Atmospheric Deposition of Polychlorinated Dibenzo-p-dioxins and Dibenzofurans at Coastal and High Mountain Areas in Taiwan

Atmospheric deposition is of great importance for the sink of air pollutants to the environment, either from local sources as well as coming from long range transport. To further understand the combined impact of both long-range transport from South East Asia and local emission sources of pollutants, the characteristics of polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs) in the ambient air of two background sites in Taiwan namely, Hengchun (coastal area) and Lulin (high mountain area), were simulated by regression of PM10 versus total-PCDD/F mass concentration, modeling of gas-particle partitioning, and simulations of both dry and wet deposition. The simulated PCDD/F concentrations at Hengchun, ranged between 0.0039–0.0106 and 0.0054–0.0138 pg WHO2005-TEQ m for 2012 and 2013, respectively. For Lulin site, the PCDD/F concentrations ranged between 0.0016–0.0029 and 0.0016–0.0032 pg WHO2005-TEQ m for 2012 and 2013, respectively. The WHO2005-TEQ ratios of PCDDs to PCDFs at both sites were less than unity, indicating that PCDF dominated the total toxicity. The results show that higher chlorinated PCDD/Fs primarily present in particulate phase for all seasons at both sampling sites especially in winter. Average dry deposition fluxes at Hengchun (57.1 pg WHO2005-TEQ m month) were approximately 5.1 times higher than those at Lulin (11.2 pg WHO2005-TEQ m month). The annual average dry deposition velocities were estimated to be 0.28 and 0.22 cm/s for Henchun and Lulin, respectively. For the whole period (2012–2013), the mean monthly wet deposition fluxes at Hengchun (averaged 11.7 pg WHO2005-TEQ m month) were 1.44 times higher than that at Lulin (averaged 8.11 pg WHO2005-TEQ m month). The calculated annual average total-PCDD/Fs-WHO2005-TEQ concentrations in the rain were 0.064 and 0.027 pg WHO2005-TEQ L for Henchun and Lulin, respectively. The estimated annual average scavenging ratios of total-PCDD/Fs-WHO2005-TEQ were 8015 and 13450 for Henchun and Lulin, respectively. Similarly, for the entire study period (2012–2013), the average annual total (dry + wet) deposition flux of total PCDD/Fs-WHO2005-TEQ in the terms of pg WHO2005-TEQ m year at Hengchun (824.9) was 3.5 times higher than those in Lulin (232.0). On the basis of total PCDD/Fs-WHO2005-TEQ, the mean fraction contributed by dry deposition at Hengchun were 78.2% and 78.9% in 2012 and 2013, respectively, while at Lulin the average fraction contributed by dry deposition at Lulin were 56.5% and 69.1% in 2012 and 2013, respectively. Higher chlorinated congeners OCDD, OCDF, 1,2,3,4,6,7,8-HpCDF and 1,2,3,4,6,7,8-HpCDD dominated in terms of mass fractions of total PCDD/F deposition fluxes for all seasons at both sites. However on the basis of total PCDD/Fs-WHO2005TEQ deposition fluxes, the most dominant congeners were 2,3,4,7,8-PeCDF, 1,2,3,7,8-PeCDD and 2,3,7,8-TeCDD during the whole sampling period at both sites. The results of this study provide useful information for both environmental impact assessment and control strategies of persistent organic compounds (POPs).


INTRODUCTION
Polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs) which belong to the dioxin family (Schecter et al., 2006) continue to be accorded increasing concern due to their toxic and ubiquitous and persistence nature in the environment (Lohmann and Jones, 1998;Oh et al., 2001;Chiu et al., 2011;Huang et al., 2011a, b).
The 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) is arguably one of the most toxic chemicals (Van den Berg et al., 1998;Mitrou et al., 2001;) and the rest of dioxin like chemicals including other of PCDD/F congeners and polychlorinated biphenyls (PCBs), which occur as mixtures in commercial and environmental samples (Lee et al., 2003) are assigned toxic equivalent factors (TEFs) relative to the 2,3,7,8-TCDD (Lohmann and Jones, 1998) according to the WHO 1998 scheme ( Van den Berg et al., 1998), which was revised later to the WHO 2005 scheme (Van den Berg et al., 2006).These TEFs are multiplied with corresponding congener concentration to give toxic equivalents.
Exposure to PCDD/Fs occurs mainly through accumulation up the food chain considering that PCDD/Fs are highly bioaccumulative (Correa et al., 2006).Additionally, high exposure incidents such as industrial accidents and spillages are also responsible for human exposure (Kulkarni et al., 2008).These compounds are classified as endocrine disruptors and act as AhR agonists that bind to AhR (Schecter et al., 2006), and cause AhR-mediated biochemical and toxic responses in animals and occur as mixtures of several congeners in the environmental matrices.From occupational and epidemiological studies, the potential health effects of PCDD/Fs have been shown to include mutagenic and carcinogenic potentials, genetic and endocrine disruptions, reproductive and developmental abnormalities (Schecter et al., 2006;Kulkarni et al., 2008;Huang et al., 2011a).
Generally, as with most of the emerging persistent organic pollutants (POPs) in the environment, the PCDD/Fs are mostly emitted from anthropogenic activities (Moon et al., 2005;Kulkarni et al., 2008;Wang et al., 2010;Huang et al., 2011a).Combustion processes (Correa et al., 2006;Shih et al., 2006;Chiu et al., 2011) involving chlorine containing organic matter (Wang et al., 2009) such as municipal solid waste incineration (Lee et al., 2003), laboratory waste incinerators (Liao et al., 2014;Wu et al., 2014), hazardous waste incinerators (Wang et al., 2014), iron ore sintering plants (Fang et al., 2011), steel and cement kilns (Van Thuong et al., 2014), landfill fires (Moussiopoulos et al., 2006), vehicle tailpipe exhausts (Chuang et al., 2010), boilers (Chen et al., 2011), fast food restaurants (Lin et al., 2011), coal fire-power plants (Lin et al., 2010b), bio-mass open burning (Shih et al., 2008;Chang et al., 2013;Chang et al., 2014) and dust storms (Chi et al., 2014) have been pointed out to be among the leading primary sources of PCDD/Fs and other dioxin like compounds.The emission of PCDD/Fs from the mentioned processes can be attributed to either their presence the feedstock, or be formed from chlorine containing precursors compounds as well as be formed through the de novo mechanism at certain temperature window (Wang et al., 2010;Fujimori et al., 2014).PCDD/Fs are also classified as semi-volatile organic compounds (SVOCs) and when they are released into the atmosphere, they are ultimately subjected to atmospheric dispersion and deposition processes (Halsall et al., 1997;Shih et al., 2006) to the environmental compartments or sinks such as water bodies, soil and vegetation (Halsall et al., 1997;Ren et al., 2007).By virtue of being SVOCs, PCDD/Fs will partition between gas and particulate phases in the atmosphere, which governs their subsequent fate (Kaupp et al., 1994;Lohmann et al., 2000).Removal of PCDD/Fs in the atmosphere can occur through both dry and wet deposition processes (Fang et al., 2011;Mi et al., 2012) as well as destruction and degradation of gaseous PCDD/Fs via photolysis and OH radical induced chemical reactions (Lohmann et al., 2000;Moon et al., 2005).This study is focus mainly on atmospheric deposition including dry and wet deposition processes.
Dry deposition is responsible primarily for removal of PCDD/Fs bound to coarse particles (Moon et al., 2005) as well as small fractions bound to fine particles and gaseous phase from the atmosphere during non-precipitation days (Giorgi, 1988).Mechanisms such as turbulent diffusion, sedimentation, Brownian motion, interception, inertial forces, thermophoresis, electrical migration and diffusiophoresis are ways in which dry deposition occurs.The governing factors of dry deposition process include wind, temperature, humidity, size and shape of particles as well as the surface characteristics of both particles and sink media (Giorgi, 1988;Oh et al., 2002;Moon et al., 2005;Wang et al., 2010;Fang et al., 2011;Amodio et al., 2014).According to (Shih et al., 2006) dry deposition process accounts for approximately 15% of the total deposition flux of the higher chlorinated homologues and half of the flux of the most volatile homologues.
Three methods of determining dry deposition fluxes include measuring ambient air concentrations and the associated deposition velocities, using mass balance approach as well as employing surrogate surfaces to directly measure the fluxes (Tasdemir et al., 2004).Artifacts affecting direct measurements include blow off from particles collected on the surfaces, uncertainties of using artificial surfaces in place of natural surfaces such as water, soil and vegetation with different surface characteristics, the appropriate time for exposure, which echoes that in the environment (Halsall et al., 1997;Wu et al., 2009).This presents drawbacks into having a universally accepted method altogether (Tasdemir et al., 2004).
Monitoring dry deposition characteristics of PCDD/Fs in different geographic sites and areas is of great importance in understanding the fate and distribution of these pollutants in the environment as well as the estimating potential exposure to the populations living in these locations.Previously, various studies have been done to elucidate PCDD/F dry deposition scenarios in an industrial coastal city in South Korea (Fang et al., 2011), Manchester and Cardiff cities in the UK (Halsall et al., 1997), in the vicinity of municipal solid waste incinerators in South Taiwan (Wu et al., 2009), commercial and residential districts of Guanzhou city in China (Ren et al., 2007), major water shed in Northern Taiwan (Lin et al., 2010a) as well as the Atlantic Ocean (Jurado et al., 2004).
Wet deposition is also the major removal mechanism of particles, especially the fine particles (Moon et al., 2005) as well as gaseous phase POPs (Melymuk et al., 2011), in atmosphere via precipitation in form of rainfall and cloud droplets, or snow in the cold weather condition (Lohmann and Jones, 1998) and is majorly responsible for the presence of higher chlorinated homologues in the environmental sinks (Shih et al., 2006;Lin et al., 2010a;Wang et al., 2010).Precipitation scavenging accounts for the majority of removing these SVOCs from atmosphere by wet deposition (Huang et al., 2011a, b).The wet deposition flux of PCDD/Fs can be obtained by combining both vapor dissolution into rain and the removal of suspended particulates by precipitation.
Studies focusing on atmospheric deposition of PCDD/Fs in mountain regions are scarce, therefore, this study aims to address the research gap by providing an insight to the characteristics of PCDD/Fs deposition fluxes in a background mountain area and compare it with a coastal region of Taiwan.This will serve as useful information for both environmental impact assessment and control strategies of persistent organic compounds (POPs).
The objectives of this study were to investigate ambient air concentrations, gas-particle partitioning in addition to dry, wet and total deposition fluxes of atmospheric PCDD/Fs at coastal and high mountain area in Taiwan via modeling and simulation.Two sampling sites, in Hengchun, Southern Taiwan and in Mount Lulin, Central Taiwan, were selected.Collected samples were analyzed for specific 17 PCDD/Fs congeners to establish atmospheric concentration level, and then gaseous and particulate concentration could be calculated.The atmospheric deposition fluxes were simulated by model calculations.Results of this study will provide long-term data of PCDD/Fs in Taiwan.

Sampling Sites
Samples used in this experiment were obtained from two astronomical observatory sites in Taiwan, which include the Hengchun coastal area (22°06′N 120°70′W, 6 m elevation above sea level) in southern tip of Taiwan and Lulin Atmospheric Background Station (23°47′N 120°87′E, 2862 m above sea level) in central Taiwan (Fig. 1).
The sampling sites were chosen, since there are few studies focusing on the difference in atmospheric deposition patterns between coastal and high mountain areas of Taiwan.Sampling period was from February 2012 to April 2013.

Meteorological Conditions and PM Concentration during the Sampling Periods
The pollutant transmissions and deposition in the atmosphere are affected by the meteorological conditions, such as wind speed, rainfall intensity, PM 10 concentrations and the atmospheric stability.In this study, the pertinent meteorological information and PM 10 concentrations for Hengchun and Lulin, respectively, for the year 2012 and 2013 were obtained from the nearby air quality monitoring stations.The meteorological conditions prevailing in the sampling areas over the whole sampling period of 2012-2013 are summarized in the Tables 1-4.
During 2012, the monthly average atmospheric temperatures at Hengchun were between 21.1°C and 29.1°C and averaged 25.6°C, while those at Lulin were between 4.2°C and 12.9°C and averaged 9.4°C.In 2013, the atmospheric temperatures at Hengchun were between 21.0°C and 28.6°C and averaged 25.4°C, while those at Lulin were between 5.5°C and 14.3°C and averaged 10.9°C.In year 2012 and 2013, the maximum monthly average temperature occurred in July and June at both sites.The monthly average atmospheric temperatures at Hengchun are much higher than that in Lulin, since Lulin is located in a high altitude area (2862 m high) with cooler temperature than coastal area like Hengchun (6 m high).Additionally, Taiwan lies in the sub-tropical regions especially Hengchun, which is in the southern part, thus as reported by Shih et al. (2006) and Wang et al. (2010) and the temperature values reported herein are similar to those reported by Huang et al. (2011) for a rural area in Taiwan.
The Taiwan air quality monitoring network collects hourly data of PM 10 using the Wedding beta gauge recommended by the US EPA (Chang and Tsai, 2003;Tsai and Cheng, 1996).In 2012, the monthly average PM 10 concentrations at Hengchun were between 15.0 and 34.0 µg m -3 and averaged 24.3 µg m -3 , while those at Lulin were between 5.0 and 18.0 µg m -3 and averaged 8.5 µg m -3 .While in 2013, the monthly average PM 10 concentrations at Hengchun were between 19.0 and 43.0 µg m -3 and averaged 28.8 µg m -3 , while those at Lulin were between 5.4 and 20.2 µg m -3 and averaged 9.7 µg m -3 .Overall, the monthly average highest PM 10 concentrations at Hengchun occurred in the month of October (averaged at 38.5 µg m -3 ) for both years.However, at Lulin, the highest monthly average PM 10 concentrations occurred in March (averaged 19.1 µg m -3 ) in both years.
Table 1.Meteorological data for gas-particle partition simulation at Hengchun.

Year
Month Temperature (°C) PM 10 (µg m -3 ) TSP (µg m -3 ) Wind speed (m s -1 ) The annual average PM 10 concentration at Hengchun (24.9 µg m -3 ) is approximately 2.7 times higher than Lulin (9.1 µg m -3 ).The lowest monthly average PM 10 concentration For the year 2012, the wind speeds at Hengchun were between 1 and 6 m s -1 and averaged 4 m s -1 , while those at Lulin were between 2 and 6 m s -1 and averaged 4 m s -1 .As for the year 2013, the wind speeds at Hengchun were between 3 and 8 m s -1 and averaged 6 m s -1 , while those at Lulin were between 2 and 6 m s -1 and averaged 4 m s -1 .It is clear there was no significant difference in wind speed at both sampling locations.
Tables 2 and 3 illustrate the prevailing rainfall conditions during the simulation period.At both sites August had the highest amount of rainfall as well as highest days with rainfall while March was the driest month for the whole simulation period of 2012-2013.During the year 2012, the monthly rainfall intensities at Hengchun were between 1.6 mm and 769.4 mm with an annual total of 2573.4 mm, while those at Lulin were between 16 mm and 886.2 mm monthly and with an annual total of 3735.8 mm.On the other hand, during 2013, the monthly rainfall intensities at Hengchun were between 8.4 mm and 534.8 mm and with an annual total of 1860.6 mm, while those at Lulin were between 4.4 mm and 1056.6 mm monthly and with an annual total of 3503.2 mm.Overall, the highest rainfall intensities at Hengchun were 769.4 mm in June 2012 and 534.8 mm in August 2013.However, at Lulin, the highest monthly rainfall intensities were 886.2 mm in June 2012 and 1056.6 mm in August 2013, respectively.Generally, Lulin recorded higher total rainfall intensity than Hengchun due to the location in high mountain area.

Sampling Procedures and Analysis
Samples were collected, for a period of five days, using PS-1 sampler (Graseby Andersen, GA) according to the T09A method which was referred by United States Environmental Protection Agency (US-EPA).The PS-1 sampler was used to collect both gas and particle-phase compounds.Particle-phase compounds were collected by quartz fiber filter, whereas the gas-phase was collected using polyurethane foam.To evaluate contamination during sampling, one field blank was taken during the individual sampling events.Field blanks were loaded into the sampling system, but no air was drawn through them.They experienced the same handling, storage, and analysis procedures as the actual samples.
All the chemical analyses in this study were carried out in an accredited laboratory, Super Micro Mass Research and Technology Centre, in Cheng Shiu University which is certified by Taiwan EPA to analyze PCDD/Fs in Taiwan.High-resolution gas chromatograph/high-resolution mass spectrometer (HRGC/HRMS) (Hewlett Packard 6970 Series, CA, USA) was used for PCDD/F analysis.Detailed analytical procedures and instrumental parameters of PCDD/Fs given in the previous work (Wang et al., 2010).
Plotting log Kp against the logarithm of the subcooled liquid vapor pressure (P L o ), gives: P L o : subcooled liquid vapor pressure (Torr), m r : slope of a plot of log Kp vs log P L o , b r : y-intercept in a plot of log Kp vs log P L o (Lohmann and Jones, 1998).Eitzer and Hites (1989) have correlated P L o of PCDD/Fs with gas chromatographic retention indexes (GC-RI) on a non-polar (DB-5) GC-column using p,p'-DDT as a reference standard (Eitzer and Hites, 1989), and the correlation has been re-developed by (Hung et al., 2002): RI: gas chromatographic retention indexes (GC-RI), referred to Donnelly and Hale (Hale et al., 1985;Donnelly et al., 1987), T: ambient temperature (K) (Hung et al., 2002).
A complete datasets on the gas-particle partitioning of PCDD/Fs in Taiwan have been reported by (Chao et al., 2004).From their data, parameters for Eq.(1) were determined as m r = -1.29 and b r = -7.2 with R 2 = 0.94.In this study, those parameters are also used for estimating the partitioning constant (Kp) of PCDD/Fs.

Dry Deposition Fluxes of PCDD/Fs
The dry deposition fluxes of PCDD/Fs in the atmosphere is a combination of both gas-phase and the particle-phase fluxes, which is given by: F T : the summation of PCDD/F deposition fluxes from both gas and particle phases, F g : the PCDD/F deposition flux contributed by the gas phase (Wang et al., 2010), F p : the PCDD/F deposition flux contributed by the particle phase, C T : the measured concentration of total PCDD/Fs in the ambient air, V d,T : the dry deposition velocity of total PCDD/Fs, C g : the calculated concentration of PCDD/Fs in the gas phase, V d,g : the dry deposition velocity of the gas-phase PCDD/Fs, C p : the calculated concentration of PCDD/Fs in the particle phase, V d,p : the dry deposition velocity of the particle-phase PCDD/Fs.Dry deposition of particle-phase PCDD/Fs occurs mainly via the gravitational settling.The dry deposition velocities of particle-phase PCDD/Fs (V d,p ) can be simulated by Eqs. ( 5) and (7).

Theory of Scavenging Ratios
The wet deposition flux of PCDD/Fs is a combination of both vapor dissolution into rain and the removal of suspended particulates by precipitation.The gas scavenging ratio, S g , can be estimated by S g : the gas scavenging ratio of PCDD/Fs (dimensionless), R: the universal gas constant (82.06 × 10 -6 m 3 atm mol -1 K -1 ), T: ambient temperature (K), H: Henry constant (m 3 atm mol -1 ).
C rain, dis : the dissolved-phase concentration of PCDD/Fs in the raindrop, C g : the concentration of PCDD/Fs in the gas phase.The particle scavenging ratio, S p , on the other hand, can be calculated by: S p = C rain,particle /C p (10) S p : the particle scavenging ratio of PCDD/Fs (dimensionless), C rain,particle : the particle-phase concentration of PCDD/Fs in the raindrop, C p : the concentration of PCDD/Fs in the particle phase.Total scavenging of precipitation (S tot ) is the sum of gas and particle scavenging, which can be calculated by: S tot : the total scavenging ratio of PCDD/Fs (dimensionless), Φ: the fraction of PCDD/Fs bound to particles.Due to the lack of real measured data for the particle scavenging ratios of PCDD/Fs, the values used in this study were referenced to those in Eitzer and Hites (1989) work.

Determination of Wet Deposition Fluxes of PCDD/Fs
Wet deposition is the removal of particles in atmosphere by precipitation (rainfall and cloud droplets) and precipitation scavenging accounts for the majority of removing SVOCs from atmosphere by wet deposition (Huang et al., 2011b).Wet deposition flux of SVOCs is a combination of both vapor dissolution into rain and removal of suspended particulates by precipitation (Bidleman, 1988;Koester and Hites, 1992).
The wet deposition flux of SVOCs can be evaluated as follows:

RESULTS AND DISCUSSION
Simulated Ambient Air PCDD/F Concentrations Fig. 2 shows the regression line obtained for five total PCDD/F mass concentrations values measured in this study versus the corresponding PM 10 data.There is a strong correlation (R = 0.99) between the PM 10 values and the total-PCDD/F mass concentrations.From the results in the study of Huang et al. (2011), it was also shown that there was a good correlation (R = 0.94) between PM 10 concentrations and total-PCDD/F mass concentration.Additionally, in the same study, the PM 10 concentration and particle phase PCDD/Fs contributed and controlled more than 90% of the atmospheric PCDD/F depositions on a mass basis.Therefore, based on these inferences, this study simulated concentrations data for PCDD/F by regression analysis of PM 10 versus total PCDD/F mass concentration.Monthly simulated concentrations of total-PCDD/Fs-WHO 2005 -TEQ are presented in Fig. 3
Feb  Hengchun.This is due to the fact that Lulin atmosphere has a lower PM 10 concentration.Hengchun and Lulin are both background sites with no significant anthropogenic sources existing in the vicinity.Particularly, Lulin being a mountain site had lower PM 10 and total-PCDD/Fs-TEQ concentrations due to the location above sea level and far distance from the possible South East Asia sources.At a height of 2862 m above sea level, Lulin is above the boundary layer, hence the presence of PCDD/Fs maybe from long range transport.For comparison, Chi et al. (2013) reports a range of 0.00187-0.0102pg I-TEQ m -3 for a mountain site in Northern Taiwan and a range of 0.00237-0.00374pg I-TEQ m -3 for the same southern Taiwan coastal site.In another study, Chang et al. (2013), reported ambient concentrations ranging from 0.00232-0.00428I-TEQ m -3 for Lulin site.Chang et al. (2013) attributed the presence of PCDD/Fs in these background sites to the fact that Taiwan lies in in the downstream side of biomass burning plumes originating from South East Asia.

Gas-Particle Partitioning of PCDD/Fs
The distribution of PCDD/Fs between the gas-particle phases determines their fates in the environment.When PCDD/Fs are emitted into the atmosphere, they can be partitioned between the gas and particulate phases based on their concentrations, vapor pressure, the atmospheric temperature, and the ambient air particle concentration (Hoff et al., 1996).Gas-particle partitioning was calculated using Eq.(1), while the Tables 1 and 2 illustrate the meteorological data for gas-particle partition simulation.Total suspended particulate (TSP) concentrations were evaluated using the following relationship TSP: PM 10 = 1.24:1 (Huang et al., 2011a) in Taiwan.
Monthly gas-particle partitioning of PCDD/Fs in the ambient air of Hengchun and Lulin in 2012 and 2013 are illustrated in Figs. 4 and 5.According to Figs. 4 and 5, for both Hengchun and Lulin, the highest particle-bound PCDD/Fs occurred in winter and is lowest in summer.The partitioning of PCDD/Fs in gas and particle phase is controlled by three key factors, the atmospheric total suspended particulate (TSP) concentration, the atmospheric temperature and the vapor pressure of PCDD/F congeners.Higher chlorinated PCDD/Fs primarily present in particulate phase for all seasons at both sampling sites similar to the observations of (Wang et al., 2010).Higher chlorinated PCCD/Fs, with higher molecular weight, have a lower vapor pressure.Thus, higher chlorinated PCDD/Fs are mostly present in particle phase than gas phase.The above findings are comparable with those reported in previous studies (Wu et al., 2009;Lin et al., 2010a;Huang et al., 2011a).These results are associated with different vapor pressures; a property strongly related to temperature and the main factor influencing partition of PCDD/Fs (Pankow, 1987).
Additionally, the results show that in winter at both sites, PCDD/Fs bound to particle phase were higher than that in other seasons.This is due to the temperature in winter (averaged 21.8°C at Hengchun and 6.3°C at Lulin) were lower than that in summer (averaged 28.5°C at Hengchun and 13.3°C at Lulin).As the temperature rose, all PCDD/Fs would probably evaporate from the particle phase to gas phase.Thus, at high temperatures, the PCDD/Fs mostly exist in the gas phase.Therefore, since Hengchun has a higher mean atmospheric temperature than Lulin, it can be explained that PCDD/Fs that is bound to particle phase are lower at this site.These findings are comparable with those reported earlier by Huang et al. (2011a) and Mi et al. (2012), where the PCDD/Fs bound to particle was found to be increased with decreasing temperature since as temperature increased, the higher chlorinated PCDD/Fs bound to particle phase decrease.

Dry Deposition of PCDD/Fs
Dry deposition velocities of total PCDD/Fs were selected to be 0.45, 0.52, 0.32, and 0.39 cm s -1 in spring, summer, autumn and winter, respectively (Huang et al., 2011a).Due to lack of measured data for PCDD/Fs, a selected value (0.010 cm s -1 ) is used here to simulate PCDD/Fs dry deposition flux contributed by their gas phase (Mi et al., 2012).The dry deposition velocity contributed by the particle phase was then calculated using Eqs.( 5) and (7).
Monthly dry deposition fluxes of total-PCDD/Fs-WHO 2005 -TEQ at both sites are illustrated in Fig. 6.During 2012, at Hengchun, the monthly dry deposition flux of total-PCDD/Fs-WHO 2005 -TEQ ranged between 22.7 and 82.2 pg WHO 2005 -TEQ m -2 month -1 , while those during 2013 ranged between 33.6 and 104.3 pg WHO 2005 -TEQ m -2 month -1 .The annual total dry deposition flux of total PCDD/Fs-WHO-TEQ at Hengchun were 602.2 and 767.5 pg WHO 2005 -TEQ m -2 year -1 in 2012 and 2013, respectively.In Hengchun, when the dry deposition flux was divided by the ambient air concentration, the resulting dry deposition velocities were estimated as 0.27 and 0.28 cm s -1 for 2012 and 2013, respectively and averaged 0.28 cm s -1 .

Monthly Wet Deposition Flux of Total PCDD/Fs-WHO 2005 -TEQ
Wet deposition is the combination of both vapor-phase and particle-bound SVOCs that are removed from the atmosphere by rain.The method to evaluate wet deposition fluxes of PCDD/Fs is as described earlier in methodology part.
Monthly wet deposition fluxes of total-PCDD/Fs-WHO 2005 -TEQ at both sites can be seen in Fig. 7.During 2012, at Hengchun, the monthly wet deposition flux of total-PCDD/Fs-WHO 2005 -TEQ ranged between 0.1 and 63.2 pg WHO 2005 -TEQ m -2 month -1 , while those during 2013 ranged between 0.8 and 30.1 pg WHO 2005 -TEQ m -2 month -1 .The annual total wet deposition flux of total-PCDD/Fs-WHO-TEQ concentration at Hengchun were 151.0 and 129.1 pg WHO 2005 -TEQ m -2 year -1 in 2012 and 2013, respectively.By using the annual total rainfall and the annual total wet deposition fluxes, the estimated total-PCDD/Fs-WHO 2005 -TEQ concentrations in the rain were 0.059 and 0.069 pg WHO 2005 -TEQ L -1 for 2012 and 2013, in January (0.1 pg WHO 2005 -TEQ m -2 month -1 ) 2012 and March (0.8 pg WHO 2005 -TEQ m -2 month -1 ) 2013.At Hengchun, lowest rainfall intensities were recorded in January 2012 (1.6 mm) and March 2013 (8.4 mm), thus the wet depositions were also lowest in those periods.
On the other hand, at Lulin, the highest wet deposition fluxes were in May 2012 (19.7 pg WHO 2005 -TEQ m -2 month -1 ) and April 2013 (21.5 pg WHO 2005 -TEQ m -2 month -1 ).Highest wet deposition flux obtained in these periods was due to high rainfall intensity were occurred in May 2012 (616 mm) and in April 2013 (427.6 mm).The lowest wet deposition flux occurred in October (0.6 pg WHO 2005 -TEQ m -2 month -1 ) 2012 and February (0.2 pg WHO 2005 -TEQ m -2 month -1 ) 2013 as a result of low rainfall intensities recoded in October 2012 (16 mm) and in February 2013 (4.4 mm).
The PCDD/Fs scavenging with raindrops, therefore, the wet deposition fluxes fluctuate dramatically from month to month.Wet depositions are strongly influenced by the meteorological factors such as ambient temperature, rainfall intensity and wind speed.The wet deposition fluxes of PCDD/Fs increase primarily with increasing rainfall intensity.The scavenging ratios of both sites can be seen in (Rahenderi, 2014).
Furthermore, particulate matter (PM 10 ) concentration also plays a big role in the wet deposition.Higher PM 10 concentration means that there are more particulates can be scavenged by wet deposition.For example, if we compare wet deposition at Lulin in 2012, highest rainfall occurred in June 2012 (886.2 mm).But, highest wet deposition occurred in the second highest rainfall period in May 2012 (616 mm).Considering the PM 10 concentration, in May 2012 mean monthly PM 10 concentration was higher (9 µg m -3 ) than that in June 2012 (5 µg m -3 ).Thus, more particles were scavenged in May than June, so that the wet deposition in May is higher than in June 2012.During the whole period (2012 and 2013), the mean monthly wet deposition at Hengchun (averaged 11.7 pg WHO 2005 -TEQ m -2 month -1 ) was 1.44 times higher than that at Lulin (averaged 8.11 pg WHO 2005 -TEQ m -2 month -1 ).However, the rainfall intensity at Lulin (monthly average of 301 mm) was 1.64 times higher than at Hengchun (monthly average of 184mm).Therefore, the concentrations of both PM 10 and PCDD/Fs and gasparticle partitioning of PCDD/Fs in the ambient air are also key factors affecting wet deposition of PCDD/Fs in addition to rainfall intensity.On the other hand, in this study, other parameters effect like snow or fog are not being observed in this simulation.

Monthly Total (Wet + Dry) PCDD/Fs-WHO 2005 -TEQ Deposition Fluxes
Total deposition fluxes were calculated by summation of dry and wet deposition fluxes.Monthly total deposition fluxes of total-PCDD/Fs-WHO 2005 -TEQ at both sites can be seen in Fig.During the whole periods (2012 and 2013), the average annual total (dry + wet) deposition flux of total PCDD/Fs-WHO 2005 -TEQ in the terms of pg WHO 2005 -TEQ m -2 year -1 at Hengchun (824.9) was 3.5 times higher than those in Lulin (232.0).These values are lower than 5.02-5.11ng I-TEQ m -2 year -1 reported near two municipal waste incinerators in southern Taiwan (Huang et al., 2011b) and far much lower than 27.0 ng I-TEQ m -2 year -1 in a drinking water treatment plant in Taiwan (Lin et al., 2010a).
Comparisons between total depositions simulated in this study and in other previous studies are illustrated in Table 5.It can be seen from the table that total deposition simulated in this study are much lower compared to other previous studies.

Contribution Fractions of Dry Deposition to the Total (Dry + Wet) Deposition Fluxes of Total-PCDD/Fs-WHO 2005 -TEQ
Monthly fraction contributed by dry deposition in monthly total (dry + wet) deposition fluxes of total PCDD/Fs-WHO 2005 -TEQ can be seen in Fig. 9.At Hengchun, during 2012, the fraction contributed by dry deposition ranged between 30.5% and 99.8%, while during 2013 ranged between 51.7% and 99.1%.On the basis of monthly average, the mean fraction of total (dry + wet) deposition fluxes of total PCDD/Fs-WHO 2005 -TEQ contributed by dry deposition at Hengchun were 78.2% and78.9% in 2012 and2013, respectively;however, at Lulin, during 2012, the

CONCLUSIONS
The results of this study can be summarized as follows: 1.The simulated total-PCDD/Fs-WHO 2005 -TEQ concentrations at Hengchun (coastal site) which had more anthropogenic activities, ranged between 0.0039-0.0106pg WHO 2005 -TEQ m -3 and 0.0054-0.0138pg WHO 2005 -TEQ m -3 for 2012 to 2013, respectively .For Lulin, a background (mountain) site, the PCDD/F concentrations ranged between 0.0016-0.Dry deposition fluxes at Hengchun (averaged 57.1 pg WHO 2005 -TEQ m -2 month -1 were approximately five times higher than those at Lulin (11.2 pg WHO 2005 -TEQ m -2 month -1 ).The lack of anthropogenic activities around Lulin site and the altitude of location may influence the results, which resulted in lower than those recorded at Hengchun.3. In Hengchun, the resulting dry deposition velocities were estimated as 0.27 and 0.28 cm s -1 for 2012 and 2013, respectively, and averaged 0.28 cm s -1 .At Lulin, the estimated dry deposition velocities were 0.24 and 0.19 cm s -1 for 2012 and 2013, respectively, and averaged 0.22 cm s -1 .4. In this study, the annual total wet deposition fluxes of total-PCDD/Fs-WHO 2005 -TEQ at Hengchun were 151.0 and 129.1 pg WHO 2005 -TEQ m -2 year -1 in 2012 and 2013, respectively, while the annual total wet deposition flux of total PCDD/Fs-WHO 2005 -TEQ at Lulin were 99.0 and 95.7 pg WHO 2005 -TEQ m -2 year -1 in 2012 and 2013, respectively.For the whole period (2012 and 2013), the mean monthly wet deposition at Hengchun (averaged 11.7 pg WHO 2005 -TEQ m -2 month -1 ) was 1.44 times higher than that at Lulin (averaged 8.11 pg WHO 2005 -TEQ m -2 month -1 ).However, the rainfall intensity at Lulin (monthly average of 301 mm) was 1.64 times higher than at Hengchun (monthly average of 184 mm).Therefore, the concentrations of both

F
w,p = C rain, particle × Rainfall (14) F w,T : the wet deposition flux of SVOCs from both vapor dissolution into rain and removal of suspended particulates by precipitation, F w,dis : the wet deposition flux contributed by vapor dissolution into rain, F w,p : the wet deposition flux contributed by removal of suspended particulates by precipitation, Rainfall: monthly rainfall (m).
highest dry deposition fluxes at Hengchun occurred in March 2012 (82.2 pg WHO 2005 -TEQ m -2 month -1 ) and April 2013 (104.3 pg WHO 2005 -TEQ m -2 month -1 ).While the lowest occurred in February 2012 (22.7 pg WHO 2005 -TEQ m -2 month -1 ) and August 2013 (33.6 pg WHO 2005 -TEQ m -2 month -1 ).Highest dry deposition fluxes at Lulin occurred in March (averaged 29.3 pg WHO 2005 -TEQ m -2 month -1 ) in both years; while the lowest deposition fluxes were in August (averaged 4.2 pg WHO 2005 -TEQ m -2 month -1 ) in both years.Similar to Huang et al. (2011a) the lowest dry deposition fluxes in this study occurred in summer but conversely this study reports the highest dry deposition fluxes at the onset of spring, while Huang et al. (2011a) reports the highest occurrence in winter.
PM 10 and PCDD/Fs in the ambient air are two key factors influencing wet deposition of PCDD/Fs in addition to rainfall intensity.5.At Hengchun, the estimated total-PCDD/Fs-WHO 2005 -TEQ concentrations in the rain were 0.059 and 0.069 pg WHO 2005 -TEQ/L for 2012 and 2013, respectively and averaged 0.064 pg WHO 2005 -TEQ L -1 .For Lulin, the estimated total-PCDD/Fs-WHO 2005 -TEQ concentrations in the rain were 0.026 and 0.027 pg WHO 2005 -TEQ L -1 for 2012 and 2013, respectively, and averaged 0.027 pg WHO 2005 -TEQ L -1 .The corresponding scavenging ratios of total-PCDD/Fs-WHO 2005 -TEQ for Hengchun area were 8050 and 7880 for 2012 and 2013 respectively and averaged 8015, while for Lulin, the estimated scavenging ratios of total-PCDD/Fs-WHO 2005 -TEQ were calculated as 13900 and 13000 for 2012 and 2013 respectively and averaged 13450.6. Highest wet deposition fluxes of total-PCDD/Fs-WHO 2005 -TEQ at Hengchun were observed in August (averaged 46.7 pg WHO 2005 -TEQ m -2 month -1 ) in both 2012 and 2013.Highest wet deposition flux of total-PCDD/Fs-WHO 2005 -TEQ obtained in these periods was due to tremendous rainfall intensity recorded in August for both years (averaged 629.1 mm).However, those of the lowest wet deposition fluxes occurred in January (0.1 pg WHO 2005 -TEQ m -2 month -1 ) 2012 and March (0.8 pg WHO 2005 -TEQ m -2 month -1 ) 2013.At Hengchun, lowest rainfall intensities were recorded in January 2012 (1.6 mm) and March 2013 (8.4 mm), thus the wet depositions were also lowest in those periods.

Table 2 .
Meteorological data for gas-particle partition simulation at Lulin.

Table 3 .
Prevailing rainfall condition during simulation period at Hengchun.

Table 4 .
Prevailing rainfall condition during simulation period at Lulin.
Month Rainfall Amount (mm) Days with Rainfall Days without Rainfall Relative Humidity (%) 8% in the total deposition flux, since the rainfall intensity in that period is the lowest (1.6 mm).Thus, fewer particles were scavenged by wet deposition.March 2013 at Hengchun also have the lowest rainfall

Table 5 .
Comparison of total (dry + wet) deposition fluxes of total PCDD/Fs-TEQ between previous studies and this study.