Contribution of the gas-phase reaction between hydroxyl radical and sulfur dioxide to the sulfate aerosol over West Pacific

. Sulfate is among the major components of atmospheric aerosols or fine particulate matters. Aerosols loaded with sulfate result in low air quality, damage to ecosystems, and influences on climate change. Sulfate aerosols could originate from that directly emitted to the atmosphere and that produced by atmospheric physicochemical processes. The latter is generated from sulfur dioxide (SO 2 ) via oxidation either in the gas phase reactions or in the aqueous phase. Several mechanisms of SO 2 oxidation have been proposed, but the differentiation of the various mechanisms and identification of the sources remain 25 challenging. To meet this need, a new method to estimate the contribution of the gas-phase reaction between hydroxyl radical (OH) and SO 2 to the sulfate aerosol is proposed and investigated. Briefly, we consider the OH-reaction rates of the respective trace gases that compete for OH radicals with SO 2 in the troposphere, and estimate the fraction of SO 2 -OH reaction in the total OH reactivity. Then the relationship between sulfate concentration and the SO 2 -OH reaction is analyzed statistically to investigate the sources of sulfate in aerosols. We test this method using the data from ground-based observations and aircraft 30 measurements made during the Effect of Megacities on the transport and transformation of pollutants on the Regional to Global scales in Asia (EMeRGe-Asia) over the western Taiwan and West Pacific regions. Our results show that the estimated SO 2 OH reactivity fraction is well-correlated with sulfate concentration. The sulfate production from SO 2 -OH reaction accounts for approximately 30% of the total sulfate in aerosols collected at the surface and near-surface (altitude < 600 m) in our study area, comparable to the estimates from other model simulations. Within its assumptions and limitations, this new method 35 provides a valuable approach to investigate the significance of SO 2 -OH reaction regionally and globally.

Abstract. Sulfate is among the major components of atmospheric aerosols or fine particulate matters. Aerosols loaded with sulfate result in low air quality, damage to ecosystems, and influences on climate change. Sulfate aerosols could originate from that directly emitted to the atmosphere and that produced by atmospheric physicochemical processes. The latter is generated from sulfur dioxide (SO 2 ) via oxidation either in the gas phase reactions or in the aqueous phase. Several mechanisms of SO 2 oxidation have been proposed, but the differentiation of the various mechanisms and identification of the sources remain 25 challenging. To meet this need, a new method to estimate the contribution of the gas-phase reaction between hydroxyl radical (OH) and SO 2 to the sulfate aerosol is proposed and investigated. Briefly, we consider the OH-reaction rates of the respective trace gases that compete for OH radicals with SO 2 in the troposphere, and estimate the fraction of SO 2 -OH reaction in the total OH reactivity. Then the relationship between sulfate concentration and the SO 2 -OH reaction is analyzed statistically to investigate the sources of sulfate in aerosols. We test this method using the data from ground-based observations and aircraft 30 measurements made during the Effect of Megacities on the transport and transformation of pollutants on the Regional to Global scales in Asia (EMeRGe-Asia) over the western Taiwan and West Pacific regions. Our results show that the estimated SO 2 -OH reactivity fraction is well-correlated with sulfate concentration. The sulfate production from SO 2 -OH reaction accounts for approximately 30% of the total sulfate in aerosols collected at the surface and near-surface (altitude < 600 m) in our study area, comparable to the estimates from other model simulations. Within its assumptions and limitations, this new method 35 provides a valuable approach to investigate the significance of SO 2 -OH reaction regionally and globally.

Introduction
Sulfate aerosol is a dominant component of particulate matters that significantly influences environmental issues, such as air quality, climate change, human health, and acid deposition. Sulfate is an important part of acid rain and ecosystem acidification, resulting from sulfur deposition (Greaver et al., 2012;Gorham et al., 1984). Using the average sulfate 40 concentrations measured from five cities in western Taiwan during 2017 -2020, reported by the Environmental Protection Administration of Taiwan, sulfate accounts for approximately 20% by weight of atmospheric particulate matters with a diameter less than 2.5 micrometers (PM 2.5 ).
Understanding the origin of sulfate is an essential prerequisite for governments to develop an environmental policy that reduces 45 air pollution and improves air quality and human health. Sulfate aerosol can be divided into two categories known as primary sulfate and secondary sulfate. The major sources of primary sulfate include sea sprays, thermal power plants, and volcanic eruptions (Allen et al., 2002), whereas sulfur dioxide (SO 2 ) is the dominant precursor of secondary sulfate through gas-phase oxidation and aqueous-phase oxidation (Stockwell et al., 2012).

50
There are primarily two mechanisms by which SO 2 is oxidized in the atmosphere. These comprise both gas-phase and aqueousphase reactions. Though oxygen (O 2 ) is the most abundant oxidant in the troposphere, the significant energy barrier of the SO 2 -O 2 endothermic reaction slows the reaction rate at atmospheric temperatures, making the reaction of O 2 with SO 2 of minor importance (Hindiyarti et al., 2007). Similarly, the reaction rate coefficients of oxidation of SO 2 with ozone (O 3 ) and the hydroperoxyl radical (HO 2 ) are also of negligible importance. Hydroxyl radical (OH), though having concentrations much 55 smaller than that of O 2 , is the most important gas-phase oxidant for SO 2 oxidation (Stockwell and Calvert, 1983). SO 2 is first oxidized to HOSO 2 by a termolecular reaction involving an inert reaction partner (M) to take up excess kinetic energy. HOSO 2 https://doi.org/10.5194/acp-2021-788 Preprint. then reacts with O 2 in the air to form sulfur trioxide (SO 3 ) and release an HO 2 . Other oxidants that could also oxidize SO 2 include stabilized Criegee intermediates (Boy et al., 2013;Kurtén et al., 2011;Ye et al., 2018), organic peroxy radicals (RO 2 , where R is an unspecified organic group) (Kurtén et al., 2011;Kan et al., 1981), and superoxide ions (O 2 -) (Tsona et al., 2018). 60 SO 3 then reacts with water vapor (H 2 O) in the gas phase to form sulfuric acid (H 2 SO 4 ). SO 3 may also react directly with H 2 O on the wet surface of aerosols or rain droplets to form H 2 SO 4 .
In contrast to the gas phase oxidation of SO 2 in the atmosphere, the aqueous-phase oxidation mechanisms of SO 2 are not as well understood. This is in part related to the relatively high solubility of SO 2 and large dissociation equilibrium constants.
A variety of plausible pathways to form SO 2 ．H 2 O, HSO 3 -, and SO 3 2have been proposed. For example, dissolved O 3 is reported to oxidize dissolved SO 2 with a rate coefficient depending significantly on the pH of the solution (Maahs, 1983;Hoyle et al., 2016). Moreover, in the presence of some transition metals acting as catalysts, such as iron ions (Fe 3+ ) and manganese ions (Mn 2+ ), dissolved O 2 can oxidize SO 2 efficiently (Herrmann et al., 2000;Harris et al., 2013;Brandt and van 70 Eldik, 1995). Hydrogen peroxide (H 2 O 2 ) (Kunen et al., 1983) and organic peroxides (Wang et al., 2019;Dovrou et al., 2019) are also soluble and important oxidants in aqueous solutions such as tropospheric aerosols and rain droplets. They react with HSO 3ions to form SO 4 2ions. The SO 2 related dissociated sulfur-containing ions are oxidized as well by the OH radical through chain reactions (McElroy, 1986). Besides, recent reports suggest that SO 2 oxidation also occurs on interfacial surfaces of acidic micro-droplets in the absence of other oxidants at pH < 3.5 (Hung et al., 2018), and that nitrogen dioxide (NO 2 ) can 75 oxidize SO 2 in droplets in the presence of NH 3 (Ge et al., 2019). Despite the progress made on our understanding of the mechanisms by which SO 2 is oxidized in the gas-phase and aqueousphase, the contributions of the respective mechanisms to the total atmospheric sulfate concentrations remain unclear. This research aims to improve the current understanding of SO 2 gas-phase oxidation by enabling the fraction of gas phase and thus 80 also aqueous phase oxidation to be estimated accurately. We develop an approach that addresses the competition for key atmospheric oxidants between SO 2 and other important trace gases. Several assumptions and approximations are made to determine the fraction of OH reacting with SO 2 in the gas phase. This we call the gas-phase reaction rate fraction and is denoted as f SO2-OH hereafter. With the ground observation in Taiwan and aircraft measurements from EMeRGe-Asia field campaign on-board the HALO research aircraft of DLR (HALO, https://www.dlr.de/content/de/missionen/halo.html), the values of f SO2-85 OH at the surface and in the air are calculated, which are further utilized to estimate the contribution of the SO 2 -OH reaction to the total production of sulfate. This method is applied to areas within Taiwan and outside of Taiwan to verify its reliability and versatility.

EMeRGe-Asia data 90
The Effect of Megacities on the transport and transformation of pollutants on the Regional to Global scales over Asia (EMeRGe-Asia) was a field experiment led by the University of Bremen, Germany, conducted in collaboration with Academia Sinica, Taiwan. The EMeRGe-Asia aircraft field campaign took place from Taiwan, from March 10 th to April 9 th in 2018. This field campaign investigated the composition and influence of pollution plumes from major population centers in Eastern and Southeastern Asia utilizing the HALO aircraft research platform (see flight tracks in Fig. S1). Trace gases, including carbon 95 monoxide (CO), nitrogen monoxide (NO), the sum of reactive nitrogen oxides (NO y ), peroxyacetyl nitrate (PAN), volatile organic compounds (VOCs), SO 2 , and O 3 , as well as sulfate aerosol (diameter from 40 to 1000 nm) mass concentrations, were analyzed in this study. VOCs measured during the campaign included formaldehyde (CH 2 O), methanol (CH 3 OH), acetonitrile (CH 3 CN), acetaldehyde (CH 3 CHO), acetone (CH 3 OCH 3 ), isoprene (C 5 H 8 ), benzene (C 6 H 6 ), toluene (C 7 H 8 ), and xylene (C 8 H 10 ). Details of these measurements are listed in Table 1. All the data are reanalyzed to obtain a consistent 1-minute average 100 dataset if there is no specific clarification. In addition, the altitudes of airmasses are rounded to hundred meters for classification.
Although most of the measurements during the campaign were taken over a large area of the West Pacific, from the Philippines to Japan, there were many regular measurements made over Western Taiwan, typically at an altitude of approximately 600 m. 105 This study focuses on this area and these altitudes. Measurements in this area were mainly carried out between 8:00 to 17:00 (in local standard time, LST); sparse measurements performed outside this period are excluded from the 1-minute average dataset.

Surface observational data
Hourly ambient concentrations of trace gases, including CO, SO 2 , NO, NO 2 , NO x , and O 3, measured at the air quality stations 110 of the Taiwan Environmental Protection Administration (EPA), are analyzed in this study. The hourly data observed between 8:00 to 17:00 LST from March 17 th to April 7 th of 2018 are included in our analyses to compare the surface observations with the airborne measurements. Further information on the instrumentation at the surface stations is listed in Table S1. In addition, samples of fine particulate matters (i.e., PM 2.5 ) were collected at 11 sites distributed in central western Taiwan during the campaign period (from March 13 th to March 31 st , 2018). The chemical composition of the PM 2.5 samples was analyzed in the 115 laboratories of RCEC, Academia Sinica, and the daytime (8:00 to 19:00 LST) sulfate concentration is used in this study. The field sampling and in-lab chemical analyses are as described in Salvador and Chou (2014). The geographic locations where we sampled PM 2.5 are illustrated in Fig. S2.

Data analysis and assumptions
In this study, OH radical is assumed to be the dominant tropospheric oxidant of SO 2 in the gas-phase. It is challenging to 120 quantify the reaction rate of SO 2 -OH oxidation directly as sophisticated instruments are necessary to accurately measure OH radicals. Though model simulation provides information about OH radical concentration, discrepancies remain between model results and observations (e.g., Griffith et al., 2016;Lew et al., 2020). OH oxidizes not only SO 2 but also many other trace gases in the air, such as methane (CH 4 ), CO, NO, NO 2 , and VOCs. We examine the competition for OH radicals among these trace gases to estimate the fraction of OH reacting with SO 2 , i.e., f SO2-OH . 125 We estimate the f SO2-OH as defined by Eq. (1), where OH reactions with major trace gases (CH 4 , CO, NO, NO 2 , VOCs, and SO 2 ) are considered. This method constrains any overestimation of SO 2 oxidation rate in the absence of accurate knowledge about the concentration of OH radicals in air masses. The OH-trace gas reactions are either second-order reactions (CH 4 , CO, NO, and NO 2 ) or third-order reactions involving air as the third body. The SO 2 -OH reaction is such a third-order reaction but 130 is considered a pseudo-second-order reaction at low altitudes because of a reasonable approximation that the nitrogen (N 2 ) and O 2 have stable concentrations over a small altitude range. As a result, Eqs. (2)-(7) describe the OH reaction rate of each major trace gas by three parameters, including the rate coefficient, the concentration of the trace gas, and the concentration of OH radical. The rate constants of these OH reactions are summarized in Table S2 (Burkholder et al., 2020).
Among the major trace gases considered here, VOCs are not measured at the ground-based stations of Taiwan EPA. Although the concentrations of some VOCs were measured onboard HALO and the sum of the measured concentrations may represent the majority of the VOCs to a reasonable approximation, significant uncertainty could still remain in the estimate of f SO2-OH in 145 case of missing one or some of the key VOCs. Considering the challenge and difficulty of measuring all the significant VOCs, we propose a different approach. We assume that the total OH reactivity of VOCs (Eq. (6)) is proportional to the sum of OH reactivity of other trace gases (Eq. (2)-(5) and (7)), as written in Eq. (8).
This assumption is consistent with previous observations in urban areas (Yang et al., 2017). In addition, the analysis using the measurements of the EMeRGe-Asia campaign also indicates that the total OH reactivity of the measured VOCs is linearly correlated to that contributed by other major gaseous pollutants, i.e., CO, NO, NO 2 , SO 2 , and CH 4 (see Fig. S3). Consequently, the f SO2-OH is redefined as Eq. (9) by applying the above assumptions.

fSO2-OH
Because the proportional parameter "t" for OH-VOCs reactions in Eq. (9) is difficult to be determined accurately and because the observations for VOCs are not available at the ground-based stations, we further simplify Eq. (9) to Eq. (10) by ignoring that scaling factor. Thus Eq. (10) gives an approximation of f SO2-OH , which is denoted as f SO2-OH * . suggesting that the average ratios of each trace gas over SO 2 are required to calculate f SO2-OH * . These ratios are derived from field measurements in this study with the slope of linear regression using the least-squares method. However, the fits have significant errors for most trace gases, which means that the derived slopes might be arguable. These low correlations may 165 result from different weather conditions, solar radiation intensity, and instrumental detection limits (see Table S1), discussed in Section 4.1. Here, it is assumed that the air parcels achieve a quasi-stationary state and the O 3 concentration represents a measure of the oxidizing capacity of such air masses. This implies that the concentration of each trace gas divided by the concentration of O 3 can be viewed as the relative amount of reductant. Consequently, Eq. (11)

175
As the reaction rate of the reaction CH 4 -OH is less than that of reactions CO-OH, NO-OH, and NO 2 -OH (see Table S2 The airborne NO 2 is approximated by subtracting NO and PAN from NO y , assuming that NO y is composed of NO, NO 2 , and PAN in the daytime, whereas HNO 3 exists mainly in the form of particulate nitrate. However, there are fewer PAN measurements than that of NO and NO y . Consequently, we determine the PAN-NO y slope within specified region and altitude to estimate the proportion of PAN in NO y (shown in Fig. S4). Then the airborne NO 2 is approximated according to Eq. (14): 185 4 Results and discussion

Importance of normalization by O 3
We assume that the observed air plume was in steady-state, and that the ambient O 3 concentration was representative of the tropospheric oxidizing capacity of the air masses. Fig. 1 demonstrates the CO-SO 2 , NO-SO 2 , and NO 2 -SO 2 relationships before 190 and after the O 3 normalization using data from the Tainan station located in southwestern Taiwan. Low correlations between these trace gases before the O 3 normalization (Fig. 1a, 1c, 1e) are in part explained by the different weather conditions and solar radiation. For example, the photochemical consumption of air pollutants in the air masses from pollution sources traveling to the observation stations can diverge under different weather conditions, leading to different aging of the plume. The interspecies correlation significantly improved after normalization by O 3 (Fig. 1b, 1d, 1f). These differences suggest that the inter-195 species correlation is likely subject to atmospheric oxidation, and the O 3 -normalization approach can efficiently reduce the relevant interferences. Taking the stations located in western Taiwan into account, R 2 of CO-SO 2 , NO-SO 2 , and NO 2 -SO 2 increase from 0.062, 0.061, and 0.109 to 0.468, 0.428, and 0.531, respectively, after normalizing by ozone. The individual values for each station and trace gas are illustrated in Fig. S5.
The normalization of the airborne measurements leads to the same effects. We divide western Taiwan into five regions, as shown in Fig. S6, where the airborne measurement is distributed. With aircraft-observed data in these regions, the trace gases-SO 2 relationships are determined with the O 3 normalization method. The average R 2 of CO-SO 2 , NO-SO 2 , and NO 2 -SO 2 for the five regions increased from 0.397, 0.698, and 0.614 to 0.672, 0.723, and 0.726, respectively, after the ozone normalization.
The higher values of R 2 after the O 3 normalization imply that slopes have an improved accuracy which propagates into the 205 calculation of f SO2-OH * .

Relationship between SO 2 -OH reaction and sulfate concentration in Western Taiwan
We use the computed slopes from the plots of CO/O 3 -SO 2 /O 3 , NO/O 3 -SO 2 /O 3 , and NO 2 /O 3 -SO 2 /O 3 to estimate the values of f SO2-OH * for surface stations by using Eq. (13), which are then compared with the sulfate concentrations. As shown in Fig. 2a, the sulfate concentration positively correlates with f SO2-OH * measured at 11 surface stations. This correlation implies that SO 2 -210 OH reaction explains the sulfate variability to a significant extent at selected locations. The regression slopes would be close to zero if the contribution of SO 2 -OH reaction to sulfate was negligible. The regression intercept is explained as the amount of sulfate independent of SO 2 -OH reaction, e.g., primary sulfate, transported sulfate, and sulfate derived from aqueous SO 2 oxidation.

215
The importance of the difference in weather conditions and the origin of the transported air plumes is now considered. Three periods with similar weather conditions are investigated. The weather type is considered to describe the daily synoptic weather event log, obtained from the Taiwan Atmospheric event Database (Su et al., 2018). The first period under consideration was March 13 rd , 14 th , and 17 th -19 th , during which the southwesterly flow of air was dominant. The second period includes March 16 th , 22 nd , and 27 th , which is under northeasterly flow. The third period comprises March 23 rd -26 th , 28 th , and 29 th , which 220 contains weak synoptic weather events. The correlations of f SO2-OH * and the amount of sulfate during these three periods are higher than that for the entire observation period, as shown in Fig. 3. This indicates that weather conditions influence the formation of sulfate.
For the airborne observation data, we divided Western Taiwan into five regions (as shown in Fig. S6) and calculated the f SO2-225 OH * separately. A positive correlation between sulfate and f SO2-OH * is also observed in the airborne observation, as shown in Fig. 2b. As the derived positive slope is marginally significant (p-value = 0.218), sulfate and f SO2-OH * analyses with adequate airborne measurement data are necessary to validate the relationship. We therefore extend the study to a larger region over the West Pacific in Sect. 4.3.

Relationship between SO 2 -OH reaction and sulfate concentration over the West Pacific 230
Considering aircraft speed, here a 15-second average dataset (instead of a 1-minute average in the previous analysis) is utilized to improve the spatial resolution. In addition, the data is classified according to their longitude and latitude; the data located within a 0.25° x 0.25° gridded region is used to calculate the f SO2-OH * . The contribution of SO 2 -OH reaction is evaluated at different altitudes and regions, including Taiwan, East China Sea, and South Japan Sea, as presented in Fig. 4. A significant correlation between sulfate and f SO2-OH * is found in various regions and altitudes, and the information is presented in Table 2. 235 Although a strong correlation exists at different altitudes, no significant trends exist between altitudes and other variables. We expect the f SO2-OH * to decrease with height as trace gases' concentrations are generally higher near to the surface. However, the calculated maximum f SO2-OH * does not decrease with altitude. This discrepancy may result from using measurements from different locations and the existence of inversion layers, which influences the distribution of trace gases. It is noteworthy that, 240 a special multi-altitudes (600 m, 900m, 1200 m) flight mission conducted over western Taiwan on April 3 rd revealed the largest value of slope of sulfate vs. f SO2-OH * at 900 m (shown in Fig. 4f). This phenomenon may be explained by the accumulation of trace gases below the ceiling of the boundary layer near 900 m.

OH-oxidation explicable sulfate percentage
According to our hypothesis, the intercept of the plot of sulfate vs. f SO2-OH * may approximate the amount of sulfate generated 245 by mechanisms other than the SO 2 -OH reaction. This includes SO 2 aqueous-phase oxidation, primary source emission, and transport. For environmental policy issues, it is valuable and necessary to separate the contribution of SO 2 gas-phase oxidation from the other sulfate sources. As a result, the SO 2 gas-phase-produced sulfate (SGS) percentage is defined in Eq. (15), enabling us to investigate the extent of SO 2 gas-phase oxidation in different areas.
SO 2 gas-phase-produced sulfate (SGS) percentage = maximum sulfate in target area -intercept maximum sulfate in target area × 100% (15) 250 Table 2 presents the SGS percentage in selected periods at the surface and selected altitudes/areas. The results show that SGS percentages are approximately 30% at the surface and at 600 m over a period of time. This result is comparable with the model simulation by Harris et al. (2013) that 76±7% of SO 2 undergoes aqueous-phase oxidation, and the rest of SO 2 is oxidized by OH radicals in the gas phase. However, for a daily analysis over a specific region, the SGS percentages for data at different altitudes vary and may be much higher than 30% (see Table 2). These high SGS cases could be related to the less amount of 255 transported sulfate. For example, the SGS of Fig. 4d is much higher than that of Fig. 4b, although they were both measured at 300 m. The backward trajectories using HYSPLIT provided by NOAA ARL (Stein et al., 2015;Rolph et al., 2017) in Fig S7ab show that the air masses of Fig. 4a-b passed through East China and loaded with abundant SO 2 and primary sulfate, while air parcels of the case of Fig. 4c-d went across the Sea of Japan, a relatively clean area. Less transported sulfate in the air parcels could contribute to higher SGS. The high SGS at 700 and 900 m over Taiwan may also connect with low long-range-260 transported sulfate (see Fig. S7c-d), and thus the importance of the SO 2 -OH gas-phase oxidation increase in these cases. Therefore, the amount of primary sulfate could fluctuate and impact the percentage of SO 2 gas-phase-produced sulfate.

Characteristics of trace gases and SO 2 oxidation at the surface and as a function of altitude 270
The characteristics of trace gases provide additional information to investigate the possible sources of pollution plumes. For example, CO-NO x at the surface and CO-NO y in the air have similar correlations. As shown in Fig. 5a, the correlations between CO and NO x at the surface, as well as CO and NO y in the air, are similar. The slopes of CO-NO x at the surface and CO-NO y in the air are 0.057±0.0003 and 0.050±0.002, respectively. Though NO x does not wholly represent NO y , we infer that NO x takes a significant part of NO y . On the other hand, as demonstrated in Fig. 5b-c, discernible disparities exist in SO 2 -CO and 275 SO 2 -NO x /SO 2 -NO y airborne to surface relationships. According to the scatter plots, the SO 2 -CO relationship can be generally subdivided into two categories: high SO 2 to CO ratio and low SO 2 to CO ratio, and so does the SO 2 -NO x /SO 2 -NO y relationship.
These two apparent features are explained by different emission types: traffic sources and industrial sources. Conforming to the Taiwan Emission Data System version 10.1 (TEDS 10.1, 2020, as listed in Table S3), the annual SO 2 to CO emission ratio of traffic sources (line sources in the TEDS) is estimated to be 4.1x10 -4 , and that of industrial sources lie between 0.3 to 20, 280 which is much higher than the ratio of traffic sources. Similar to the annual SO 2 to CO emission ratio, the annual SO 2 to NO x emission ratios of traffic sources and industrial sources are assessed to be 7.7x10 -4 and 0.37 to 0.79, respectively. According to Fig. 5b-c, airborne trace gases have high SO 2 to CO ratios and high SO 2 to NO y ratios. These relationships imply that industrial sources may significantly contribute more to the airborne trace gases over Taiwan than traffic sources during 285 the spring and the intermonsoon weather. The industrial-produced plumes are transported to the upper air more quickly through the tall stacks of these sources and the thermal driving force than traffic emissions released close to the surface. Here, we classify the EPA surface stations into five regions defined in Section 4.1 and calculate the average f SO2-OH * for each region. As shown in Fig. 6, the orders of f SO2-OH * at the surface and in the air are B=C>E>D>A and E>B>D>C>A, respectively. Region B has high rate fractions at the surface and in the air since a massive coal-fired power plant and vast industrial areas stand in 290 region B. The distribution of f SO2-OH * coincides with the locations of massive power plants and industrial areas, indicating that industrial sources significantly influence f SO2-OH * and the sulfate concentrations. Nonetheless, region E has the highest f SO2-OH * in the air but is the third highest at the surface. In addition to a coal-fired power plant and industrial areas located in region E, the other reason for the observed behavior could be the strong prevailing north wind, which transports SO 2 from upwind sources to southern areas and leads to higher sulfate concentration in region E (see Fig. S8 for average wind speeds in each 295 region). Nevertheless, limited measurements from surface stations and in the air might not represent the whole region. More observations are necessary to verify our arguments.

Conclusions
Our study investigates and presents an approach to evaluate the contribution of SO 2 -OH gas-phase reaction to the total sulfate.
We use knowledge of the reactions between OH and the major trace gases (SO 2 , CO, NO, NO 2 ) to estimate the fraction of OH 300 reactivity due to SO 2 -OH reaction, f SO2-OH * . The surface observations over western Taiwan and airborne EMeRGe-Asia field measurements over the West Pacific are analyzed. This study shows that the value of f SO2-OH * has a significant correlation with the sulfate concentration at the surface and 600 m altitude above Taiwan. This correlation also appears in different areas and altitudes, though more observational data are needed to confirm the significance. Based on the relationship between sulfate and f SO2-OH * , it is concluded that approximately 30% of sulfate at the surface and an altitude of 600 m is produced by SO 2 -OH 305 reaction, which could vary significantly with the influences of primary emission or long-range transported sulfate. Meanwhile, the observations indicate that SO 2 and sulfate from SO 2 -OH reaction in the airspace of western Taiwan