found with mixing ratios by as much as 1.6 ppb in South America ( Figure 1). These hot spots can be ascribed to the large quantities of CH 3 COOH produced by the reactions (1) and (2a) in these regions. Peroxyacetyl radicals are abundant in the atmosphere over South America due to the large quantities of biogenic volatile organic compound precursors that are released from the forest. Isoprene is one of the predominant precursors to CH 3 CO 3 and has a large emission flux in these regions. In addition, the relatively unpolluted section of the troposphere above the forest contains moderately low levels of nitrogen oxides pollutants of about 500 pptv. Therefore, the abundant CH 3 CO 3 are not consumed by reactions with nitrogen oxides but instead react with HO 2 and RO 2 radicals to generate CH 3 COOH.

Surface Distribution
The higher concentrations of CH 3 COOH are found during the dry season due to a slower loss rate owing to lower levels of deposition and corresponding lower concentrations during the wet season. This is indeed the case in Figure 1 for the northern parts of South America (e.g., Brazil) with comparatively higher CH 3 COOH concentrations during the dry season (J-J-A) and the comparatively lower CH 3 COOH concentrations during the wet season (D-J-F). Similar agreement between dry and wet seasons and CH 3 COOH concentrations exists for the other hot spots of CH 3 COOH located in Africa. For countries in central Africa (e.g., Central African Republic) the dry season falls between December and February and the wet season lies between July and September but Angola, located on the west coast of Southern Africa, has distinct dry (J-J-A) and wet (D-J-F) seasons. Thus, the higher CH 3 COOH concentrations are found in central Africa and Southern Africa during D-J-F and J-J-A, respectively.
Including the oceanic emissions of acetaldehyde and isoprene in STOCH-AO increases by 8.9 Tg/year CH 3 COOH chemical production globally which increases the concentration of CH 3 COOH relative to the STOCH-Base by up to 30 ppt (400%) at the equator in the Pacific and Indian Ocean during J-J-A season and up to 25 ppt (700%) over the Southern Hemisphere (SH) oceans during D-J-F season ( Figure 2). The large secondary production of CH 3 COOH from acetaldehyde and isoprene in the marine environment confirms the significant source of CH 3 COOH over the oceans.
The zonal distribution plot shows that the peaks of CH 3 COOH (up to 0.20 ppb) are found at the tropics because of larger biogenic emissions within this region (Figure 3a). The high levels of CH 3 COOH (up to 0.1 ppb) at the top of the model domain in STOCH-Base are found in the tropics because of the ubiquitously distributed CH 3 CO 3 throughout the troposphere after forming from the oxidation of most nonmethane volatile organic compounds (VOCs). The secondary production of CH 3 COOH from oceanic sources in STOCH-AO increases the CH 3 COOH concentration by up to 250% in the SH oceans ( Figure 3b). The oceanic acetaldehyde

Journal of Geophysical Research: Atmospheres
and isoprene emissions can produce large amounts of CH 3 CO 3 and peroxy radicals resulting in increasing the mixing ratios of CH 3 COOH in these regions.
An increase of 12% in the CH 3 COOH level was found in the marine environment after considering the water complexation of peroxy radicals and the subsequent reactions of the water complexed HO 2 and RO 2 with CH 3 CO 3 (Figure 4a). Upon updating the emission inventory of isoprene in the STOCH-IM simulation gave a further increase (up to 20%) of surface CH 3 COOH over the forested regions of South America, South Africa and Australia (Figure 4b) because of the increased formation of CH 3 CO 3 from the oxidation of isoprene.

Model-Measurement Comparison
The comparison between measurement and model is not straight forward for CH 3 COOH and several caveats must be kept in consideration. The meteorological data for the year 1998 was used in the model; thus, the meteorology can have impact on the model values when compared with measured data for other years.
In the study, a comparison of modeled vertical profiles of CH 3 COOH with aircraft measurements was conducted for four designated campaigns ( Figure 5)  (Table S3). There was a general underprediction of CH 3 COOH by the model over all altitudes (mean bias of 0.05 ppb for STOCH-Base simulation). An improvement in the accuracy of the model with the reduction of mean biases of 14%, 6%, and 16% is apparent for all flight data following the updating the acetaldehyde emission inventory in STOCH-AO, increasing rate of the reaction of CH 3 CO 3 + HO 2 /RO 2 due to water complexation of peroxy radicals in STOCH-WC and updating the isoprene emission inventory in STOCH-IM, respectively.
The vertical profile of CH 3 COOH for remote oceans in STOCH-AO ( Figure 5) shows a significant increase of the model CH 3 COOH in the Tropical Pacific, Easter Island, Hawaii, Christmas Island and a slight increase in the Tahiti and Fiji. A further improvement of the model CH 3 COOH in the tropical oceans are seen in STOCH-WC due to the complexation of peroxy radicals with water resulting in faster rates of peoxy-peroxy radical reactions. The concentrations of RO 2 .H 2 O in tropical areas are as high as 5 ppt, which correspond to approximately 12% of the peroxy radicals (Khan, Cooke, Utembe, Archibald, Derwent, et al., 2015). The formation of complexes of peroxy radicals with water reduce the energy barrier of the transition state for the peroxy-peroxy radical reaction, which enhance the formation of CH 3 COOH in the tropical oceans. Despite an increase in CH 3 COOH concentration between the STOCH-Base and STOCH-AO simulations, the STOCH-AO concentrations remain up to twofold to threefold lower than the measured concentrations for Fiji and Tahiti. One of the possible reasons for the large discrepancies may be the sample air mass which was affected by a biomass burning plume from Southeast Asia or Australia during the PEM Tropics-B campaign.
The location of Ireland is such that the atmospheric composition is heavily impacted by SW prevailing winds and is therefore representative of marine Atlantic air. Adding the oceanic emissions in STOCH-AO still underpredicts the concentration of CH 3 COOH in Atlantic oceans. These comparisons emphasize the need for further investigation into the origins of CH 3 COOH in the troposphere, with a focus on midlatitudinal oceanic regions. Further data were collected over Maine in the United States and Newfoundland in Canada as part of the SONEX study. A similar increase in CH 3 COOH concentration persists over the four STOCHEM simulations, and the model underpredicts when contrasted with measurement data. One possible explanation is higher anthropogenic emissions of CH 3 COOH in the northeastern United States and Canada than are predicted in the emissions inventory. The underprediction is also visible for the campaigns, TexAQS and ITCT2K4, which were heavily influenced by anthropogenic pollution and anthropogenic + biomass burning plumes, respectively. Acetic acid can be produced from vinyl alcohol (Archibald et al., 2007), which are likely to be produced from combustion sources (anthropogenic or biomass). Inclusion of vinyl alcohols and their chemistry in STOCHEM-CRI can reduce the discrepancy between model and measured CH 3 COOH in these campaign sites.
In general, the model runs, STOCH-Base, STOCH-AO, STOCH-WC, and STOCH-IM, produce a better modelmeasurement comparison for tropical region compared with midlatitudinal regions. We compared midlatitudinal (e.g., East Atlantic, Ireland, Newfoundland, Maine, ITCT2K4, and TexAQS II) measured CH 3 COOH with model CO (tracer for anthropogenic and biomass burning emissions) and found an excellent correlation (R 2 = 0.83) between them ( Figure S1). A sensitivity experiment was conducted with an arbitrary additional anthropogenic emission of CH 3 COOH (100 Tg/year) introduced to STOCH-IM in a run hereafter referred to as STOCH-ANTH (orange lines in Figures 5 and 6). The vertical profiles of SONEX, TexAQS, and ITCT2K4 flight campaigns show a significant increase of tropospheric CH 3 COOH in STOCH-ANTH, which brings the model into closer agreement with most of the measurements suggesting that there could be significant amount   (Table S4; Andreae et al., 1988;Arlander et al., 1990;Haase et al., 2012;Harrington et al., 1993;Helas et al., 1992;Khwaja, 1995;Kumar et al., 1996;Norton, 1992;Preunkert et al., 2007;Puxbaum et al., 1988;Quesada et al., 2001;Sanhueza et al., 1996;Schultz Tokos et al., 1992;Talbot et al., 1988). Talbot et al. (1988) monitored the gas phase atmospheric concentration of CH 3 COOH over a 15-month period at a site in eastern Virginia. Data from this study for each of the 12 months of the year are plotted alongside concentrations from the STOCHEM simulations in Figure 6. This plot provides a useful insight into how well STOCHEM replicates the seasonality of CH 3 COOH at a given location. In the field campaign, a marked annual seasonality was noted with higher concentrations in the growing season (March-September) than the nongrowing season (October-February). Peak concentrations were registered in the summer month of July. The model data display moderate agreement with the trend in measurement data, replicating lower levels of CH 3 COOH in the winter months of December, January, and February, and generate much higher concentrations in the peak growing season months of June-July-August. However, there is a large disagreement between model-measurement data throughout the year with a twofold to threefold lower CH 3 COOH level in model data than that in measurement data. A good correlation is found (R 2 = 0.61) between measured CH 3 COOH and model isoprene (tracer of biogenic emissions) for this site ( Figure S2). A sensitivity experiment was conducted to investigate the missing biogenic sources of acetic acid by considering an arbitrary addition of biogenic emissions of 50 Tg/year to STOCH-IM in a run refereed as STOCHEM-BIO (violet lines in Figures 5 and 6). The model-measurement agreement has been improved significantly for the forested and rural sites (e.g., Los Caracas, Venezuela; Amazonia region, Brazil; Surinam; Eastern Virginia, United States) when additional biogenic emission is included; this supports the importance of unknown biogenic precursors with large yields of CH 3 COOH which were not considered in the model. Sanhueza et al. (1996) measured the gas phase concentration of CH 3 COOH at a number of sites in Venezuela during the dry and wet seasons. Average monthly concentrations from a coastal site at Los Caracas are Figure 6. The monthly variation of model and measured CH 3 COOH mixing ratios. Red squares, blue diamonds, green triangles, yellow rounds, violet stars, and orange crosses represent the model data for STOCH-Base, STOCH-AO, STOCH-WC, STOCH-IM, STOCH-BIO, and STOCH-ANTH simulations, respectively. Black squares represent measurement data, and the black error bars represent measurement variability.

10.1029/2018JD028529
Journal of Geophysical Research: Atmospheres compared here with the STOCHEM simulation data. Good agreement is found between model and measurement data with simulated concentrations falling within the measurement data error bars (see Figure 6). The model-measurement agreement is also found to be reasonable for the DECAFE experiment and ICARTT campaign where the CH 3 COOH was measured from the tropical forest of the People's Republic of the Congo in February 1988  and from the Appledore Island, ME, United States, in July-August 2004 (Haase et al., 2012). Excellent agreement is observed between model and measurement data recorded during April-May of the 1987 wet season at a site in the Amazonian region in Brazil (Talbot et al., 1990). In addition to the high concentrations of peroxyacetyl radicals and peroxy radicals in the Amazonian region, there are persistently high relative humidity levels. Thus, STOCH-WC brings the model into closer agreement with measurement at this site during the wet season. However, the model underpredicts CH 3 COOH abundances relative to measurement data recorded during the dry season at this site. The increased isoprene emissions in STOCH-IM slightly increase the CH 3 COOH concentration, but model data remain up to twofold to threefold lower than the measured data during dry season. It is possible that the biomass burning in the dry season exceeds the emissions predicted in our model and is responsible for the higher measurement data concentrations.
Data for the Pacific Ocean and Indian Ocean sites have been taken from a cruise of the SH Pacific Ocean and Indian Ocean by Arlander et al. (1990) during which the measurements of CH 3 COOH were recorded. The measurement points at 0.28 ppb and 0.29 ppb represent the average concentration of CH 3 COOH in SH Pacific Ocean and Indian Ocean recorded for this leg of the cruise. A slight increase in CH 3 COOH concentration is simulated at the SH Pacific and Indian Oceans when the oceanic emissions of acetaldehyde and isoprene are considered in STOCH-AO. However, the model still underpredicts CH 3 COOH concentrations relative to the measurement data. There are several possible explanations for this. First, there may be direct oceanic sources of CH 3 COOH, which have not been accounted for in the STOCHEM model and are contributing to the higher concentrations of CH 3 COOH observed in the measurement data. Second, there may be other secondary sources of CH 3 COOH (e.g., the degradation of marine organic aerosol; Keene et al., 2015) that are not included in STOCHEM. In the study, the average CH 3 COOH mixing ratio values for the cruise were used for the model validation, so there is a possibility several of the values at higher latitudes were impacted by anthropogenic emissions, or biomass burning plumes. Paulot et al. (2011) also observed their model underpredicted CH 3 COOH concentrations in remote marine environments and therefore further investigations of these discrepancies are highly desirable.
The model-measurement agreement is also found to be within one standard deviation of the mean measured values for the semirural site in Agra, India, during the winter and summer (Kumar et al., 1996) and for the Vallot observatory in French Alps (4,360 m above sea level) during September 2004 (Preunkert et al., 2007). Puxbaum et al. (1988) measured the gas phase concentration of CH 3 COOH at three sites in eastern and northern Austria. Unlike the previous stations discussed, the values of the concentrations for all model runs are similar in magnitude for the station in Austria, with concentrations from the simulations found to be highest for August. The concentrations of NO 2 over mainland Europe are expected to be higher than those of the previous remote/rural sites discussed and therefore most of the peroxyacetyl radicals present in the troposphere react preferentially with NO 2 species to form peroxyacyl nitrate. Thus, they are not participating in reactions that produce CH 3 COOH in these regions.
A similar trend in these data exists for the modeled concentrations of CH 3 COOH over Japan, New York, United States, and California, United States. Output concentrations from all simulations are found to be similar in magnitude. While the model underpredicts abundances for the site in Austria by a factor of 2, the modeled concentrations for Japan, New York, and California are significantly lower than the measurement data points. The measurement data points for Japan and New York represent the average concentrations of CH 3 COOH measured on a particular day in August and October at an urban site in Yokohama, Japan, and a semiurban site in New York, United States. The measured data points for California are representative of the concentrations of CH 3 COOH for forested regions of the Sierra Nevada that are impacted by pollutants generated in the central valley region of California (Harrington et al., 1993). It is probable that the local anthropogenic emissions from motor exhaust emissions and industrial sites can be responsible for the high CH 3 COOH concentrations recorded during these three studies. We found a good correlation between measured CH 3 COOH and model n-butane (tracer of anthropogenic emissions) for the urban sites (e.g., Yokohama, Japan, New York, United States, California, United States, and Austria; Figure S3). The additional anthropogenic emissions of 10.1029/2018JD028529

Journal of Geophysical Research: Atmospheres
CH 3 COOH in STOCH-ANTH reduce the discrepancies significantly between model and measurements for these sites ( Figure 6) supporting that there could be large unknown anthropogenic emissions of CH 3 COOH either directly or through secondary production from the anthropogenic precursors.
The photochemical degradation of secondary organic aerosol can be another source of CH 3 COOH in the urban areas (Malecha & Nizkorodov, 2016;Pan et al., 2009;Sareen et al., 2013). Considering the photolysis rate of SOA as J SOA = 4 × 10 À4 .J NO2 (Hodzic et al., 2015) yielding 100% CH 3 COOH, we examined the production of CH 3 COOH from the photodegradation of SOA from the STOCHEM model run (referred to as STOCH-ASOL) and found a negligible amounts of CH 3 COOH production globally (100 kg/year) from this reaction channel. The heterogeneous oxidation process can be a substantial sink of SOA mass in the troposphere (Hodzic et al., 2016). If we consider the lifetime of organic particles toward heterogeneous oxidation is 2 days (Molina et al., 2004) and the oxidation process produces 100% CH 3 COOH, the STOCHEM simulation (STOCH-ASOL) produces a further 205 kg/year of CH 3 COOH. Table 1 presents the global budget of CH 3 COOH for the six simulations performed during this investigation, in addition to the global budget from Paulot et al. (2011), von Kuhlmann et al. (2003, Ito et al. (2007), and Baboukas et al. (2000). In all simulations presented in the study, the direct emissions only represent 20-26% of the total sources, which are in excellent agreement with the study of Paulot et al. (2011), von Kuhlmann et al. (2003, and Baboukas et al. (2000; see Table 1). The large percentage production contribution (74-80%) from chemical reactions dominated by CH 3 CO 3 + HO 2 and CH 3 CO 3 + RO 2 in the simulations (STOCH-Base, STOCH-AO, STOCH-WC, and STOCH-IM) highlights the importance of the peroxyacetyl radical in the production of CH 3 COOH in the troposphere.

Global Budget
Total CH 3 COOH sources of 102.8 Tg/year in the STOCH-Base simulation were found to be higher than the studies of Paulot et al. (2011), von Kuhlmann et al. (2003, and Ito et al. (2007) who estimated sources to be 85.5, 93.0, and 73.0 Tg/year, respectively. The better representation of CH 3 CO 3 during oxidation of biogenic VOC (e.g., isoprene) in the STOCH-Base (Khan, Cooke, et al., 2017) gives a chemical production of CH 3 COOH (75.9 Tg/year). The updated acetaldehyde emission inventory in STOCH-AO, the rate enhancement of the reaction of CH 3 CO 3 + HO 2 /RO 2 due to water complexation of HO 2 and RO 2 in STOCH-WC, and the updated isoprene emission inventory in STOCH-IM increase the chemical production of CH 3 COOH to 84.8, 91.8 and 107.4 Tg/year, respectively (Table 1). Baboukas et al. (2000) predicted a total chemical production of atmospheric CH 3 COOH of 120 Tg/year, which is much higher than that shown in our study. As von Kuhlmann et al. (2003) explained, it is likely that Baboukas et al. (2000) overpredicted the production of CH 3 COOH from the reaction of peroxyacetyl radicals with HO 2 due to the selection of a branching ratio yielding a proportion of CH 3 COOH that is too high.
The global sinks of CH 3 COOH are dominated by wet deposition (53%) followed by dry deposition (25%) and the loss by OH (22%) for the STOCH-Base simulation. The increased wet deposition loss is associated with the Journal of Geophysical Research: Atmospheres increased concentrations of CH 3 COOH in the tropical rainforest region brought about by their substantial production from peroxyacetyl radical reactions in low NO x environments. The scavenging coefficients used in the model was estimated from the relationship between Henry's law solubility coefficients and scavenging coefficients of a suit of VOCs. These estimated scavenging coefficients could be overestimated values, which increase the wet deposition loss in our model study compared with Paulot et al. (2011) study. The global burden (0.45-0.61 Tg) of CH 3 COOH for all simulations in the study is in best agreement with that of the value from Paulot et al. (2011), but the tropospheric lifetime of CH 3 COOH (1.6-1.8 days) is found to be lower than those obtained in the study of Paulot et al. (2011). The model-measurement discrepancies suggest the possible unknown or underestimated sources which can contribute to the global burden of CH 3 COOH.

Loss of CH 3 COOH by Criegee Intermediates
Recent studies suggested that the reaction of Criegee intermediates with organic acids may be of importance for the tropospheric chemistry of CH 3 COOH Welz et al., 2014). An effort has been made to establish the scale of the potential contribution of these reactions to the tropospheric levels of CH 3 COOH. The results illustrate that the reaction of CH 3 COOH with Criegee intermediates exceeds 60% of the total chemical loss of CH 3 COOH in South America, South Africa, Australia, and parts of Southeast Asia (Figure 7). The high concentrations of CH 3 COOH in the Amazon rain forest area coincide with high concentrations of sCI, and their reactions could significantly reduce the atmospheric concentration of CH 3 COOH. Chhantyal-Pun et al. (2017) found that these reactions show a relatively small temperature dependence over the temperature range found in the lower troposphere. Incorporating these reactions of Criegee intermediates with CH 3 COOH in the model results in reduction of CH 3 COOH in the Amazon rain forest area. The identification of this significant loss process of CH 3 COOH would indicate that the sources of CH 3 COOH are currently underpredicted even further.

Conclusion
The production of CH 3 COOH from the reactions of the peroxyacetyl radical has been shown to be the principal source of tropospheric CH 3 COOH, which is consistent with the results of other studies. CH 3 COOH concentration hot spots are located primarily in the tropics owing to the high concentration of peroxyacetyl radicals and low nitrogen oxide concentrations in this region. There is reasonable agreement between the STOCHEM model and measurement data, but the model significantly underpredicts concentrations in urban and biomass burning environments suggesting the potential unknown sources and/or errors in the sink processes of CH 3 COOH. The reaction of CH 3 COOH with sCIs is shown to be a potentially significant loss process of CH 3 COOH in the troposphere. The Criegee intermediates may be responsible for over 60% of the chemical loss of CH 3 COOH in South America, South Africa, Australia, and parts of Southeast Asia. Thus, the Criegee intermediates make a significant contribution to the reduction of tropospheric lifetime of CH 3 COOH. The highly oxygenated species are likely to be produced from this reaction which can lead to nucleation and secondary organic aerosol (SOA) formation. However, the products of this reaction need further investigation to understand its impact on SOA formation regionally and globally.