Physicochemical Characteristics and Source Apportionment of Atmospheric Aerosol Particles in Kinmen-Xiamen Airshed

The objective of this study was to characterize the chemical properties of atmospheric particles sampled in the KinmenXiamen Airshed located on the west bank of the Taiwan Strait. Seven particulate matter (PM) sampling sites in the KinmenXiamen Airshed, including three sites at Kinmen Island and four in urban Xiamen, were selected for this particular study. Regular sampling was conducted to collect PM10 with high-volume samplers twice a month from March 2008, while intensive sampling was conducted to collect PM2.5 and PM2.5–10 with dichotomous samplers and PM10 with high-volume samplers in the spring and winter of 2008–2009. After sampling, the metallic contents of PM10 were analyzed with an inductively coupled plasma-atomic emission spectrometer (ICP-AES). Ionic species and carbonaceous contents of PM10 were analyzed with an ion chromatograph (IC) and elemental analyzer (EA), respectively. Finally, the source identification and apportionment of PM were analyzed by principal component analysis (PCA) and receptor modeling (CMB), respectively. The results from PM10 sampling indicated that atmospheric aerosol particles had a tendency to accumulate in Xiamen Bay all year round, particularly in spring and winter. The five sampling sites at the center of Xiamen Bay had relatively higher PM10 concentrations than the two sampling sites outside Xiamen Bay, suggesting that local emissions from Xiamen Bay were more significant than emissions transported over a long distance by the Northeastern Monsoon. The phenomenon of superimposition was regularly observed during air pollution episodes at Xiamen Bay. Moreover, the results of chemical analysis showed that the main chemical components of the PM were SO4, NO3, NH4, OC, and EC and crustal elements (Ca, Mg, Fe, and Al) in the aerosol particles in the Kinmen-Xiamen Airshed. The neutralization ratios (NR) of PM were generally smaller than unity, indicating that the atmospheric particulates were mostly acidic. The averaged sulfur oxidation ratio (SOR) ranged from 0.20 to 0.51, and the nitrogen oxidation ratio (NOR) ranged from 0.10 to 0.41 for all seasons. The ratios of sulfur and nitrogen oxidation were generally higher than 0.25 and 0.10, respectively, suggesting that secondary sulfate and nitrate aerosols came mainly from across-boundary transportation and could be further accumulated in the Kinmen-Xiamen Airshed. The results from CMB receptor modeling showed that the major sources of atmospheric PM10 in the Kinmen-Xiamen Airshed were soil dusts, secondary aerosols, the petroleum industry, motor vehicle exhausts, the iron and steel industry, the cement industry, diesel vehicle exhausts, marine aerosols, and biomass burning.


INTRODUCTION
Xiamen Bay has subtropical monsoon weather with an annual average temperature of about 21°C, rainfall of 1,043 mm, and prevailing winds from the northeastern monsoon (from October to April of next year) and the southwestern monsoon (from May to August).Xiamen and Kinmen Islands located in Xiamen Bay are a scenic coastal cities in south-eastern China, covering areas of 132.5 and 150.0 km 2 , respectively.There are huge industrial emissions in metro Xiamen, including these three coal-fired power plants, stone processors, ceramic industry, porcelain products, clothing manufacturers which are next to the coastal region at Xiamen Bay.There are no large-scale industrial emission sources on Kinmen Island.According to source emission data of 2009, the stationary sources in urban Xiamen were approximately twenty times higher than those on Kinmen Island.
Particulate matter with aerodynamic diameter (d pa ) ≤ 10 μm (PM 10 ), particularly the fine particle fraction (d pa ≤ 2.5 μm), has been shown to be associated with health problems, such as asthma (Anderson et al., 1992;Dockery et al., 1993;Dockery and Pope, 1994;Pope et al., 2002).Previous studies reported that the physicochemical characteristics of atmospheric aerosols correlate closely with ambient air quality, atmospheric visibility reduction.(Yuan et al., 2002;Lee et al., 2005;Yuan et al., 2006), and human health.Our recent studies indicated that the main chemical composition of PM 10 are secondary aerosols in water-soluble ionic species, crustal elements in metallic contents, and carbons in Xiamen City in winter and spring.Ammonium sulfate and nitrate are generally the major common components of secondary aerosols in the atmosphere, converting from sulfur dioxide (SO 2 ) and nitrogen oxides (NO x ), respectively.The sulfur oxidation ratio (SOR) expresses the degree of oxidation of sulfur in terms of the ratio of sulfate to total sulfur (sulfate plus SO 2 ).Similarly, the nitrogen oxidation ratio (NOR) expresses the degree of oxidation of nitrogen in terms of the ratio of nitrate to total nitrogen (nitrate plus NO x ).High SOR and NOR suggest that photochemical oxidation tends to form secondary aerosols in the atmosphere (Colbeck and Harrison, 1984;Ohta and Okita, 1990).
In recent years, the ambient air quality of the Kinmen-Xiamen Airshed has deteriorated gradually, and PM 10 is responsible for the poor air quality in spring and winter.High percentages of poor air quality (Pollutants Standard Index, PSI > 100), ranging from 5.5 to 14.0% at Kinmen Island in the years 2002-2008, show that it had the worst air quality among seven Air Quality Zones (AQZ) in Taiwan.PSI is the abbreviation of Pollutant Standards Index based on five criteria air pollutants: PM 10 , sulfur dioxide, carbon monoxide, nitrogen dioxide, and ozone.For each air pollutant, a value of 100 is assigned to be the maximum permitted concentration of that air pollutant.After determining the sub-index value of each air pollutant, the highest sub-index of the five air pollutants is reported as the Pollutant Standards Index (PSI) of the day.High levels of PM 10 (> 125 mg/m 3 ) found in Kinmen Islands during the northeast monsoon seasons are likely blow from the upwind emission sources.Similar seasonal variations of low PM 10 levels from June to August and high PM 10 levels from October to March of next year, in the Xiamen Bay.Moreover, more than 700 ceramic and title factories mainly with coal-fired furnaces in Jinjiang City located at the northeastern region of the Xiamen Bay, where highly influenced the air quality of Xiamen Bay under the northeastern winds (Wu, 2011).
Consequently, this study investigated the tempospatial distribution of PM 10 , including the mass concentration and physicochemical properties in the Kinmen-Xiaman Airshed.The source identification and apportionment of PM 10 were further analyzed by principal component analysis (PCA) and receptor modeling (CMB).The SURFER model is seldom used in combination with the CMB model to look for PM 10 hot spots and estimate the contribution of various sources, and this is the aim of the current study.

Sampling Protocol
The locations of the seven PM 10 sampling sites are illustrated in Fig. 1, including three sites on Kinmen Island and four sites in urban Xiamen.Among these, two sampling sites at Zhangzhou campus of Xiamen University (A1) (118°02′N, 24°22′E) and Jinjing Elementary School (A4) (118°36′N, 24°34′E) are located outside Xiamen Bay.The other five sampling sites at the main campus of Xiamen University (A2) (118°05′N, 24°26′E), Daderng High School (A3) (118°19′N, 24°33′E), Lieyu Junior High School (B1) (118°14′N, 24°25′E), Jinding Elementary School (B2) (118°20′N, 24°26′E), and Jinsha Elementary School (B3) (118°24′N, 24°29′E) are located inside Xiamen Bay.Sampling location, altitude, and surrounding environment of each site are summarized in Table 1.Sampling sites A2 and A4 located in the downtown Xiamen and Jinjing, are next to a street and likely to be influenced by direct emissions from vehiclular exhausts and textile industries.Site A1 located in the Zhangzhou campus of Xiamen University at the foot of hills has a rural area characters.Air pollutants from Zhangzhou Harbor, Xiamen Harbor, Songyu power plant and the Xiamen metropolitan area would influence the ambient air quality at site A1 during the northeastern monsoon periods, whereas air pollutants emitted from Houshi power plant could be transported to the downwind sites.Site A3 located at an island of Dadeng in northern Xiamen is a tourism township with a cargo harbon.Kinmen Islands is recognized as a national park and its local emission sources are well controlled.Thus, sites B1, B2, and B3 in Kinmen Islands are more likely to be influenced by regional air pollution, especially from the upwind northern and northeastern industrial areas in Jinjiang River and Jinjing during the northeastern monsoon seasons.
Regular and intensive PM sampling were conducted from March 2008 to February 2009.Regular sampling was conducted to collect PM 10 with a high-volume sampler for 24 hours at each site twice a month from March 2008, while intensive sampling was conducted to collect 24-hr PM 2.5 and PM 2.5-10 with a dichotomous samplers at the Jinding Elementary School site (B2) and 24-hr PM 10 with a highvolume sampler at each site on January 6-10 and March 16-20, 2009.Ambient particles with aerodynamic diameters below 10 μm (PM 10 ) were divided into two separate fractions (i.e., PM 2.5 and PM 2.5-10 ) using a virtual impactor with a 10 μm cutpoint at the inlet of the sampler.These two fractions were classified as fine (PM 2.5 ) and coarse (PM 2.5-10 ) particles.In this study the space analysis software (SURFER) was used to assess the PM 10 hot spots in Kinmen-Xiamen Airshed and to draw the concentration contours of each month from 2008 to 2009.The SURFER model has been commonly used to describe the spatial distribution of PM 10 for metropolitan district, industrial area, rural area, etc. (Otvos et al., 2003;Shaocai et al., 2004;Tsai et al., 2010).

Chemical Analysis
After sampling, quartz filters were temporarily stored at 4°C and then transported back to the Air Pollution Laboratory in the Institute of Environmental Engineering at National Sun Yat-Sen University for further conditioning, weighing, and chemical analysis.All PM 10 sampling filters collected by high-volume samplers were analyzed for chemical composition.One part of the quartz filter was analyzed for water-soluble ionic species.The filter analyzed for ionic species was put inside a 15-mL PE bottle for each  Another part of the quartz filter analyzed for metals was initially digested in a 20 mL mixed acid solution (HNO 3 : HClO 3 = 3:7) at 150-200°C for 2 hrs, and then diluted to 25 ml with distilled de-ionized water (D.I.H 2 O).During the digestion, D.I. H 2 O was added to the residual solution two or more times in order to eliminate the acid content of the digestion solution.The metallic species of the PM, including Na, Ca, Al, Fe, Mg, K, Zn, Cr, Ti, Mn, Ba, Sr, Ni, Pb, and Cu, were then analyzed with an Inductively Coupled Plasma-Atomic Emission Spectrometer, ICP-AES (Perkin Elmer, Model Optima 2000DV).
Two parts of quartz filters were further used to measure the carbonaceous contents of the PM.The carbonaceous contents, including elemental, organic, and total carbons (OC, EC, and TC), were measured with an Elemental Analyzer (Carlo Erba, Model 1108).Before sampling, quartz filters were pre-heated to 900°C for 1.5 hr to remove potential carbon impurities from the filters.The preheating procedure could minimize the background carbon in the quartz filters and matrix, which might cause interference with the analytical results, leading to an overestimation of the carbonaceous content of the PM.The Elemental Analyzer (EA) was operated using the procedure of oxidation at 1020°C and reduction at 500°C, with continuous heating for 15 mins.Additionally, one-eighth of the quartz filter was heated in advance using nitrogen gas at 340-345°C for at least 30 mins to expel the organic carbon (OC) fraction, after which the amount of elemental carbon (EC) was determined.Another one-eighth of the quartz filter was analyzed without heating to determine total carbon (TC).The amount of OC was then estimated by subtracting EC from TC.

Quality Assurance and Quality Control
The quality assurance and quality control (QA/QC) for both PM sampling and chemical analysis were conducted in this study.Prior to conducting PM sampling, the flow rate of each PM sampler was carefully calibrated with a film flowmeter (MCH-01 SENSIDYNE Inc.).Quartz filters were then carefully handled and placed on the PM 10 samplers to prevent potential cracking during the sampling procedure.After sampling, aluminum foil was used to fold the quartz filters, which were then temporarily stored at 4°C and transported back to the central laboratory for chemical analysis.The sampling and analytical procedure was similar to that described in various previous studies (Cheng and Tsai, 2000;Lin, 2002;Yuan et al., 2006;Tsai et al., 2008;2010, 2011).Both field and transportation blanks were used for PM sampling, while reagent and filter blanks were used for chemical analysis.The determination coefficient (R 2 ) of the calibration curve for each chemical species was required to be higher than 0.995.Background contamination was routinely monitored by using operational blanks (unexposed filters), that were proceeded simultaneously with field samples.The background interference was insignificant in this study, and can thus be ignored.At least 10% of the samples were analyzed by spiking with a known amount of metallic and ionic species to calculate their recovery efficiencies.

Principal Component Analysis (PCA) and Chemical Mass Balanced (CMB) Receptor Model
The concentrations of ionic species, metallic elements, and carbonaceous contents for PM 10 samples were used to calculate varimax rotated principal component analysis to identify the number of principal components having eigenvalues > 1.0 (Tandon et al., 2008;Deshmukh et al., 2011).The source apportionment of ambient PM 10 was assessed using a receptor model based on the chemical mass balance (CMB) (Ke et al., 2007;Kothai et al., 2008;Wang et al., 2008;Yatkin and Bayram, 2008).Since the detailed descriptions of CMB receptor model (e.g., CMB8) are available elsewhere, only a brief summary is presented below.
The CMB receptor model uses the emission profiles of prominent sources to estimate their contribution to a specific receptor.It is assumed that the total concentration of a particular chemical species at the receptor is the linear summation of each individual contribution from various sources.The CMB receptor model uses the results of the least-square regression analysis of the aerosol chemical composition to estimate the most appropriate contributions of source apportionment.Therefore, this model consists of a least-square solution to a set of linear equations.This solution expresses each receptor concentration of a chemical species as a linear summation of the products of source profiles and source contributions.Source profiles (the fractional amount of each species in the emissions from each source type) and receptor concentrations, each with realistic uncertainty estimates, serve as input data to the CMB receptor model.The model output consists of the contribution from each source type to the total ambient aerosol mass, as well as to individual chemical species concentration.The CMB8 model results are evaluated by using several fit indices, such as R 2 (≥ 0.8), χ 2 (≤ 4.0), T statistics (≥ 2.0), and percentage of mass accounted for 0.8-1.2.The source profiles used in this study were reported by USEPA, Southern California Air Quality Study, and the researches were studied the chemical composition and source profile in Taiwan.Table 2 summarizes the source profiles which were used in this case of PM 10 in the Kinmen-Xiamen Airshed.

Chemical Transformation of SO 2 and NO x
Sulfate and nitrate are the major components contained in the atmospheric aerosols in urban areas.To determine the degree of atmospheric transformation of SO 2 to SO 4 2and NO x to NO 3 -, the sulfur and nitrogen oxidation ratios (i.e., SOR and NOR) were employed, and these are defined as follows (Colbeck and Harrison, 1984;Ohta and Okita, 1990;Kaneyasu et al., 1999): where nss-SO 4 is the excess sulfate that was calculated by subtracting the amount of SO 4 2-of marine from that of SO 4 2-in the atmosphere (Kaneyasu et al., 1995;Cheng et al., 2000).The units of nss-SO 4 , S SO2 , N NO3 , and N NO3 -are neq/m 3 .The average concentrations of SO 2 and NO x during each sampling period were obtained from the ambient air quality monitoring station in Kinmen Island.

PM Concentration and Size Distribution
As illustrated in Table 3 and Fig. 2, regular sampling of PM 10 concentrations at seven sampling sites at Xiamen Bay was conducted from March 2008 to February 2009.The average concentration of 24-hr PM 10 varied from 43.5 to 148.53 μg/m 3 , with an average of 91.1 ± 25.9 μg/m 3 in spring; from 30.8 to 83.2 μg/m 3 , with an average of 48.5 ± 14.6 μg/m 3 in summer; from 42.0 to 90.7 μg/m 3 , with an average of 66.8 ± 14.7 μg/m 3 in fall; and from 45.49 to 136.4 μg/m 3 , with an average of 93.5 ± 45.9 μg/m 3 in winter.
Among the seven sampling sites, the average concentrations of 24-hr PM 10 frequently violated the ambient air quality standard of 125 μg/m 3 in winter.The highest PM 10 concentrations mostly occurred in winter, while the lowest generally occurred in summer.The PM 10 concentration contour shows that the highest PM 10 concentration was observed along the northern coast of Xiamen Bay.The results from PM 10 sampling indicate that PM 10 had a tendency to  accumulate in Xiamen Bay all year round, particularly in spring and winter.The five sampling sites (A2, A3, B1, B2, and B3) located at the center Xiamen Bay had relatively higher PM 10 concentration than the other two sampling sites (A1 and A4) outside Xiamen Bay.The PM 10 concentrations at the upwind sites, including site A4, were not much higher than those at the downwind site A1 during the Northeastern Monsoons in spring and winter.The highest PM 10 concentrations occurred at site A3 in the Kinmen-Xiamen Airshed, especially in December.The results suggest that a superimposition phenomenon was regularly observed during air pollution episodes at Xiamen Bay.Local emissions from the Xiamen Bay were thus more significant than emissions transported over a long distance by the Northeastern Monsoons.
During the intensive sampling periods, fine (PM 2.5 ) and coarse (PM 2.5-10 ) particles were simultaneously sampled at site B2 with a dichotomous sampler.The concentrations of fine and coarse particles and their mass ratios (PM 2.5 / PM 2.5-10 ) are summarized in Table 4.During the first intensive sampling period in winter, the highest PM 10 and PM 2.5 concentrations were 115.2 and 56.0 μg/m 3 , respectively.
(1) Spring  During the second sampling period in spring, the highest PM 10 and PM 2.5 concentrations were 172.3 and 121. 2 μg/m 3 , respectively.The mass ratios of PM 2.5 to PM 10 ranged from 40.9 to 57.2% and from 63.5 to 72.6% for the first and second intensive sampling periods, respectively.Generally speaking, fine particles were more common than coarse particles in the Kinmen-Xiamen Airshed, suggesting that high PM 10 concentration was probably attributed to local anthropogenic sources (i.e., industrial and vehicular emissions) adjacent to the sampling sites during the intensive sampling periods.

Tempospatial Variation of Ambient PM 10
The PM 10 monitoring data at seven sampling sites was used to investigate the spatial distribution and temporal variation of PM 10 concentration.Figs. 2 and 3 illustrates the daily variation of wind direction and monthly variations of the PM 10 concentration contour of each sampling sites in the Kinmen-Xiamen Airshed.In fall, winter, and spring (from September 2008 to May 2009), the Daderng High School (A3) site and northern Xiamen island generally had the highest PM 10 concentrations in Xiamen Bay.In summer (from June to August, 2008), the PM 10 concentration was relatively higher in the region between urban Xiamen and Kinmen Islands.High PM 10 concentrations were always observed at the sites adjacent to industrial areas along the northern coast of Xiamen Bay.PM 10 concentrations in winter (37.6-192.9μg/m 3 ) were higher than those in summer (23.5-83.2μg/m 3 ), because the weather system in the Kinmen-Xiamen Airshed is dominated by Northeastern Monsoons, which blow air pollutants from the eastern coast of China to Xiamen Bay from approximately March to February.However, local emissions from the surrounding region of Xiamen Bay were as important as long-range transportation by Northeastern Monsoons, and thus a superimposition phenomenon was regularly observed during the air pollution episodes at Xiamen Bay.
In winter and spring, the PM 10 concentration at site A3 was the highest, followed by those at site A1 located at Xiamen Island and site B1 located at Kinmen Island.In summer, the PM 10 concentration was relatively higher at sites A1 and B3.In fall, the PM 10 concentration was higher at sites A1, A2, and A3, located in urban Xiamen.It should be noted that the sampling sites with high PM 10 concentrations were always adjacent to either industrial areas or main traffic arteries.Fig. 4 illustrates the concentration contours and prevailing wind direction in the Kinmen-Xiamen Airshed.The concentration contours of PM 10 show that the PM 10 hot spots of were always located between Daderng Island and urban Xiamen in the Kinmen-Xiamen Airshed.The hot spots of PM 10 in the Kinmen-Xiamen Airshed were highly affected by the wind field of Xiamen Bay.The highest concentrations of PM 10 were observed in January, mainly due to the Northeastern Monsoons.Northern winds could   transport atmospheric particles from the upwind emission sources (e.g., Jinjiang River Basin) to the downwind sites (Kinmen Island), causing a significant increase in atmospheric particle concentrations at Kinmen Island.The Hybrid Single-Particle Lagrangian Integrated Trajectory is a widely used model that plots the trajectory of a single air parcel from a specific location and height above ground over a period of time.The 96-hour backward trajectories starting at 118°14′N, 24°25′E at the heights of 100, 350, and 500 m, respectively, to study the transport pathways of air parcels that arrived at Xiamen Bay on four different days are shown in Fig. 5 (Chen et al., 2012).The level of atmospheric PM 10 is affected by meteorological condition, thus PM 10 concentrations in spring and winter was much higher than those in fall and summer.Results from backward trajectories showed that the concentrations of PM 10 blown from the north were generally higher than those from the south.

Chemical Composition of PM 10
Figs. 6-8 illustrate the seasonal variations in the chemical composition of PM 10 in the Kinmen-Xiamen Airshed.As shown in Fig. 6, the abundant water-soluble ionic species of PM 10 were SO 4 2-, NO 3 -, and NH 4 + at Xiamen Bay.The main chemical species of PM 10 were probably ammonium sulfate ((NH 4 ) 2 SO 4 ) and ammonium nitrate (NH 4 NO 3 ) (Yao et al., 2003;Han et al., 2007;Kocak et al., 2007;Tsai et al., 2012).The sulfate concentrations ranged from 3.0 to 37.0 μg/m 3 with an average of 13.0 μg/m 3 , the nitrate concentrations ranged from 1.4 to 25.3 μg/m 3 with an average of 9.2 μg/m 3 , and the ammonium concentrations ranged from 1.0 to 19.1 μg/m 3 with an average of 6.5 μg/m 3 , which were generally higher in spring and winter than those in summer and fall.
The metallic contents of PM 10 sampled in the Kinmen-Xiamen Airshed are shown in Fig. 7. Crustal elements (Ca, Mg, Fe, and Al) and anthropogenic elements (Zn, As, and Pb) contributed the major metallic contents of PM 10 .Among the crustal elements, Ca and Fe were the most abundant metals.In particular, the concentration of Ca was much higher than that of other crustal elements (Fe, Mg and Al).The highest concentrations of Ca in spring and winter were 3.4 and 5.7 μg/m 3 , respectively.Zn and Pb were the major components among the other metallic elements (Cr, Mn, Ni, and Cd).Pb is a toxic heavy metal (Barrat, 1990;Akhler and Madany, 1993;Pirrone et al., 1996), and the highest concentrations of Pb were observed at sites A2 and A3.Observations showed that the incinerator combustion and heavy vessel travel were always heavy around Xiamen Bay, resulting in high Zn and Pb emissions.High Pb emissions may also came from incinerator, leaded-gasoline combustion.High Pb concentrations could be also contributed form deuse heavy vessels which were quite busy around Xiamen Bay.The oil burning for heavy vessels is one of the major cause of chemical characteristic for Pb.The high Pb concentration were also contributed form heavy vessel which were always heavy around Xiamen Bay.The oil of vessel were one of the cause of the high marker for the metal of Pb.
The concentrations of Ca and Fe at sites A2 and A3 in the rural open lands and construction areas were higher than those at other sites.Al and Ca are the main elements in the earth's crust and particles emitted from the cement industry, and the wind-blown dusts and frictional works from construction sites increased the atmospheric loading of dust particles (Huang et al., 1994).In addition, there are many stone processing plants located to the north of site A3.Moreover, anthropogenic metallic elements could be transported to Xiamen Bay from the eastern coast of China through long-range transportation by the Northeastern Monsoons.
The carbonaceous contents of PM 10 sampled in the Kinmen-Xiamen Airshed are illustrated in Fig. 8. Elemental carbon (EC), which has a chemical structure similar to impure graphite, originates primarily from direct emissions from combustion.Organic carbon (OC) is emitted from primary anthropogenic sources and secondary organic aerosols formed by chemical reactions in the atmosphere.

Spring Summer Fall Winter
In this study, the concentrations of OC were always higher than EC for all seasons at each sampling site.The mass ratios of OC to EC (OC/EC) ranged from 1.1 to 2.7, and were larger than unity for PM 10 at all sampling sites.The average OC/EC ratio of 1.6 at urban Xiamen was higher than that of 1.4 at Kinmen Island.The results indicate that OC was the major carbonaceous species of PM 10 in the Kinmen-Xiamen Airshed.The OC/EC ratio can be used to identify the formation of secondary organic aerosols when the OC/EC ratio exceeds 2.2 (Turpin et al., 1990;Chow et al., 1996).The order of OC/EC ratios for PM 10 at all sampling sites was A4 > A2 > A3 > A1 > B3 > B2 > B1.In this study, the highest average OC/EC ratio of PM 10 was observed at site A4, which was adjacent to major emission sources such as textile plants, industrial boilers, heavy oil and coal burning.Certain meteorological conditions (i.e. the Northeast Monsoons) and the huge amount of volatile organic compounds emitted from various sources (e.g., textile industry at Jinjing River Basin) could enhance the formation of secondary organic aerosols.

Chemical Transformation of SO 2 and NO x
The chemical transformations of SO  The values of SOR were generally higher in spring than those in other seasons, since the major emissions of SO 2 were mostly contributed from both local emissions and longrange transportation.The average SOR of PM 10 at Kinmen Island ranged from 0.20 to 0.51.Previous studies reported that SOR and NOR are less than 0.25 and 0.10 for primary pollutants, respectively, while the chemical oxidation of SO 2 and NO x would occur in the atmosphere when SOR and NOR are greater than 0.25 and 0.10, respectively (Ohta and Okita, 1990).The high SOR and NOR obtained in this study suggest that the formation of SO 4 2-and NO 3 -from SO 2 and NO x could occur in the atmosphere.The average NOR of PM 10 at Kinmen Island ranged from 0.10 to 0.41.A previous study reported that NOR is generally lower than SOR (Colbeck and Harrison, 1984).The high NOR obtained in this study suggests that the formation of NO 3 -from NO x occurred in the atmosphere during the sampling period on Kinmen Island.6-Jan.7-Jan.8-Jan.9-Jan.10-Jan.14-Jan.20-Feb.28-Feb.5-Mar.16-Mar.17-Mar.18-Mar.19-Mar.20-Mar.
SOR or NOR SOR NOR Fig. 9.The SOR and NOR ratios of PM 10 in the Kinmen-Xiamen Airshed.

Source Apportionment of PM 10
PCA has been commonly used as an exploratory tool to identify the major sources of aerosol emissions and to statistically select independent source tracers.One of the objectives of this study was to distinguish the emission sources of PM 10 sampled in the Kinmen-Xiamen Airshed.The concentrations of ionic species, metallic elements, and carbonaceous contents of PM 10 were used to calculate varimax rotated principal component analysis to identify the number of principal components having an eigenvalue > 1.0 (Tandon et al., 2008) involved in the emission of PM 10 .The source apportionment of PM 10 was further assessed using a receptor model based on chemical mass balance (CMB) (Ke et al., 2007;Kothai et al., 2008;Wang et al., 2008;Yatkin and Bayram, 2008).
Previous investigation reported that SO 4 2-, Al, OC, and EC are from diesel vehicular exhausts (Wang et al., 2003;Cheng et al., 2010).NH 4 + , SO 4 2-, and NO 3 -are from secondary inorganic aerosols, while Na + and Cl -are from oceanic spray (Khemani et al., 1985).Pb is highly associated with vehicular exhausts, while Ca, Fe, and Al are associated with fugitive soils and road dusts (Kumar et al., 2001).Cr, Pb, and Zn originate from anthropogenic sources, such as iron and steel plants, power plants, and industrial boilers., and Cu, Mn, and Pb are regarded as road dusts emitted from heavy traffics and vehicular emissions (Kumar et al., 2001;Davis et al., 2001).K is mainly derived from biomass burning.Ni is straight from lumber and heavy oil combustion.Tables 5 and 6 summarize the results of PCA for the ionic species, metallic elements, and carbonaceous contents of PM 10 on Kinmen Island and in urban Xiamen.Five major factors (KF1-KF5) contributing to PM 10 were identified on Kinmen Island (see Table 5).The highest component loading associated with NH 4 + , NO 3 -, SO 4 2-, EC, and OC was identified as Factor KF1, which accounted for 27.16% of the variance.Moderate loading of OC, NO 3 − , SO 4 2− , and were also observed in Factor KF1, and these are secondary aerosols and the byproducts of combustion.Hence, the first factor can be identified as industrial emission plus secondary aerosols, Factor KF2 was heavily loaded with Ca, Fe, and Al with the percentage variance of 19.79%, which can be interpreted as the crustal materials.Thus, the second factor can be identified as road dusts.Factor KF3 explained 13.38% of the variance and was highly loaded with Pb, Ni, amd K, which are markers for biomass burning, heavy oil combustion (Kowalczyk et al., 1982;Harrison et al., 1996), and waste incinerator.The high concentrations of Pb were contributed mainly form heavy vessels which were always heavily deuse around Xiamen Bay.The used by vessels contains Pb which can be used as a markers for vessels.The third factor can thus be identified as biomass and fuel combustion and waste incinerator.Factor KF4 explained 11.61% of the variance and was highly loaded with Na + and Cl -, and thus the fourth factor can be identified as oceanic spray (Khemani et al., 1985).Factor KF5 explained 8.38% of the variance and was highly loaded with Cu and Mn, and thus the fifth factor can be identified as industrial emissions.
Four factors (XF1-XF4) accounting for 80.45% of the total variance were identified for PM 10 in urban Xiamen (see Table 6).Factor XF1 explained 42.84% of the variance and presented high loadings of Ca, Fe, and Al, which can be interpreted as the crustal contribution.Moderate loadings of OC, NO 3 − , SO 4 2− and NH 4 + were also observed in Factor XF1.These are secondary aerosols and the byproducts of combustion, and thus the first factor can be identified as the crustal contribution plus secondary aerosols.Factor XF2 explained 16.54% of the variance and was highly loaded with Na + and Cl − .Hence, the second factor can be identified as oceanic spray.Factor XF3 explained 14.07% of the variance and was highly loaded with Cu, Ni, and K. K is derived from biomass burning, while Ni is straight from lumber and heavy oil combustion.The third factor can thus be identified as biomass burning and heavy oil combustion.Factor XF4 explained 6.99% of the variance and was highly loaded with Cr, which originates from anthropogenic sources, such as iron and steel plants, power plants, and industrial boilers.Hence, the fourth factor can be identified as industrial emissions.According to the PCA results, the major sources in the Kinmen-Xiamen Airshed were secondary aerosols, crustal materials, biomass burning, oceanic spray, and heavy oil combustion.Table 7 summarizes the source apportionment of PM 10 in the Kinmen-Xiamen Airshed, and an obvious seasonal variation can be observed.In spring, soil dusts (19.93-24.51%)were the main source of PM 10 , followed by secondary aerosols (16.39-22.86%),industrial boilers (11.94-19.48%),motor vehicular exhausts (8.48-14.41%),diesel vehicular exhausts (3.12-7.33%),petrochemical plants (2.21-5.33%),oceanic spray (3.55-5.34%),steel plants (3.40-5.86%),biomass burning (2.12-8.11%),and cement plants (1.65-4.97%) in the Kinmen-Xiamen Airshed.Among these, soil dusts, petrochemical plants, and vehicular exhausts were the major sources in metro Xiamen, whereas soil dusts and biomass burning dominated in Kinmen Island.In particular, the concentration of oceanic spray on Kinmen Island (3.89-5.34%)was similar to that observed in Xiamen Island (3.55-4.87%).In summer, soil dusts, vehicular exhausts, and secondary aerosols were the major sources that contributed to PM 10 .In fall, soil dusts, secondary aerosols, industrial boilers, and vehicular exhausts were the major sources in metro Xiamen.The seasonal variation of the contribution of vehicular exhausts to PM 10 in urban Xiamen was always higher than at other sites, since urban Xiamen had the heaviest traffic in the Kinmen-Xiamen Airshed.Although vehicular exhausts were the dominant source, biomass burning and secondary aerosols were also significant sources causing an increase in PM 10 at urban and suburban areas in the Kinmen-Xiamen Airshed.In winter, soil dusts, secondary aerosols, vehicular exhausts, and industrial boilers were the major sources of PM 10 .The Northern Monsoons transported suspended particles from the upwind emission sources (e.g., Jinjiang River Basin) to the downwind sites (B1-B3) on Kinmen Island, causing a significant increase in secondary aerosols mainly composed of sulfate and nitrate.Moreover, the contribution of secondary aerosols to PM 10 in metro Xiamen was generally higher than that in Kinmen Island.In contrast, the contribution of biomass burning to PM 10 in Kinmen Island was higher than that in metro Xiamen.

CONCLUSIONS
This study investigated the tempospatial distribution, physicochemical characteristics, and source apportionment of atmospheric particles in the Kinmen-Xiamen Airshed.The results revealed that high PM 10 concentrations were usually found at the sampling sites adjacent to the industrial areas along the northern coast of Xiamen Bay.The PM 10 concentration contour showed that the highest PM 10 concentrations generally occurred in the region between metro Xiamen and Kinmen Island.A superimposition phenomenon was regularly observed during air pollution episodes at Xiamen Bay, resulting from both local emissions from Xiamen Bay and long-range transportation from the eastern coast of China by Northeastern Monsoons.
The mass ratios of fine and coarse particles (PM 2.5 / PM 2.5-10 ) ranged from 40.9 to 57.2% and 63.5 to 72.6% during the first and second intensive sampling periods, respectively.Fine particles were more common than coarse particles in the Kinmen-Xiamen Airshed, suggesting that the high PM 10 concentration could mainly be attributed to local anthropogenic sources (i.e.industrial emissions and vehicular traffics) adjacent to the sampling sites in urban Xiamen.Chemical analysis indicated that the major watersoluble ionic species of PM 10 were SO 4 2-, NO 3 -, and NH 4 + in Xiamen Bay, suggesting that PM 10 was mainly composed of ammonium sulfate and ammonium nitrate, which were generally seen in higher amounts in spring and winter than in summer and fall.The most abundant metals of PM 10 were Ca, Mg, K, Fe, Al, Zn, Pb, Mn, and Cu, while OC was higher than EC for all seasons in the Kinmen-Xiamen Airshed.The mass ratios of OC to EC (OC/EC) ranged from 1.1 to 2.7, and were larger than unity for PM 10 at all the sampling sites.SOR and NOR of PM 10 on Kinmen Island ranged from 0.20 to 0.51 and 0.10 to 0.41, respectively.High SOR and NOR suggested that the formation of SO 4 2and NO 3 -from SO 2 and NO x could occur in the atmosphere of the Kinmen-Xiamen Airshed.The results of the principal component analysis and CMB receptor modeling showed that major sources of PM 10 in the Kinmen-Xiamen Airshed were soil dusts, secondary aerosols, petrochemical industries, vehicular exhausts, industrial boilers, oceanic spray, and biomass burning.Northern winds may have transported suspended particles from the upwind emission sources (e.g., Jinjiang River Basin) to the downwind sites (Kinmen Island), causing a significant increase in the concentration of atmospheric particle on Kinmen Island.

Fig. 3 .
Fig. 3. Monthly variation of PM 10 concentrations at seven sampling sites in the Kinmen-Xiamen Airshed from March 2008 to February 2009.

Fig. 4 .
Fig. 4. Monthly variation of PM 10 concentration contours and prevailing wind direction in the Kinmen-Xiamen Airshed from March 2008 to February 2009.

Fig. 8 .
Fig. 8. Tempsoparial variation of carbonaceous contents and their mass ratios (OC/EC) ratio of PM 10 in the Kinmen-Xiamen Airshed.

Table 1 .
Sampling location and environmental description for seven sampling sites.

Table 2 .
Source Profiles of PM 10 were used in chemical mass balance.
Wang et al., 2006The source profiles used in this study were mainly obtained from the researcher's finding of the chemical composition of PM 10 emitted from various emission sources.Only limited source profiles are referred from USEPA and Southern California Air Quality Study, and local emission source profiles.

Table 3 .
The mass concentration for different fraction of particles in the Kinmen and Xiamen region.

Table 4 .
Concentrations of fine and coarse particles and their mass ratio during the intensive sampling periods.

Table 5 .
Principal component analysis of PM 10 sampled on Kinmen Island.

Table 6 .
Principal component analysis of PM 10 at metro Xiamen.

Table 7 .
Source apportionment of PM 10 sampled in the Kinmen-Xiamen Airshed.