Distribution of Ozone and Related Compounds in the Marine Boundary Layer of the Northern South China Sea in 2010

To investigate the characteristics of air pollutants transported from the Asian continental regions to the marine boundary layer of the northern South China Sea (SCS), we recorded the continuous measurements of meteorology, sea surface radiative budget, and ozone (O3) and related compounds in the marine boundary layer near Taiwan during 2010 cruises. For the marine field campaign investigation, the contaminated O3 and related compounds (e.g., NO2, NO, CO, CH4, and NMHC) have been eliminated from research vessel's exhaust by using various meteorological factors. The mean values of O3 and its major precursors (NO2, NO, and CO) were 25 ± 9.9, 3.2 ± 1.8, 2.9 ± 1.7, and 204 ± 54 ppb, respectively. A high O3 mixing ratio is related to extreme shortwave radiation, high air temperature, less precipitation, and low wind speed and specific humidity. The results indicated a higher ∆O3/∆CO ratio of 0.2 at sunset during the March 2010 cruise mainly due to a long-range transport of aged plumes (> 3 days) originating from the super-region of Asian continent and a lower ∆O3/∆CO ratio of 0.12 at midday, mostly associated with the proximity to local sources of fresh plumes (< 2 days) during April 2010 and July 2010 cruises. The O3 and related compounds mixing ratios over the ocean are affected by emission source, Asian monsoon, wind speed, gas deposition, gas solubility, the chemical enhancement factor, the frontal inversion and the boundary layer height. A clear bulge in the diurnal cycle was observed between early morning and late afternoon for NOx, CO, and O3. Moreover, 5-d backward trajectories and Southeast Asia surface wind fields suggested that the southward export of the air masses in spring originated from Mongolia and the East Asian continent.


INTRODUCTION
O 3 plays a crucial role in transferring atmospheric radiation, atmospheric oxidation, maintaining air quality, and the enhancement of the greenhouse effect (Lelieveld et al., 2004).The major source of O 3 in the troposphere is the photochemical reaction involving an interaction of sunlight with carbon monoxide (CO), nitrogen oxides (NO x ), methane (CH 4 ), nonmethane hydrocarbons (NMHC), and volatile organic compounds (VOCs).High ozone and CO concentrations plus positive correlation of O 3 -CO in springtime indicated that the enhancements of O 3 probably resulted from photochemical O 3 production (Lin et al., 2011).
Because the atmospheric lifetime of O 3 ranges from a few weeks to a few months, O 3 serves as an effective indicator for distinguishing air-mass sources from the Asian continent or the marine boundary layers.Ou-Yang et al. (2013, 2015) indicated that the strong northeasterly winds arising from the winter Asian monsoon may have transported polluted air masses from the northern continent to the northern SCS as indicated by elevated O 3 mixing ratio.
Meteorology plays a crucial role in the formation, dispersion and transport of O 3 and related compounds.Debaje and Kakade (2009) indicated that an abrupt supply of NO x and NMHCs and increased incoming shortwave radiation (SW) result in a high photochemical O 3 mixing ratio.Cheng et al. (2013) observed a strong correlation between a high O 3 concentration and the anticyclone weather type, which is typically associated with clear skies, strong solar insolation, high diurnal temperature ranges, and warm temperature, which are favorable for a high O 3 mixing ratio.
Air samples observed from a ship can be contaminated by pollutants from the ship's stack when the ship remains stationary.A few studies have described several methods to avoid the possible contamination of measurement.Rhoads et al. (1997), Lal et al. (1998), and Stubbins et al. (2006) discarded data to avoid data distortion because of air sample contamination from the ship's exhaust when the ship was stationary.Rhoads et al. (1997), Lal et al. (1998), Burkert et al. (2003), andAlvala et al. (2004) have used a Teflon inlet filter mounted on a prow tower before gas analyzers.Alvala et al. (2004) eliminated the samples that were collected when the winds were directed from the stern to the prow.
The O 3 in the northern SCS is mostly derived through the transport of O 3 and its precursors from the Asian continental regions exhibiting marked anthropogenic activities.To gain additional insight into the spatial patterns of O 3 , 5-d back trajectories reaching the ship of 3 sites with high O 3 concentrations following the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model from the NOAA ARL website were analyzed.Ou-Yang et al. (2013), Sheu et al. (2013), and Wang et al. (2013) reported a similar trajectory result after investigating the origin of air masses to the downwind regions of the northern SCS.
This paper presents the results of the continuous shipbased measurement of O 3 and related compounds in the marine boundary layer of the Taiwan Strait and the northern SCS during 2010 cruises.The objectives of this study were as follows: (1) to determine ocean O 3 and related compounds concentrations excluding possible contamination; (2) to discuss the ∆O 3 /∆CO ratio associated with aged plumes or fresh plumes, and to understand the wind-induced effect on O 3 and the difference of local or super-region O 3 contribution; (3) to examine the association of spatial distributions and diurnal variations of O 3 with various meteorological factors; and (4) to analysis the property of O 3 and CO along the path from HYSPLIT.

Shipboard Site Description
Three research voyages were conducted, first during March 14 to March 20 in 2010 (OR1-921) through the research vessel Ocean Researcher 1 (R/V OR1), second during April 12 to April 15 in 2010, (OR3-1444) and finally, during July 11 to July 16 in 2010, (OR3-1474) through the R/V OR3 over the Taiwan Strait and the northern SCS.The tracks of 3 cruises are presented in Fig. 1.The March 2010 cruise was a part of the 7 Southeast Asian Studies (7-SEAS)/Dongsha campaign; the project was sponsored by National Central University of Taiwan.The detailed background of the 7-SEAS/Dongsha project is well presented by Lin et al. (2013).
The meteorological conditions of 3 campaigns were presented in Fig. 2. The near-surface air temperature (T a ) ranged between 297 K and 304 K and the lowest T a was recorded in the afternoon on March 16 while the cool front was passing during the March 2010 cruise (Fig. 3(b)).The highest temperature was recorded near the Kaohsiung port during the July 2010 cruise, whereas the lowest temperature was recorded during early morning in a rim of the Cigu Lagoon.During the March 2010 cruise, the average T a was 300 ± 0.9 K.The boundary layer height (BLH) along the cruise ranged from 100 m to 872 m according to the ERA-Interim reanalysis data (with a 0.75° grid and 3 hours intervals).Diurnal variations of BLH were observed during the April 2010 and the July 2010 cruise.The surface pressure (SP) was 1015 ± 1.7 hPa, and the sea-surface specific humidity (SH) was 14 ± 2.5 g kg -1 .The average wind speeds (WS) of the 3 cruises were 8 ± 3.7, 5 ± 3.9, and 7 ± 5.4 m s -1 , respectively.Prior to the passing of the cool front (Fig. 3(a)), the lowest WS was recorded at night on March 15 and the BLH was a marked increase (Fig. 2).On sunny days, the SW reached 960 W m -2 during March 2010 and April 2010 cruises, whereas it reached 1000 W m -2 during July 2010 cruise.This value was similar to those recorded on clearsky days (May 2, May 22, and June 29 in 1998) observed at Dungsha Island (20°42′N, 116°43′E) (Lin et al., 2002) during the SCS monsoon experiment (May-June 1998).In addition, numerous studies have reported that the O 3 mixing ratio exhibits a positive correlation with T a and a negative correlation with relative humidity (RH) (David and Nair, 2011;Nair et al., 2011;Toh et al., 2013).A high O 3 mixing ratio is related to extreme SW, less precipitation, low WS and low RH (Pulikesi et al., 2006).

Measurement Techniques and Instrumentation for O 3 , Related Compounds and Meteorology
Ozone and related compounds (NO x , CO, CH 4 , NMHC, and SO 2 ) were continuously measured during 3 cruises; O 3 was measured using an ultraviolet photometric analyzer (Model 45, Thermo Scientific, Franklin, MA, USA), and NO, NO 2 , and NO x were measured using a chemiluminescence analyzer (Model 42, Thermo Scientific, Franklin, MA, USA), consistent with the methodology used by Rhoads et al. (1997) and Torres and Thompson (1993).CO was measured using a gas filter correlation analyzer (Model 48, Thermo Scientific, Franklin, MA, USA), consistent with the methodology used by Rhoads et al. (1997).The CH 4 and NMHC were measured using a flame ionization detector (Model 462, DANI Instruments S.p.A., Contone Switzerland), and SO 2 was measured using a pulsed fluorescence analyzer (Model 43A, Thermo Scientific, Franklin, MA, USA).All of these gases were sampled at 1-min intervals.The detailed descriptions of the instruments are listed in Table 1, including the sample range, response time, detection limit, and precision.
Each analyzer was plugged into a constant voltage regulator.To avoid contamination by sea salt aerosols, the tube inlet was capped using an inverted funnel; a hydrophobic polytetrafluoroethene membrane with a 47-mm diameter and 5.0-μm pore size (LSWP04700, Merck Millipore, Billerica, MA, USA) was used as a filter before each analyzer, and was frequently replaced during the cruise.Sampling was performed using a Teflon tubing, with a funnel collector installed over the roof of the main deck in the front of the vessel, approximately 10 m above sea level (ASL).The sampling location and the filter type were similar to those used by Alvala et al. (2004), Nair et al. (2011), andRhoads et al. (1997).The sample air was passively delivered through the Teflon tube by using a blower motor.Before being directed toward each analyzer, the sample air was mixed in an 8-port octopus manifold.(The residence time was approximate 20 s.) Subsequently, the exhaust gas and mist were actively excluded using a moisture trap.Hatakeyama et al. (1995) also used a gas manifold to achieve the sampled air-tube connection of each instrument.The calibration gases were carried aboard the vessel for multipoint calibration before sailing, and again after returning to the port.The coefficients of determination (r2) during the cruises were > 0.99 for ozone and related compounds instruments.
The meteorological instruments were installed on the foremast of either R/V OR1 or R/V OR3 with T a , SP, WS, SH, SW, and atmospheric longwave radiation (LW).Those instruments included a 3D ultrasonic anemometer (CSAT3, LICOR, Lincoln, Nebraska, USA), infrared (IR) open-path H 2 O/CO 2 gas analyzer (LI-7500, LI-COR, Lincoln, Nebraska, USA), two sets of shortwave radiometers (PSP, Eppley Laboratory, Newport, USA) and two sets of longwave radiometers (PIR, Eppley Laboratory, Newport, USA).The CSAT3 was mounted 10 m ASL, which measured the 3 orthogonal wind components (u, v, and w) at a frequency of 10 Hz.In addition, a gyro-enhanced orientation sensor (3DM-G, MicroStrain, USA) was installed 2 m below the 3D anemometer to measure the ship's rotation angles (roll, pitch, and yaw) at 10 Hz.Using the GPS and gyro data, the measured wind vector was corrected for the ship's movement according to the method described by Edson et al. (1998).

Methodology for Data Screening
The wind direction, R/V position, and directory were constantly monitored to avoid contamination from the R/V exhaust itself (Nair et al., 2011).The criteria for deleting invalid data are presented in the following equation.
where us t is the relative wind component (the threshold of the data discarded as us t > -0.1 m s -1 , u  is the wind vector (m s -1 ), s  is the navigation vector (GPS) (m s -1 ), unit a  is the unit vector for the R/V major axis (m s -1 ), u 1 is According to Eq. ( 1), the relative WS and navigation vectors play vital roles in avoiding any interference from the vessel's smokestack.When the relative wind vector is directed from the prow to the stern, the air sample would not be contaminated by the vessel's smokestack.By contrast, the air sample would be contaminated when the relative wind vector is directed from the stern to the prow.Alvala et al. ( 2004) used a similar concept to eliminate contaminated data based on the wind direction, but it was not an explicit definition.

Wind Fields and Backward Trajectories
In order to investigate how the regional characteristics of O 3 and related compounds related to their potential source and surface air-mass transport paths, this study uses an ERA-Interim reanalysis data (with a 1.5° grid) to represent the Southeast Asia (10°S-40°N, 60°E-150°E) regional surface wind field and 2-m T a on March 18 and July 13, 2010.The ERA-Interim reanalysis was produced with a sequential data assimilation scheme, which was advanced in time using 12-h analysis cycles combined with the available observations (Dee et al., 2011).Backward trajectory analysis was conducted to determine the air-mass transport paths at 2 height levels at 3 high O 3 sites.The 5-d backward trajectories were calculated using the NOAA HYSPLIT model, with the NCEP-GDAS (2006 to present) meteorological data set.The elevations at both sampling sites were calculated every 3 h (Label Interval: 6 h), and were set 10 m, and 1000 m as the ending level height of the trajectory paths.

RESULTS AND DISCUSSION
The surface measurements of O 3 and related compounds were conducted during the 3 campaigns over the Taiwan Strait and the northern SCS in March, April, and July of 2010 on either R/V OR1 or R/V OR3.The mean results and standard deviation for each cruise compared with the other ocean observations are listed in Table 2.The detailed characterization of meteorology over the northern SCS in March 2010 as 7-SEAS/Dongsha experiment has been described by Wang et al. (2011) andYen et al. (2013).

Valid Data Excluding Interference from the R/V Exhaust
The contaminations produced by the vessel itself should be considered when equipment is used for continuous monitoring over the ocean.In this study, most of the chimney plume drifted toward the vessel stern when R/V OR1 was cruising.Rhoads et al. (1997), Lal et al. (1998), and Stubbins et al. (2006) discarded data to avoid data distortion introduced by the ship's exhaust when the ship was stationary.In this study, some no distorted data were   detected during the stationary periods of the ship, depending on the relative WS and direction.Nevertheless, according to Eq. ( 1), the WS, wind direction, gyro vectors, and navigation direction are the major factors for estimating interference from R/V emissions.When the wind approached from the direction of the stern and the wind vector was greater than the navigation vector were determined as contaminated by the smokestack.Outside the Kaohsiung Harbor region, most of the noise in the data of high pollutant concentrations was caused by emissions from the vessel's chimney.

Regional Characteristics of O 3 and Related Compounds in Air Masses
The temporal characteristics of the O 3 , NO 2 , NO, CO, SO 2 , CH 4 , and NMHC mixing ratios and relative wind components during the 2010 cruises are presented in Fig. 4.Over the 3 cruises, the mean values of O 3 , NO 2 , NO, CO, SO 2 , CH 4 , and NMHC were 25 ± 9.9 ppb, 3.2 ± 1.8 ppb, 2.9 ± 1.7 ppb, 204 ± 54 ppb, 1.7 ± 1.0 ppb, 1.7 ± 0.1 ppm, and 0.2 ± 0.2 ppm, respectively.For the March 2010 cruise tracks on March 15 to March 16 in the vicinity of Dongsha Islands, the O 3 mixing ratio (26.5 ppb) was closed to that reported by Wang et al. (2013) (28 ± 5 ppb) at Dongsha Islands during 7-SEAS/Dongsha experiment, and the CO mixing ratio was equal in quantity, and exhibited a tendency to decrease from March 14 to March 16.Taiwan   ([123.0°E, 24.5°N]) observed an accumulation of the CO mixing ratio (more than 400 ppb).This could be because they were located in the downwind region and frontal inversion created by the passing of the cool front on March 16, 2010 (Fig. 3(b)).The extent of the cool front in Fig. 3(b) is decreased and BLH is increased near the Dongsha Islands (Fig. 2) because, as a highpressure area moved from the land to the open ocean, the cool dry air is warmed up.These events are consistent with CO results in the post-frontal stage in Hong Kong on March 16, 2010 and are well captured by the USEPA Community Multi-scale Air Quality model (Wai and Tanner, 2014).The emission source is similar but the increase in CO mixing ratios at site A ([116.4°E, 20.7°N]) on March 16, 2010, is small, compared to that at Kenting and Yonagunijima (Fig. 4) because of the increasing BLH and WS.A majority of the average CH 4 concentrations were slightly higher than those reported by Alvala et al.Here ∆CO and ∆O 3 were defined to be the background concentration subtracted from the measured concentration (the background concentrations for CO and O 3 were estimated to be 100 and 10 ppb).The ∆O 3 /∆CO ratio indicated the extent of atmospheric oxidation during the air-mass transport.A higher ∆O 3 /∆CO ratio of 0.20 during March cruise can be mainly attributed to a longer distance of transport which was a lower value than that of Wang et al. (2013).They found a ∆O 3 /∆CO ratio of 0.23, produced by plumes originating from elevated regions over northern China and Mongolia and passing through the coastal areas of China during 2010 7-SEAS/Dongsha experiment.There is further evidence for long-distance transport on March 18, 2010, where the high O 3 and CO events occurred at sunset, suggesting that these might not be due to local influence of O 3 enhancement with photochemical production.The lower ∆O 3 /∆CO ratios of 0.16 and 0.12 found during the April 2010 and the July 2010 cruises, respectively, might have been due to local influences.Lal et al. (1998) reported that CO exhibited variations that were similar to those exhibited by O 3 in the Indian Ocean, and Rhoads et al. (1997) reported the same phenomenon over the equatorial Indian Ocean.The distributions of the NO x concentrations were similar to those reported by Torres and Thompson (1993), Shon et al. (2001) and Hatakeyama et al. (1995).Torres and Thompson (1993) reported the highest NO concentrations near the equatorial Pacific, where the sea surface can be a weak NO source, but the NO concentrations rapidly decreased between the equator and the 10°N latitude.In this study, a moderate variation was observed in the NO concentrations over the northern SCS, consistent with the observations reported by Torres and Thompson (1993).The O 3 precursor gases (NO 2 and CO) showed high concentrations in regions with high O 3 concentrations (Nair et al., 2011).In 3 cruises, O 3 was mainly derived through the transport of O 3 and its precursors from the Asian continent or the surrounding regions.The scatter plots of NO 2 versus WS with the natural logarithm trend line are presented in Fig. 5(b), and those of CO are presented in Fig. 5(c).Both NO 2 and CO exhibited negative correlations with the WS, and the correlation coefficient of the natural logarithm trend was approximately 0.45.Lower precursor gases are typically produced under higher WS, which reduce the rate of photochemical reaction for O 3 production.The CO mixing ratio at Kenting and Yonagunijima were dramatically increased on March 16.The surface winds on March 16, 2010 from NCEP operational dataset (Fig. 6) indicated the WS at EPA Kenting station was lower than Yonagunijima station, resulting in a higher CO accumulation than at Yonagunijima station under the same frontal inversion and the Asian continent emission.
In addition, the loss of O 3 and related compounds in the marine boundary layer are affected by WS, gas deposition, gas solubility, the chemical enhancement factor (α), frontal movement, the frontal inversion and BLH.Chang et al. (2004) reported a 5-fold increase in O 3 deposition to the sea surface as WS increases from 0 to 20 m s -1 , indicating the importance of WS-induced turbulent gas-transfer.The deposition velocity can be determined as the reciprocal of total resistance, which is the sum of aerodynamic resistance (r a ), gas-phase film resistance (r b ) and aqueous-phase film resistance (r c ) (Lan et al., 2010).The 1/r a and 1/r b as a function of WS-induced friction velocity and 1/r c as a function of WS-induced gas-transfer velocity implied that the loss of O 3 and related compounds increases with WS.SO 2 is a highly soluble gas; O 3 , NO 2 and CO are slightly soluble gases.Lan et al. (2010) indicated the effective Henry constant of slightly alkaline seawater being higher than that of fresh water and the loss of sea surface O 3 and related compounds into ocean being due to increasing α.Chang et al. (2004) showed that deposited O 3 can react with iodide Fig. 6.Surface winds (m s −1 ) on March 16, 2010 from NCEP operational dataset; site A, Dongsha Islands, EPA Kenting station of Taiwan and Yonagunijima, Japan were marked as ▲.
(I -), dimethylsulfide (DMS) and alkenes in seawater, and that its α (5.3) is larger than 1.In addition, prior to the passing of the cool front, the decreased atmosphere stability and increased BLH generated the ventilation effect and diluted gases concentrations on March 15, 2010 (Figs. 2 and Fig. 3).Because of WS, gas solubility, α and increasing BLH, the O 3 mixing ratio was slightly lower than Wang et al. (2013) at Dongsha Islands, Kenting and Yonagunijima station during the March 2010 cruise.
A favorable correlation between the 3 cruises data and EPA Kenting station were presented in Fig. 7.The raw dots (○) denote the original measurements during the cruises.The correct (corr.)dots (■) denote the contaminated data from the R/V's exhaust that were discarded when the raw dots (○) of the relative wind vector (us t ) in Eq. ( 1) were more than -0.1 m s -1 (positive to the prow as Fig. 4).The corr.dots reduced the scatter and improved the correlation between cruises observation and Kenting station.

Backward Trajectories and Atmospheric O 3 and Related Compounds from the Asian Continent Emissions
The cruise tracks along the Taiwan Strait and the northern SCS were typically influenced by the northeast Asian monsoon at the surface level during spring (Fig. 8(a)).This could result in the transport of O 3 and related compounds from the Asian continent to the downwind region (e.g., the northern SCS during March 2010).The strong northeasterly winds arising from the winter Asian monsoon may have transported polluted air masses from the Asian continent to the downwind region (e.g., the marine boundary layer of the northern SCS during March 2010) as indicated by elevated O 3 mixing ratio.The prevailing northeast wind occurred in all the 3 cruise areas, and the 5-d backward trajectories of the air masses, ending on March 16, 2010, at a 10-m height (Fig. 9(a)), were computed using the HYSPLIT model, which shows that the trajectory paths originated from the northern China to site A, passing the Korea, Shanghai city, East Asian coastline and Taiwan.The trajectory paths originated from Mongolia and northern China, passing the East Asian coastline and Yonagunijima, Japan and reached EPA Kenting station with a 10-m height.Arriving with a 1000 m height, the trajectory path reaching at site A, originating from the northern SCS and the Philippines, has a relatively lower emission than that reaching EPA Kenting, originating from and passing along the East Asian coastline.The navigated locations of site B ([118.2°E, 19.3°N]  2010, the CO peak in spring appears first in Yonagunijima's and Kenting's time series, then in site B's, indicating that the trajectory paths for the three sites are from the similar source.Around the anthropogenic pollutant sources, the conversion of VOCs and NO x to Peroxyacetyl Nitrate (PAN) plays an important role in O 3 production.PAN will decompose and regenerate NO x , leading to O 3 production at far downwind regions.Lin et al. (2013) reported the ∆O 3 /∆CO ratio of aged plumes (> 3 days) was 0.22, which is close to the ratio of 0.2 in this study.A similar trajectory result was derived from an investigation of the air-mass paths to the downwind region of the northern SCS conducted by Ou-Yang et al. (2013), Sheu et al. (2013), and Wang et al. (2013) during 2010 7-SEAS/Dongsha experiment.By contrast, during summer, the air masses of the northern SCS were primarily conveyed from the deep SCS or Western Pacific Ocean (WPO) (Fig. 8(c)).The Asian summer monsoons may convey relatively clean marine air masses with low concentrations of O 3 and related compounds.The highest hourly O 3 mixing ratio of 32.5 ppb was observed at 12:00 PM local time (LT).The 5-d backward trajectories ending at 2:00 PM on July 13, 2010 of site C were computed using the HYSPLIT model, which indicated that the air masses originating from the SCS and WPO arrived at the Taiwan Strait.The trajectory path originating from the WPO reached Kenting station (Fig. 9(c)) with relatively clean air masses, and the NO 2 , CO and O 3 concentrations were lower than cruise observation during July cruise.Thus, the enhanced concentrations over the Taiwan Strait (site C) may indicate the contributions from local transported O 3 .The relative fresh plumes of site C (< 2 days) have a lower ∆O 3 /∆CO ratio of 0.12 than the aged plumes of site B (> 3 days) with a ratio of 0.2.Li et al. (2012) reported a higher local contribution than super-region contribution of O 3 in the summer; it is possibly associated with the weaker WS.Despite the uncertainties and the numerous trajectories not being enough to provide accurate differences in CO and O 3 concentrations between aged plume and fresh plume, the significant difference in ∆O 3 /∆CO ratio and high-O 3 event at sunset give strong evidence that a long-distance transport from Asian continent might affect the northern SCS in March 2010.

Spatial Distributions and Diurnal Variations for All Air Masses
The spatial distributions of O 3 and related compounds are shown in Fig. (10).The high O 3 , CO, NO, and SO 2 mixing ratios were recorded near the East Asian continent and Taiwan.In the absence of human activity, the atmospheric CO concentrations are controlled by balanced photochemical production and destruction through the reaction of CO with hydroxyl radicals (OH•).O 3 is titrated by fresh NO emissions via O 3 + NO → NO 2 + O 2 , this study presents O 3 + NO 2 for the overall oxidant abundance.This quantity includes the amount of O 3 that is temporally lost by NO titration and the NO 2 from local emissions (Wang et al., 2009).Besides a high O 3 event of site B, Fig. (10) shows that the elevated O 3 + NO 2 grids are located at industrial harbor (Kaohsiung) and Fujian coastal cities where there are local sources.A negative correlation was observed between the air samples and the distance from Kaohsiung Harbor (Fig. 11(a)) and the coastline (Fig. 11(b)), which was consistent with the spatial distributions of the air samples.The distance from the coastline is based on the distance of each cruise location to the nearest coastline.The maximal O 3 concentrations were recorded proximal to the shore within 50 km of the coastline and all the air masses exhibited higher concentrations near the coastline, as shown in Fig. 11.Fig. 12 shows the average diurnal variations for the measured O 3 , NO 2 , NO, CO, SO 2 , CH 4 , and NMHC and EPA background station (Kenting) data during the 2010 cruises.The diurnal patterns of SO 2 , CO, and O 3 were similar as those observed at the EPA Kenting station.Regarding diurnal variations, other than NO x , the mixing ratios of the other air masses were relatively lower than those recorded at the background station.Typically, NO x , CO, and O 3 exhibited relatively flat values from nighttime to the early morning hours.The minimal concentration occurred at approximately 8:00 AM LT, its concentration increased, reached a maximal value at 2:00 PM LT, and then remained stable after sunset.These O 3 diurnal variations are consistent with those reported by Lin et al. (2011) for a background site in Asia.The diurnal patterns of CO, NO, and NO 2 were consistent with those reported by Lin et al. (2011); however, the maximal NO and NO 2 concentrations were occurred earlier, and the maximal CO concentrations were occurred later than the maximal O 3 concentrations reported by Lin et al. (2011).The CO concentrations increased in the morning, but gradually decreased after noon.The CH 4 concentrations increased during night, decreased during early morning, and then increased again before nightfall.Burkert et al. (2003) indicated that CO, CH 4 , and NMHC play vital roles in the production and depletion cycles of O 3 .The O 3 was photochemically produced after the NO 2 , CO, CH 4 , and NMHC concentrations increased during daytime.

CONCLUSIONS
This study presented the measurements of O 3 and related compounds over the Taiwan Strait and the northern SCS from cruises running in March to July of 2010 (R/V OR1 and R/V OR3).Continuous shipboard observation is characterized by the prevention of interferences such as gas moisture, sea salt aerosols, gas attached at tube and tube inlet, sample air was passively delivered and ship's chimney exhaust, and by overcoming the instrument calibration.According to Eq. (1), the wind direction, R/V position, and directory were constantly monitored to avoid any contamination from the R/V exhaust itself.Over the 3 cruises, the mean concentrations of O 3 , NO 2 , NO, CO, SO 2 , CH 4 , and NMHC were 25 ± 9.9 ppb, 3.2 ± 1.8 ppb, 2.9 ± 1.7 ppb, 204 ± 54 ppb, 1.7 ± 1.0 ppb, 1.7 ± 0.1 ppm and 0.2 ± 0.2 ppm, respectively.The correlation coefficient between CO and O 3 of all cruises was 0.50, and the ∆O 3 /∆CO ratio was 0.16.The results indicated a higher ∆O 3 /∆CO ratio of 0.20 at sunset during the March 2010 cruise can be mainly attributed to a long-distance transport of aged plumes (> 3   The trajectory paths for the three sites (Yonagunijima, Kenting and site B) are from the similar source, as suggested by the small delay in the appearance of the CO peak in time series on March 18, 2010.By contrast, a lower ∆O 3 /∆CO ratio of 0.16 and 0.12 found during the April 2010 and the July 2010 cruises, respectively, might have been due to local sources of fresh plumes (< 2 days).
The 5-d backward trajectory paths ending at the Taiwan Strait (site C) during the July 2010 cruises indicated that the air masses originated from the SCS and WPO passing the coastline of Fujian (< 1 day).Without passing the coastline of Fujian, the trajectory paths reaching EPA Kenting station shows relatively low and steady NO x , CO and O 3 mixing ratios.
Overall, the O 3 and related compounds mixing ratios over the ocean are affected by emission source, Asian monsoon, WS, gas deposition, gas solubility, α, the frontal inversion and BLH.Arriving with a 1000-m height, the trajectory path reaching at site A, originating from the northern SCS and the Philippines, has a relatively lower emission than that reaching EPA Kenting, originating from and passing along the East Asian coastline (Fig. 9(a)).Regarding the spatial distributions and diurnal variations of the air masses, the spatial distributions revealed higher SO 2 , NO, CO, and O 3 mixing ratios proximal to a continent or an island.The distribution of O 3 + NO 2 presented the amount of O 3 that is temporally lost by NO titration and the NO 2 that is directly emitted by local source.
Despite the uncertainties and the numerous trajectories not being enough to provide accurate differences in CO and O 3 concentrations between aged plume and fresh plume, the significant difference in ∆O 3 /∆CO ratio, wind direction, the peak of CO in time series at three sites (Yonagunijima, Kenting and site B) and high-O 3 event at sunset give strong evidence that the air masses in spring mainly originated from the Asian continent emissions.

Fig. 2 .
Fig. 2. Time series of 10-m air temperature, boundary layer height (ERA Interim data), atmospheric pressure, 10-m WS, specific humidity, incoming solar radiation and atmospheric longwave radiation during the periods from March 14 to March 20 of 2010 in RV/OR1, from April 12 to April 15 of 2010 and from July 11 to July 16 of 2010 in RV/OR3 in the SCS.

Fig. 3 .
Fig. 3. Weather Chart of surface analysis on (a) March 15 and (b) March 16 of 2010.(data was provided from the Central Weather Bureau, Taiwan) the x component of u  , v 1 is the y component of u  , φ is positive for the R/V prow yawed clockwise from north, s x is the x component of s  , and s y is the y component of s  .

Fig. 4 .
Fig. 4. Time series of hourly O 3 and related compounds observations on March 14-20, April 13-15, and July 11-16, 2010 over of the Taiwan Strait and the northern SCS.The black diamonds (◊) represent the relative wind components (the threshold of data discarded as us t > −0.1 m s −1 in Eq. (1)), the black circles (○) represent the raw data and the blue squares (■) represent data excluding interference from the R/V exhaust and the navigated locations of site B and site C had the highest O 3 concentrations in March and July 2010, respectively.
(2004), with an average of 1.75 ± 0.07 ppm.During the cruise toward the north (March 17 to March 19 of 2010), the CO concentrations in the northern SCS varied between 100 ppb and 200 ppb, which were similar to the concentrations observed byBurkert et al. (2003) in February, March, and April of 1999 over the Indian Ocean, and by Stubbins et al. (2006) during April-May 2000 along the same latitude over the Atlantic Ocean.CO is produced mainly from burning of fossil fuels and it is a precursor of O 3 .High CO periods (March 18, April 14, and July 13) coincided with sharp increases in O 3 concentrations in Fig 4. Tsai et al. (2012) also indicated the increasing O 3 was observed from Hengchuen station, Taiwan in April 13, 2010.Fig. 5(a) presented a correlation between the increase in ∆CO and the elevated ∆O 3 concentrations over the northern SCS and Taiwan Strait.

Fig. 5 .
Fig. 5. (a) Scatter plot of three cruises O 3 versus CO mixing ratio with linear trend lines, the black line is total trend, the blue dotted line is March 2010 cruise trend in March, the block dotted line is April 2010 cruise trend in April and the red dotted line is July 2010 cruise trend in July, (b) scatter plot of NO 2 versus WS with natural logarithm trend line, and (c) scatter plot of CO versus WS with natural logarithm trend line.

Fig. 7 .
Fig. 7. Scatter plot of three cruises O 3 and related compounds versus EPA environmental background station (Kenting) data, the black circles (○) represent raw data; the red squares (■) represent data excluding interference from the R/V exhaust.

Fig. 12 .
Fig. 12.Diurnal variations of SO 2 , NO 2 , NO, CO, O 3 , CH 4 , and NMHC over the Taiwan Strait and the northern SCS during 2010 cruises, the blue circles (○) represent valid data, and the red squares (□) represent background station (Kenting) data.
from the Asian continent of super-region contribution.

Table 1 .
Specifications of SO x , and CO analyzers.

Table 2 .
A comparison of oceanic O 3 and related compounds concentrations in this study with other observations.
Environmental Protection Administration (EPA) background Kenting station and Japan Meteorological Agency Yonagunijima station