Vertical Profile and Spatial Distribution of Ozone and Its Precursors at the Inland and Offshore of an Industrial City

This study investigated the ozone formation mechanism and air mass trajectory via simultaneous air quality sampling around the coastal region of urban Kaohsiung. Vertical concentration profiles of O3 and its precursors (NOx and VOCs) were sampled and measured at inland and offshore sites during eight intensive sampling periods. The intensive sampling periods were divided into three categories based on meteorological condition: the sea-land breeze period, the northeastern monsoon period, and the mixed period. Vertical profile results showed that the stratification of O3 was commonly observed at 40 out of 64 sampling sites accounting for 62.5% of the total O3 measurement. The results obtained from VOCs measurement indicated that the major species of VOCs was acetone, which accounted for 16.25–64.05% of total TVOCsC2 in the offshore region, while the major species of VOCs in the inland region was toluene, which accounted for 6.41– 43.77 % of total TVOCs-C2. Backward trajectories showed that air pollutants emitted from land sources could transport to the offshore region, resulting in a high concentration of oversea NOx and VOCs. Major species of VOCs with high O3 formation potential were found to be aromatics in the low atmosphere around the coastal region of metro Kaohsiung.


INTRODUCTION
Kaohsiung-Pindong air quality zone (KP air quality zone), covering Kaohsiung City and Pindong County, has been officially labeled by Taiwan's Environmental Protection Administration (TEPA) as the worst air quality region (6-8% of poor air quality with PSI > 100) among seven Air Quality Zones (AQZ) in Taiwan (Tsai et al., 2010).Moreover, major ambient air pollutants in metro Kaohsiung are mostly higher than that those at other AQZs in Taiwan (Tsai et al., 2008).Among them, O 3 and PM 10 are two major air pollutants that are responsible for poor air quality in metro city (Cheng et al., 2010;Chakraborty and Gupta, 2010;Huang et al., 2010;Lee et al., 2010;Reddy et al., 2011;Stone et al., 2011).Previous investigators reported that high O 3 concentration formation had a relationship with nitrogen oxides and hydrocarbon emissions, and under certain weather conditions (eg.high temperature, low wind speed, high radiation, and low humidity), the related photochemical reactions become more vigorous, causing O 3 to accumulate in the low atmosphere (Ball and Bernard, 1978;Comrie and Yarnal, 1992;Han et al., 2011).
Ozone is considered as the principal product of tropospheric photochemical chemistry.High O 3 concentration is a major environmental concern due to its adverse impacts on human health and its key role in the photochemical reaction cycles affecting the formation and fate of its precursors (NO x and VOCs).In other words, it is of importance to reduce the ambient concentrations of NO x and VOCs in order to decrease O 3 concentration.There is a non-linear relationship between the concentrations of these two species and the formation concentration of O 3 (Na et al., 2005), and the ratios of [TVOCs-C 2 ]/[NO x ] in the offshore regions near Kaohsiung are high, indicating that O 3 formation in these regions is NO x -limited.Thus, NO x emission must be controlled to prevent the formation of O 3 .However, the ratios of [TVOCs-C 2 ]/[NO x ] in the inland regions are relatively lower, the ratio of some cases were even below 4, showing that O 3 formation is VOCs-limited.Consequently, VOCs emission must be controlled in the inland regions to prevent the formation of O 3 .In this work, TVOCs-C 2 is defined as a total of 101 species of volatile organic compounds (TVOCs) with the exception of ethylene, acetylene, and ethane (C 2 ) species.Generally, C 2 dominated TVOC concentration, which might cause misjudgments for the qualitative analysis of GC/MS spectrum (Jenkin and Clemitshaw, 2000).
Sea-land breezes (SLBs) play an important role in transporting air pollutants to and from urban areas on the coast (Tsai et al., 2008(Tsai et al., , 2010(Tsai et al., , 2011)).Previous studies have investigated the effects of SLBs on the spatial distribution and transport of ambient air pollutants, particularly for O 3 episodes (Nester 1995;Venkatesan et al., 2002;Evtyugina 2006;Tsai et al., 2008).Many studies indicate that SLBs may transport air pollutants from coastal polluted areas to inland and spread O 3 to an onshore distance of 20-60 km (Kitada and Kitagawa, 1990;Seinfeld and Pandis, 1998;Hastie et al., 1999).Cheng (2002) examined the impact of SLBs on the horizontal distribution of O 3 concentration in the coastal region of Central Taiwan.He concluded that the weak southeastward breeze is the dominant factor affecting the occurrence of O 3 episodes.Moreover, Liu et al. (1994) found high O 3 episodes occur at the downwind of sea breeze in Taipei and Kaohsiung.
Previous investigators reported that, in recent decades, towers, tethered balloons, aircraft, and lidar have all been applied to measure vertical profiles (Chen et al., 2002).In this study, the vertical concentration profiles of O 3 and its precursors (NO x and VOCs) at eight sites were sampled by tethered balloons with unreactive air sampling diaphragm pumps and black tedlar bags.This method has been used to investigate the vertical profile and the tempospatial distribution of ozone and its precursors in the low atmosphere.This study further investigated the ozone formation mechanism and air mass trajectory via simultaneous air quality sampling over the coastal region of metro Kaohsiung.This work presents the vertical distributions of ozone concentrations and its precursors, as measured via tethered balloons and sampling pumps at Kaohsiung City.

EXPERIMENTAL METHODS
Ozone and its precursors were simultaneously sampled both inland and offshore during eight intensive sampling periods on August 16-18 and November 2-4, 2006;January 24-26, March 6-8, and May 2-4, 2007;October 30-November 1, 2008;and March 11-13 and July 15-17, 2009.Sampling of inland O 3 and its precursors was conducted at sixteen sites located in southern Taiwan, including ten Taiwan Air Quality Monitoring Network (TAQMN) sites and six other sites established for this particular study.The six latter sites for intensive sampling were located at National Kaohsiung University (NKU), National Kaohsiung First University of Science & Technology (NKFUST), Zuoying Junior High School (ZYJ), Chisian Junior High School (CSJ), Jhongyun Junior High School (JYJ), and Fooyin College (FYC).At each inland sampling site, a mobile air quality monitoring vehicle with an O 3 analyzer (ML 9812) and a NO x analyzer (DANI 200A) was used to simultaneously monitor O 3 and NO x , as shown in Table 1.After sampling, O 3 and NO x in the tedlar bag were analyzed with on-line instruments immediately, and volatile organic compounds (VOCs) were sampled with a stainless steel canister and further analyzed with a gas chromatography-mass spectrometry (HP 6890N GC/MS) and a thermal-desorption unit (ENTECH 7100A Preconcentrator).A 0.25 mm Chrompack DB-1 60 m-length capillary column was employed for qualitative analysis.Spectrum gases for calibration curves in quantities follow the standard gas of NIEA A715.14B (i.e.US EPA Method TO-15).
Sampling of offshore O 3 and its precursors were conducted at two offshore locations, including Hsiau-Liu-Chiou (HLC), an island approximately 10 km away from the southwestern coast of Kaohsiung City, and a mobile air quality monitoring boat navigated to positions approximately 5-12 km away from the coastline of Kaohsiung City.The locations of inland and offshore sites for sampling O 3 and its precursors around the coastal region of Kaohsiung City are shown in Fig. 1.During each intensive sampling period of 48 h, sampling of O 3 and its precursors were conducted simultaneously at both offshore and inland sites.Offshore O 3 and its precursors were sampled at the top floor on the front deck of the boat to prevent the interferences from the exhaust tail gases of the boat and from oceanic spray.Like the inland sampling, offshore O 3 and its precursors were also measured with O 3 and NO x analyzers, and VOCs was preserved by stainless steel canister.
During the intensive sampling periods, O 3 and NO x concentrations were measured over a 48 h period, while VOCs were sampled using two black tedlar bags (SKC, 233-08B) at 3 h intervals.This bag was frequently used for collecting and protecting light-sensitive compounds to avoid the requirement for cumbersome outer containers.In addition, the analytical method for the US EPA Method TO-15 used in this work involves using a high resolution GC/MS (HP, 6890N) at each sampling site.The meteorological data was also monitored over a 48 h period to obtain the hourly average value for the ground level time-intensive sampling data.The vertical profiles of O 3 , NO x , and VOCs were also simultaneously measured in the atmosphere.During the  intensive sampling periods, vertical O 3 and NO x concentration profiles were recorded at 3 h intervals-8:00 am, 11:00 am, 2:00 pm, 5:00 pm, 8:00 pm, 11:00 pm, 2:00 am, and 5:00 am Local Time (LT), starting at 8:00 am.Among them, the first to fifth samples were taken at an altitude of 100 m above ground from 2006 to 2007, while the last three samples were taken at an altitude of 500 m at NKU, NKFUST, FYC, ZYJ, and HLC, respectively.In addition, a mobile air quality monitoring boat collected a total of 34 sampling data sets of O 3 and NO x at an altitude of 200 m from 2008 to 2009.A total of 16 sampling data sets of VOCs at the altitudes of 0, 100, 300, and 500 m above ground were collected every 6 hours for a 48 h sampling starting at 8:00 am Local Time (LT) at FYC, JYJ, and HLC, respectively.This study used a tethered balloon technique to acquire the vertical and spatial distributions of O 3 and its precursors at the altitudes of 0 m, 100 m, 200 m, 300 m, and 500 m above ground (Cheng, 2002).The vertical profile measurements of O 3 and its precursors given here were obtained by using a tethered balloon with the attachment of electric winch, ropes, air pumps (SKC 224-PCXR miniature diaphragm pumps for personal or environmental sampling of gases and vapors in air), and tedlar bags, which could reach an altitude up to 500 m.A specially designed lightweight (936 g) air sampling pump with a timer was utilized to inflate two tedlar bags.After sampling, the winch retracted the tether line to retrieve the tedlar bags.The time setting was different depending on the sampling heights.In general, the higher the balloon was raised, the longer the delay time was set before sampling began.For example, the delay times were set as 4, 12 and 20 min, when the sampling points were set at 100, 300, and 500 m above ground, respectively.When the balloon reached its proper height, the timercontrolled sampling pump was automatically turned on and provided a constant flow rate to the tedlar bags for 6 min, and the sampling pump was then automatically shut down and the unit lowered down the tethered balloon by using the electric winch.Two tedlar bags were then removed.Tetherd balloons have been widely applied to suck air samples above ground into two tedlar bags by an air pump (Cheng, 2002;Wu et al., 2010;Sangiorgi et al., 2011).After sampling, one tedlar bag was quickly connected to the air inlet of an air quality monitoring automobile for measuring O 3 and NO x concentrations, respectively.Measurement of O 3 and NO x for each bag took approximately 2 min.To minimize sampling artifacts by cross contamination, the pump was equipped with a diaphragm membrane.Another tedlar bag was immediately transferred to a stainless steel canister and then brought back to the laboratory for further analysis of VOCs within one week.Moreover, each tedlar bag was covered with a light-shielding bag to avoid photochemical reactions occurred inside the tedlar bag.The device used for the vertical profile sampling is shown in Fig. 2.
Backward trajectory analysis were simulated with an Air Quality Trajectory Model (version 1.1) developed by Taiwan EPA (Chang et al., 1983).The assimilation method incorporates the Barnes objective method to interpolate spatial values and the variation-kinematical model was adopted to correct the effects of complex terrain and to produce hourly wind field data using 24 ground surface stations spread over southern Taiwan (Barnes 1973;Chang et al., 1983).These included two meteorological stations from the Central Weather Bureau of Taiwan (CWBT) and 22 ambient air quality monitoring stations from TAQMN.By utilizing the hourly wind fields over the ground surface, backward trajectories of air parcels were simulated for 10 h at seven selected air quality monitoring stations.The trajectories were simulated by using the horizontal wind components only, with a segment time resolution of 1 h, and the interpolation was linear in both time and space domain.The initial time of backward trajectory for each monitoring station was determined whenas the maximum hourly O 3 concentrations were observed.

Meteorological Measurement Data
Taiwan is geologically located in the East Asian subtropical zone, and is significantly influenced by the northeastern monsoon (NEM) blown from late fall to late spring and by the southwestern monsoon blown from summer to mid-fall.A sea breeze is superimposed on these monsoons when meteorological conditions are suitable to produce daytime northwesterly or southwesterly flow.This study collected the meteorological data from ambient air quality monitoring stations operated by TAQMN and CWBT in the KP air quality zone during eight intensive sampling periods from August 2006 to July 2009.Wind speed and wind direction data shown in Table 2 were used to characterize the intensive sampling periods into three categories: the sea-land breeze (SLB) period (the first, fifth, and eighth sampling periods), the northeastern monsoon (NEM) period (the third, fourth, and seventh sampling periods), and the mixed (MIX) period (the second and sixth sampling periods).During the SLB periods, the wind direction changed for at least 90° between daytime (daytime sampling was conducted at 8:00 am, 11:00 am, 2:00 pm, and 5:00 pm) and nighttime (nighttime sampling was conducted at 8:00 pm, 11:00 pm, and 2:00 am, and 5:00 am), and the wind speeds of sea breezes varied much more significantly than those of land breezes.There was an obvious interaction between sea and land breezes, when the wind direction transformed a land breeze to a sea breeze at 9:00-10:00 am, and a sea breeze to a land breeze at 9:00-11:00 pm, and the wind speed also varied significantly during the interaction periods.During the NEM periods, the prevailing wind blew from the north (300-60°) with the average speeds of 1-4 m/s.During the MIX periods, the wind direction varied significantly from 270° to 90° with the average wind speeds of 1-3 m/s.In addition, previous investigators reported that the sea breeze speed is higher than land breeze speed, since sea regions have less friction than land ones (Ozoe et al., 1983).

Vertical Distribution of O 3 Concentration
The spatial distribution and temporal variation of O 3 concentration were monitored over the coastal region of metro Kaohsiung during the SLB, NEM, and MIX periods.A single modal variation of O 3 was observed over the coastal region.During the SLB periods, O 3 appeared after sunrise and its level at the inland sites was relatively lower than the offshore sites in the early morning.Photochemical reaction forming O 3 in the atmosphere resulted in the highest O 3 concentration mostly at noon at the inland sites.The highest O 3 concentration of approximately 150 ppb at the inland sites was observed on May 2-4, 2007.At nighttime, O 3 and its precursors accumulated in the coastal region had a tendency to transport from the land to the sea due to land breezes, which resulted in relatively higher O 3 concentration at the offshore sites than that at the inland sites.O 3 was formed mainly in the daytime and then accumulated at nighttime.During the MIX periods, field measurement results showed that the highest O 3 concentrations were observed at the inland sites in the daytime and at the offshore sites at nighttime.The highest O 3 concentration of approximately 130 ppb was observed at the inland sites at 14:00 on October 31, 2007.During the NEM periods, O 3 concentrations with an average of 48.54 ppb at the inland sites were mostly higher than those with an average of 32.17 ppb at the offshore sites.On the contrary, high O 3 concentrations observed at inland during the NEM periods were mainly influenced by the northerly winds, which transported O 3 from the northern region to metro Kaohsiung.
Field measurement results also indicated that the variation trend of O 3 concentration at the ground level was similar to that at 100-500 m above the ground, although O 3 concentration at 100 m was higher than that at the ground level over the coastal region.However, no matter in the daytime or at nighttime, the average O 3 concentration at the ground level was at least 3 times higher than that above the ground (100 m) at four sampling sites on March 6-8, 2007.The results implied that O 3 could be formed and accumulated at 100 m over the coastal region.It suggested that the sunshine at 100 m was somehow stronger than that at the ground level, which influenced the lifetime of ozone formation/sink and the time constant of mixing.Additionally, NO x emitted from high chimney could titrate with O 3 and reduce its concentration above 100 m, resulting in a maximum O 3 concentration at 100 m above ground (Wu et al., 2010).Further investigation on O 3 accumulation showed that O 3 concentration at 500 m was 40.45 ppb at the offshore HLC site during the secondary intensive sampling period.At nighttime, O 3 concentration at the ground level and above the ground was relatively high at the offshore sites.Vertical profiles of O 3 concentration at inland and offshore sites are illustrated in Fig. 3. On October 30, a high O 3 concentration was observed at 2:00 pm at the offshore site HCL.Ozone concentration at 100-500 m was higher than the ground O 3 level, with a core up to 85 ppb at 300-500 m, contributing to serious O 3 episodes at the downwind region of metro Kaohsiung.At nighttime, O 3 concentration tended to decrease at the coastal region due to land breezes.A backward trajectory model was used to determine the transportation routes of O 3 before they arrived at the sampling locations.

Stratification Phenomenon of O 3 and its Precursors (NO x and VOCs)
In this study, TVOCs-C 2 is defined as a total of 101 species of volatile organic compounds (TVOCs) with the exception of ethylene, acetylene, and ethane (C 2 ) species.Generally, C 2 dominated TVOCs concentration, which may cause misjudgments for data analysis.In this study, stratification factor was defined as the ratio of the concentrations of air pollutants sampled at 100, 300, and 500 m above ground to those sampled on the ground.The data shown in Table 3 are the mean value of stratification factor in the daytime and at nighttime.The turn-over of TVOC-C 2 concentrations were observed for different altitudes (100, 300, and 500 m above ground) at 19 sampling sites, accounting for 63% of total 30 sampling sites in the daytime.The results showed that the stratification phenomenon occurred frequently at both inland and offshore sites.At nighttime, the turn-over of TVOC-C 2 concentrations for different altitudes were measured at 15 sampling sites, accounting for 50% of total 30 sampling sites, and the results showed that the stratification phenomenon occurred mainly at inland sites (a total of 10 sites).The turn-over of NO x concentrations at the altitudes of 100, 300, and 500 m above ground had 17 sampling sites, accounting for 53% of total 32 sampling sites of the daytime, and the stratification phenomenon occurred at inland and offshore sites; at nighttime, the turn-over of NO x concentrations at different altitudes were measured at 13 sampling sites, accounting for 41% of total 32 sampling sites, and the stratification phenomenon occurred mainly at offshore sites (a total of 11 sites).Overall, during the eight intensive sampling periods, the stratification of O 3 concentration occurred at 40 out of 64 sampling sites, in which its precursors (NO x and VOCs) demonstrated their stratification phenomenon at 30 sampling sites, accounting for 75% of total O 3 stratification.This phenomenon is mainly attributed to the facts that air pollutants can be accumulated in the mixing layer over the coastal region (Lin et al., 2007).

Major Species of Volatile Organic Compounds for Different Wind Field Categories
In this study, 46 species of TVOCs-C 2 were divided into four categories: alkanes, alkenes, aromatics, and carbonyls, as illustrated in Fig. 4.During the SLB periods (Fig. 4 2. b CSJ site was chosen for sampling periods #1-#5 and JYJ site was chosen for sampling periods #6-#8.3. Stratification factor is defined as the ratio of the concentrations of air pollutants sampled at 100, 300, 500 m above ground to those sampled on ground level.The data shown in Table 3 are the mean value of stratification factor in the daytime and at nighttime.4. "-" means no sample available; "< 1" means no stratification; "> 1" means stratification occurrence. 5. Daytime sampling was conducted at 8:00 am, 11:00 am, 2:00 pm, while 5:00 pm, while nighttime sampling was conducted at 8:00 pm, 11:00 pm, 2:00 am, and 5:00 am. μg/m 3 , respectively.A significant vertical variation of TVOCs-C 2 ratio was observed from ground level to a high altitude in the offshore region.The major species of TVOCs-C2 were alkanes and carbonyls, which accounted for 69.83% and 74.01% of TVOCs-C 2 at ground level in the daytime and at nighttime, respectively.The major species of TVOCs-C 2 were aromatics, which accounted for 49.26% and 35.71% of TVOCs-C 2 at 100-200 m above ground in the daytime and at nighttime, respectively.The major species of TVOCs-C 2 were carbonyls, which accounted for 65.74% and 40.33% of TVOCs-C 2 at 300-500 m above ground in the daytime and at nighttime, respectively.The major species of TVOCs-C 2 were alkanes, aromatics, and carbonyls in the inland region.In general, for petrochemical processing and storage tank operations, the major species of TVOCs-C 2 are alkanes, alkenes, and aromatics, while of the major species of TVOCs-C 2 for mobile sources are alkanes and carbonyls.At the offshore region, boat exhaust is the major VOC emission source (Morikawa et al., 1998).In addition, boats coated with painting can emit VOCs to the air (Malherbe et al., 2007).It concluded that TVOCs-C 2 were mainly emitted from industrial sources and automobiles at the inland region, while boat exhaust and painting were the major sources for TVOCs-C 2 at the offshore region.

Analysis of Air Mass Aging
Previous investigators reported on the transmission processes of VOCs at offshore and inland regions in order to determine O 3 episodes derived from VOCs and NO x sources (Hsieh and Tsai, 2003;Wang et al., 2008).This study used the life cycles of VOCs species to estimate air mass aging and photochemical reactivity.Air mass aging analysis was conducted by using the m,p-xylene/ethylbenzene (m,p-X/E) ratio.In general, the air mass is fresher as the m,p-X/E ratio is larger (Hsieh and Tsai, 2003;Wang et al., 2008).According to the photochemical reactivities, the reaction rate of m,p-xylene was faster, while that of ethylbenzene was slower.
Table 4 lists the m,p-X/E ratio for three wind field categories.During the SLB periods, the sampling altitudes were divided into 0-100 m, 100-200 m, and 300-500 m above ground, and the m,p-X/E ratios in the offshore region were 2.12 ± 0.32, 1.73 ± 0.37, and 1.64 ± 0.21, respectively, while the m,p-X/E ratios in the inland region were 3.52 ± 0.14, 3.64 ± 0.11, and 3.33 ± 0.32, respectively.The results showed that the m,p-X/E ratios in the offshore region were lower than those in the inland region, indicating that the air mass in the offshore region was older than that in the inland region.Moreover, as the altitude reached 300-500 m, the air mass showed an aging trend.During the NEM periods, the m,p-X/E ratios in the offshore region were 2.05 ± 0.11, 2.33 ± 0.52, and 1.82 ± 0.17, respectively.Compared to two other periods, the m,p-X/E ratios at different heights during the NEM period did not vary much probably due to relatively stable atmosphere.In the inland region, the m,p-X/E ratios were 3.55 ± 0.25, 3.22 ± 0.18, and 3.43 ± 0.26, respectively, and thus the m,p-X/E ratios did not vary much.While the altitude reached 300-500 m, the air mass showed an aging trend in the offshore region.The air mass in the offshore region was generally older than that in the inland region due to the transport of air from the land to the sea, similar to the SLB periods.During the MIX periods, the m,p-X/E ratios in the offshore region were 3.43 ± 0.26, 2.73 ± 0.15, and 2.44 ± 0.33, respectively.In the inland region, the Table 4.The concentration ratio of m,p-Xylene and Ethylene (m,p-X/E ratio) at different sampling altitudes for three wind field categories.
The [TVOCs-C 2 ]/[NO x ] ratios at different sampling heights for three wind field categories are shown in Table 5.In this study, the [TVOCs-C 2 ]/[NO x ] ratio influenced the ground and upper levels of O 3 formed by the photochemical reactions between NO x and VOCs (Blanchard and Fairley, 2001).The [TVOCs-C 2 ]/[NO x ] ratio at offshore were higher than 8, or even higher than 15.It was showed more based on NO x -limited, thus reducing O 3 for the generation must be controlled of NO x reduction.However, the [TVOCs-C 2 ]/[NO x ] ratio at inland were lower than 8, or even lower than 4, showing that O 3 was VOCs-limited, which addressed the reduction of VOCs concentration, could effectively reduce O 3 concentration.Previous study reported that O 3 formation is mostly VOCs-limited in the vicinity of emission sources, while O 3 formation is mostly NO x -limited for other places.It is attributed to the fact that the photochemical reaction of NO x has relatively short life cycle (3 to 6 h) and VOCs in the photochemical reaction cycle need 24 h or more.As a result, the [VOCs]/[NO x ] ratio would be larger.The results suggested that air mass at the altitude of 300-500 m in the offshore region were NO x -limited during the NEM and MIX periods.High-altitude was mainly affected by longrange transport air mass blown from the central and northern parts of Taiwan due to northeastern monsoon.For other two periods, ground to the altitude O 3 was generated by the impact of both VOCs and NO x .The results showed that the offshore sites were also influenced by the emissions from inland sources.During the NEM periods, the O 3 formed in the inland region was VOCs-limited, showing that local emission sources were superior to long-range transportation.SLB periods 0-100 m 8.12 ± 0.21 6.92 ± 0.12 100-300 m 8.41 ± 0.11 8.58 ± 0.11 300-500 m 9.43 ± 0.17 6.14 ± 0.24 NEM periods 0-100 m 6.81 ± 0.36 3.73 ± 0.18 100-300 m 9.32 ± 0.25 6.41 ± 0.22 300-500 m 15.05 ± 0.44 8.64 ± 0.15 MIX periods 0-100 m 6.62 ± 0.32 8.15 ± 0.12 100-300 m 5.97 ± 0.14 6.61 ± 0.13 300-500 m 16.85 ± 0.47 9.36 ± 0.25

Backward Trajectory Analysis
A backward trajectory model was used to determine the transportation routes of O 3 before it arrived at the monitoring locations.Backward trajectory modeling involves tracing the transportation of air parcels through the meteorological field.Prior to modeling, the meteorological data (mainly wind velocity and wind direction) for the transport of air parcels was gathered from the Taiwan Environmental Protection Administration (TEPA) and Taiwan Central Weather Bureau (TCWB).The backward trajectories of air parcels transported toward the inland air quality monitoring sites around the coastal region of southern Taiwan during the sampling periods are illustrated in Fig. 5.
During the SLB periods (August 16-18, 2006, May 2-4, 2007, and July 15-17, 2009)  In addition to using the m,p-X/E ratio, the analysis of air mass aging in the backward trajectory analysis also showed air pollution transport from the north caused TVOCs-C 2 concentrations to be higher and the age of air mass to be older in the offshore region during the NEM periods.During the SLB periods, the fresh air mass transport from the offshore to the inland regions by sea breezes in the morning caused TVOCs-C 2 concentrations to be lower in the inland region over the coastal region of metro Kaohsiung.

CONCLUSIONS
This study investigated ozone formation mechanisms and air mass trajectories via simultaneous air quality sampling over the coastal region of metro Kaohsiung.This study examined eight sampling periods, and found that the stratification of O 3 concentration appeared to be 40 out of 64 sampling sites, in which its precursors demonstrated stratification at 30 sampling sites, accounting for 75% of total O 3 stratification.At nighttime, aromatics dominated TVOCs-C 2 in the offshore region, with an average concentration of 139-246% at the daytime, indicating that when a land breeze was blowing at nighttime, the effect of industrial pollution became more important.The major species of TVOCs-C 2 were aromatics, which accounted for 29.79-51.27% of TVOCs-C 2 in the inland region.The TVOCs-C 2 were mainly emitted from industrial sources and automobiles in the inland region, while boat exhaust and painting were the major sources of TVOCs-C 2 in the offshore region.The analysis of air mass aging showed that air mass in the offshore region was generally older than in the inland region, and the ground air mass was fresher than that at 300-500 m above ground, due to pollution from emissions sources at the ground level.In the offshore region, the m,p-X/E ratios in the MIX periods were higher than those in the SLB and NEM periods, showing that the air mass was fresher.Due to unstable atmosphere, the air mass transmission processes was less clear in the inland and offshore regions in metro Kaohsiung, and was also affected by local boat exhausts.Backward trajectory analysis further showed that air pollutants transported from the north caused higher TVOCs-C 2 concentrations and the air mass became older in the offshore region during the NEM periods.During the SLB periods, fresh air mass transported from the offshore region to the inland region by sea breezes in the morning caused lower TVOCs-C 2 concentrations in the inland region.

Fig. 1 .
Fig. 1.Location of inland and offshore sampling sites over the coastal region of metro Kaohsiung.

Fig. 2 .
Fig. 2. Illustration of the tethered balloon sampling and analytical instruments in the field.

Fig. 3 .
Fig. 3. Vertical distribution of O 3 concentration at inland and offshore sites over the coastal region of metro Kaohsiung in the unit of ppb.

Fig. 4 .
Fig. 4. TVOC-C 2 composition ratio of offshore and inland regions in three wind field categories (a) the SLB periods, (b) the NEM periods, and (c) the MIX periods.
(see Figs. 5(a)-(c)), sea breezes blown in the early morning could transport the offshore O 3 back to the inland region in southern Taiwan, which resulted in relatively high O 3 concentrations at the inland sites in the afternoon.During the NEM periods (January 24-26, March 6-8, 2007, and March 11-13, 2009) (see Figs. 5(d)-(f)), O 3 was brought to metro Kaohsiung mainly by northerly winds which transported air mass originating from the northern regions (i.e., Chiayi and Tainan Counties).During the MIX periods (November 6-8, 2006 and October 30-November 1, 2009) (Figs.5(g)-(h)), the wind field was dominated by the northeastern monsoons, and this likely inhibited the influences of sea-land breezes.

Fig. 5 .
Fig. 5. Backward trajectories of air parcels transported toward the inland sites over the coastal region of southern Taiwan during the SLB periods (a)-(c), the NEM periods (d)-(f), and the MIX periods (g)-(h).

Table 1 .
The principles and standard methods of O 3 and NO x analyzer.
a NIEA: National Institute of Environmental Analysis, Taiwan.

Table 2 .
Wind direction, wind speeds, and wind field categories for eight intensive sampling periods.

Table 3 .
Means of stratification of O 3 and its precursors (NO x and VOCs) at inland and offshore sites during eight intensive sampling periods.

Table 5 .
The [TVOCs-C 2 ]/[NO x ] ratio at different sampling altitudes for three wind field categories.