Influences of natural and anthropogenic particles on ambient particulate air quality during typhoon season: From Bashi Channel to Kaoping River Valley

Aeolian dust episode (ADE) is an emerging disaster occurred from the bare lands of the Kaoping River Valley in southern Taiwan because of typhoons. Four manual sampling sites located along the Kaoping River Valley conducted to collect PM10 (aerodynamic diameter ≤ 10 μm) with high‐volume samplers during the ADE occurred by Typhoon Doksuri, and on regular days. Mass percentages of sea‐salt particles (SSs) in PM10 accounted for 5.47–8.91% on regular days and 11.66–14.05% in phase II. Average mass percentage of Ca2+ in phase I increased twice than those on regular days. Cl− deficit percentages were much lower during the ADE (7.37–14.13%) than on regular days (31.69–42.78%), indicating acidic particles mainly produced by chemical reactions of acidic aerosols with aeolian dust and SSs. Even alkaline aeolian dust is a dominant source of the ADE, the atmospheric particles are attributable to acidic particles in the air. Hence, anthropogenic sources play a key role for the worst air quality during typhoon season.


| INTRODUCTION
Aeolian dust episode (ADE) triggered by specific weather conditions of particular landscapes have been emphasized (Park et al., 2004;Lin and Yeh, 2007;Kumar et al., 2012;Tsai et al., 2012;Chen et al., 2016). Aeolian dust (AD) is a sedimentary deposit produced from the finest (silt-sized) fraction of planetary regolith that is carried in suspension and distributed by atmospheric activity (Ruff et al., 2014). Taiwan Environmental Protection Administration (Taiwan EPA, 2008) reported that AD is a main contributor of ADE in the atmosphere at the major rivers in central Taiwan in winter. Most rivers flow westwards or eastwards except Kaoping River Valley because the Central Mountain Range is located in middle Taiwan. Water level of the Kaoping River, the largest watershed (3,257 km 2 ) in Taiwan, drops rapidly following rainy days. Typhoon Morakot hit Taiwan with exceptional rainfall, over 50-year record, and caused severe flooding and landslides in August 2009 (Ge et al., 2010). Huge amounts of fine sands were eroded by floods from the upstream of the Kaoping River and then depositing on the surface at the downstream. ADEs exhibited distinct characteristics with specific weather conditions in southern Taiwan during typhoon season (May-September). More than half of the riverbed can turn into bare lands due to the strong solar radiation, thus becoming a potential source of AD during the typhoon season.
Sea-salt particles (SSs) that originated from the surface of oceans by the bursting of whitecap bubbles are dominant particle matters at islands and along the coasts (Chow et al., 1996;Cohan et al., 2008). They could influence the radiative balance of the atmosphere and cloud formation (Zhuang et al., 1999;Varun Raj et al., 2009). Thus, SSs play an important role in islands' air ambient quality in the atmospheric and oceanic chemistry Park et al., 2015;Chen et al., 2016).
However, few studies focused on the influence of both natural particles (i.e., AD and SSs) and anthropogenic particles during the ADE of the typhoon season in island regions. Accordingly, the present study focused on the water-soluble ionic species (WSIs) characteristics in PM 10 and ascertained how natural and anthropogenic particles influence atmospheric PM 10 from Bashi Channel to Kaoping River Valley during the ADE of Typhoon Doksuri.  Figure 1a. Ambient PM 10 was collected simultaneously in a sampling network on regular days and during the ADE. Regular sampling was conducted to collect 24-hr PM 10 , starting from 0800 LST (local standard time), with highvolume samplers (TE-6070D) at four sampling sites on July 7-14, 2012. The sampling during the ADE was simultaneously conducted in two phases to collect PM 10 with highvolume samplers (TE-6070D) on June 29-30, 2012, while the trajectory of the Typhoon Doksuri passed through the Bashi Channel as shown in Figure 1b. Phase І was conducted from 0100 to 0400 LST on June 29, 2012, whereas phase ІI was carried out from 0400 to 0800 LST of the sequential day on June 29-30, 2012. The sampling flow rate of high-volume samplers was operated at 1.4 m 3 /min based on the standard method of NIEA A102.12A.

| MATERIALS AND METHODS
WSIs were analysed for ambient PM 10 sampled on regular days and during the ADE. All collected PM 10 samples were divided into half pieces. We only need one piece of them for further analysis of WSIs. The remaining one piece was on the purpose for redoing partial experiments if we were in need. One quarter of the quartz fibre filter was added inside a 15-mL bottle made of polyethylene. Each bottle was filled with distilled de-ionized water, and then vibrated ultrasonically for 60 min. The mixed solution obtained from each bottle was filtered to avoid the column damped for extending the life span of the analytical instrument. were measured with an ion chromatography (Dionex, DX-120). The quality assurance and quality control for the analysis of the WSIs were conducted in this study. At least 10% of the samples were analysed by spiking with a known amount of WSIs to determine the recovery rates. The recovery rates varied between 96 and 103%. In addition, duplicate analysis results showed that the relative percentage differences ranged from 3 to 4% for all chemical species. The sampling and analytical procedures were similar to those described in previous studies (Tsai et al., 2010;Li et al., 2015).

| RESULTS AND DISCUSSION
The Taiwan Central Weather Bureau recorded the ADE occurring on June 29, 2012, while Typhoon Doksuri passed along the Bashi Channel with an anticyclone outflow circulation as shown in Figure 1b. A separate outflow caused by the typhoon entered the Kaoping River Valley through its estuary in the south because the Kaoping River flows southwards. The prevailing winds were in the range of 160-200 at the DL station and 180-300 at the MN station. The wind speeds varied in the range of 4.1-7.1 m/s at the DL station and 1.2-5.4 m/s at the MN station from 0800 to 1900 LST. Even the wind speed at the MN station was less than that at the DL station, the wind speeds measured at both stations were over the threshold wind speed (3.05 m/s) for re-suspending dust in the air . According to Figure 1a-c, we concluded that the variation of PM 10 at the DL station was affected by the bare lands located at the estuary of the Kaoping River while the PM 10 concentrations at the MN station was influenced by the bare lands formed along the Kaoping River. This study further summarized the monthly variation of PM 10 from 2007 to 2013, which were recorded by the four Taiwan EPA's air quality monitoring stations located along the Kaoping River Valley as shown in Figure 2a. It indicated that the monthly averaged PM 10 concentration at the DL station was 39.8 AE 6.8 μg/m 3 in the typhoon season (June-October). The 24-hr PM 10 concentration showed significant temporal variation an the average of 414.5 AE 432.4 μg/m 3 , as shown in Figure 2b, exceeding the 24-hr PM 10 ambient air quality standard (125 μg/m 3 ) by 3.3-fold at the DL station on June 29, 2012. Compared to the monthly PM 10 concentration (39.8 AE 6.8 μg/m 3 ) at the DL station on regular days, the average PM 10 concentration was over 10-fold on June 29, 2012. The results evidenced that AD was the major source to deteriorate the ambient air quality along the Kaoping River during the ADE. Site MS1 had the highest hourly average PM 10 concentrations of 677.4 μg/m 3 in phase I as shown in Table 1, indicating that PM 10 concentrations rose as high as 17.8-to 30.6-fold higher than the average PM 10 concentrations on regular days (22.1-37.9 μg/m 3 ). Lower hourly average PM 10 concentrations were 216.1 μg/m 3 at site MS2. Hourly average PM 10 concentrations at sites MS3 and MS4 located at the left bank of the Kaoping River Valley were in the range of 73.4 and 97.6 μg/m 3 . The mass percentages of WSIs to PM 10 in phase I were higher than those on regular days by 1.17-to 1.38-fold as shown in Figure 3, indicating that PM 10 in phase I was rich in moisture content from the Bashi Channel to the inland area of the Kaoping River Valley, causing in high mass percentages of WSIs in PM 10 . Among WSI species, Na + and Cl − commonly recognized as tracers of SSs (Chow et al., 1996). Ca 2+ and K + are related to AD (Chen et al., 2004;Lin et al., 2005;Taiwan EPA, 2008;Tsai et al., 2012), and SO 4 2− , NO 3 − , and NH 4 + are emitted from anthropogenic sources (Chen et al., 2016). Figure 3 indicated that the mass percentages of SSs (i.e., Na + and Cl − ) and AD species (i.e., Ca 2+ and K + ) varied greatly between the periods of regular days and phase І.
The average mass percentages of 2.87 and 3.29% for Na + and Cl − rose more than twice to 4.34 and 6.93% in phase І.
Additionally, AD species of Ca 2+ and Mg 2+ also increased significantly from mass percentages of 4.34-6.93% and 2.18-3.11% in phase І. However, as the Typhoon Doksuri passed through the Bashi Channel in phase ІІ, the mass percentages of Ca 2+ dropped to the range of 2.70-3.71% and far lower than those in phase І, but still relatively higher than that on regular days. Similar to Ca 2+ , the SSs of Na + and Cl − also decreased to 2.06-3.48% and 2.70-5.11%, respectively. These proved that the SSs and AD increased in phase І due to the influence of Typhoon Doksuri to the Kaoping River Valley. However, as Typhoon Doksuri moved away from Taiwan Island, the surface wind speeds in the Valley decreased gradually, the amounts of SSs and AD were then reduced correspondingly. Moreover, a significant diurnal variation was also found for SO 4 2− and NO 3 − in PM 10 during the ADE. In phase І, the average mass percentages were 13.81 and 9.43% for SO 4 2− and NO 3 − and then decreased to 12.13 and 7.67% in phase ІІ, which was possibly related to the formation of secondary inorganic aerosols in the daytime resulting from the heterogeneous reactions of SO 2 and NO x with AD during the transporting process (Varun Raj et al., 2009). The main chemical pathways involve the gas-phase photochemical reactions with solar radiation and oxidants such as O 3 and OHÁ, causing SO 2 and NO x to form sulphate and nitrate correspondingly. High RHs and excess ammonium facilitate the transformation of gaseous SO 2 and NO x in aqueous phase to form sulphate and nitrate (Huang et al., 2014). In this study, meteorological data were monitored at the DL and PT air quality monitoring stations close to the sampling sites. During the ADE periods, the RHs were 75.6 AE 5.99% and 74.1 AE 9.6%, respectively, with no rainfall, indicating that the RHs were relatively high, which were favourable for the formation of WSIs from gaseous precursors. Moreover, ambient RHs were higher than 62% of deliquescence RH for ammonium sulphate and 60% of ammonium chloride (Hu et al., 2011), indicating that sulphate and chloride did exist in the aqueous phase, which favour the absorption of gaseous precursors to the aqueous phase (Sun et al., 2015). The blowing AD was supposed to be a mixture of natural and anthropogenic particles mainly emitted from various sources along the Kaoping River Valley. Tsai et al. (2010) reported that SSs are one of the abundant atmospheric aerosols at coastal regions in southern Taiwan. We estimated the amounts of SSs using Equation (1), which is appropriate for investigating the proportions of SSs in PM 10 (Quinn et al., 2003): Sea salts particles = 1:47 × Na where 1.47 is the mass ratio of (Na + + K + + Mg 2+ + Ca 2+ + SO 4 2− + HCO 3 − )/Na + . Equation (1) revealed that, in  phase I, the mass percentages of SSs to PM 10 concentrations accounting for 11.66-14.05% were obviously higher than those on regular days (5.47-8.91%) and in phase II (5.74-9.89%). The results are summarized in Table 1 that evaluated using Equation (1) was reasonable for explaining that the relatively higher SSs in PM 10 during the ADE. As AD accompanied by typhoon outflow circulation, it could bring huge amounts of SSs from the surface of the Bashi Channel to Kaoping River Valley (Park et al., 2004;Tsai et al., 2010). SSs played as nuclei for the adsorption of sulphur dioxide and the deposition of sulphate, which could strongly enhance the oxidative capability of aerosol particles because of the release of halogen radicals in the marine surroundings (Park et al., 2015). However, Cl − is often partially depleted because of its reactions with acidic compounds such as sulphuric and nitric acids (Tsai et al., 2010). Previous studies (Quinn et al., 2003;Xu et al., 2013;Spada et al., 2015) indicated that the estimated SSs are based on the assumption that Na + and Cl − are mainly derived from seawater and hence excluding the contribution from non-sea-salt ionic species (nss-WSIs) such as K + , Mg 2+ , Ca 2+ , SO 4 2− , and HCO 3 − , thereby allowing Cl loss from SSs through chemical reactions with acidic constituents of ambient particulate matter. The studied area is not far from ocean and the prevailing winds blown from Bashi Channel could enter the Kaoping River Valley from its estuary during the ADE. To reflect the actual ambient conditions, the favourable chemical reactions among different species of the WSIs should be further considered for enhancing the accuracy of estimation processes. The concentrations of nss-K + , nss-Mg 2+ , nss-Ca 2 + , and nss-SO 4 2− could be estimated using Equations (2)-(5) (Park et al., 2004;Kumar et al., 2012), nss-Mg 2 + = Mg 2 + − 0:12 × Na + , nss-Ca 2 + = Ca 2 + −0:038 × Na + , nss-SO 4 2− = SO 4 2− −0:251 × Na + : Zhuang et al. (1999)  ] can thus be determined using Equation (6) (Zhuang et al., 1999;Yao et al., 2003),  Figure 4 demonstrates that the mass fractions of ss-WSIs and nss-WSIs in PM 10 and their spatiotemporal variations estimated on regular days and during the ADE. The major nss-WSIs and ss-WSIs in PM 10 were anthropogenic  The temporal variation of SO 2 and NO x concentrations monitored at the DL and PT air quality monitoring stations on June 29, 2012 during the ADE nss-SO 4 2− to total SO 4 2− (i.e., sum of nss-SO 4 2− and ss-SO 4 2− ) (91.9-93.0%) in phase I were slightly lower to those on the regular days (92.1-93.7%) and in phase II (92.9-95.9%). According to the concentrations of SO 2 and NO x monitored at the DL and PT stations ranged from 2 to 15 ppb and from 5 to 21 ppb, respectively, during the ADE periods as shown in Figure 5, highly nss-SO 4 2− of PM 10 could be supposed to originate from anthropogenic sources near the Kaoping River Valley. High SO 2 and NO x concentrations remained along the Kaoping River Valley during the ADE mainly because the topography of Kaoping River that is located near the foot of the Central Range reduces wind speed as the air passes across the mountains, and consequently causes the accumulation of gaseous precursors. Moreover, Kaoping River valley is close to several industrial complexes in Kaohsiung City, which has major heavy industries, including petroleum refinery, petrochemical industry, iron works, shipbuilding industry, coal-fired power plants, municipal solid waste incinerators, etc. Large amounts of air pollutants could be emitted and transported to the nearby areas, which could cause the ambient air quality much worse than other cities in Taiwan. Therefore, SO 2 and NO x emitted from those major heavy industries can transfer to secondary ions (such as nss-SO 4 2− and nss-NO 3 − ) reacting with the blowing AD and SSs along the Kaoping River Valley during the ADE (Huang et al., 2014). Additionally, the mass percentages of nss-SO 4 2− in PM 10 were always higher than those of the nss-NO 3 − on regular days and during the ADE, because the reaction time for SO 2 to nss-SO 4 2− is shorter than that for NO 2 to nss-NO 3 − (Krolla and Seinfeld, 2008).
AD is generally alkaline and can neutralize acid air pollutants, including gaseous sulphur dioxide and nitric acid vapour, molecules that diffuse rapidly to particle surfaces and retrained through chemical acid-base neutralization reactions. This phenomenon has proved by several studies (Chen et al., 2004;Park et al., 2004;Tsai et al., 2010;, indicating the amounts of crust-related constituents (such as CaSO 4 , Ca[NO 3 ] 2 , and Mg[NO 3 ] 2 ) in particle matter noticeably increased during the ADE since the increased nss-Ca 2+ and nss-Mg 2+ might easily react with nss-SO 4 2− and nss-NO 3 − . Even alkaline AD is dominant source of the ADE, the atmospheric particles are attributable to acidic particles in the atmosphere. Hence, anthropogenic sources play key roles in the worst air quality in the typhoon season. In developing control strategies for the ADE, local governments need strictly regulate industries or anthropogenic activities during the ADE and take cost-effective measures for curbing the bare lands of the estuary before the typhoon season. Chen et al. (2016) reported that the acid-base neutralization reactions were responsible for most of the chloride deficit. The Cl − deficit reactions could reach 100% if enough reaction time and precursors are available. The deficit of Cl − could be determined using Equation (7) where 1.8 × [Na + ] meas is the Cl − concentration estimated from SSs, any loss of Cl − or all Na + in the aerosol particles is of SSs origin, Cl − meas is the Cl − concentration measured in the aerosol particles. Table 1 illustrates the Cl − deficit estimated using Equation (7) in the present study. Compared with the Cl − deficit, their values ranged from 31.69 to 40.38% on regular days, from 7.37 to 14.13% in phase I, and from 9.97 to 27.39% in phase II. Kumar et al. (2012) reported that the deficit of Cl − with respect to Na + was mainly caused by the reaction of Cl − in marine particles with other acidic species (i.e., NO 3 − and SO 4 2− ) to form NaNO 3 and Na 2 SO 4 , respectively, and thus reduced the Cl − concentration in the atmospheric particles (Zhuang et al., 1999;Park et al., 2015). This is one of the reasons why the concentrations of nss-NO 3 − and nss-SO 4 2− in PM 10 during the ADE were higher than those of PM 10 on regular days. Much lower Cl − deficit was observed during the ADE since huge amounts of AD and SSs persistently reacting with acidic aerosols, emitted from the anthropogenic activities. Thus, the Cl − deficit phenomenon could be regarded as valuable references for differentiating the influence of AD and SSs in PM 10 on regular days and during the ADE. Though Cl − in atmospheric particles originated from a variety of anthropogenic sources such as open burning of agricultural waste , Pingtung County Government Environmental Bureau (2013) reported that biomass burning mostly occurs in winter and early spring due to the needs for shifting cultivation. The occurring periods of ADE and biomass burning are obviously different along Kaoping River Valley. Biomass burning should not be an important source of Cl − in PM during the ADE in the typhoon season. Consequently, air flow accompanied by a substantial amount of SSs resulted in a low Cl − deficit for the AD under the effects of strong surface winds exerted by typhoon outflow circulations.

| SUMMARY
A unique ADE occurred in Kaoping River Valley during the period of Typhoon Doskuri was firstly observed in Taiwan. Anthropogenic sources close to Kaoping River still play an important role in the ambient air quality during the ADE as those in regular periods, even though the sea salts and AD were dominant natural sources accompanied by outflow circulation of Typhoon Doksuri entering into the Kaoping River Valley during the ADE. Furthermore, chloride deficit is worthy of being investigated due to the inter-reactions derived by SSs and AD in the Kaoping River Valley in the short-term transporting process during the ADE.