Sensitivity Analysis of Atmospheric PM 2 . 5-Bound Content and Dry Deposition of Total PCDD / Fs-TEQ : In the Case of Xiamen and Zhangzhou , China

This study investigated atmospheric PM2.5 concentration, PM2.5/PM10 ratio, total PCDD/Fs concentration, PCDD/F phase distribution, PM2.5-bound total PCDD/Fs-WHO2005-TEQ content, and dry deposition of PCDD/Fs for Xiamen and Zhangzhou Cities during 2015–2017, and sensitivity analysis of both atmospheric PM2.5-bound total PCDD/Fs-WHO2005TEQ content and dry deposition of PCDD/Fs in these two cities. During 2015–2017, the three-year average concentration of PM2.5 in Xiamen was 27.6 μg m, while that of Zhangzhou was 33.9 μg m; this level is still higher than the WHO annual PM2.5 standard (10.0 μg m). In addition, the summer PM2.5-bound total PCDD/Fs-WHO2005-TEQ content in Xiamen and Zhangzhou Cities was 0.131 ng-WHO2005-TEQ g and 0.161 ng-WHO2005-TEQ g. And it is lower than the average of the other three seasons. In Xiamen, the average monthly dry deposition flux in these three years was 322 pg WHO2005-TEQ m month, while that of Zhangzhou was 378 pg WHO2005-TEQ m month, respectively. Sensitivity analysis of atmospheric PM2.5-bound total PCDD/Fs-WHO2005-TEQ content showed that the most sensitive parameters are total PCDD/F mass concentration and PM10 concentration, followed by atmospheric temperature and PM2.5 concentration; in addition, the sensitivity analysis of atmospheric dry deposition is similar to those of atmospheric PM2.5bound total PCDD/Fs-WHO2005-TEQ content. The results of this study provide useful information for better understanding PM2.5, particle-bound PCDD/Fs content and PCDD/Fs dry deposition in the ambient air of urban cities.


INTRODUCTION
Particulate matter (PM), polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDD/Fs) in the ambient air have precipitated great public concern because studies have shown that they are significantly associated with the incidence of lung and heart disease (Ito et al., 2006;Wang et al., 2018).Particulate matter (PM) is a suspension of solid or liquid particles in the form of a type of an aerosol in the atmosphere (Ghosh et al., 2014).PM can be divided into TSP (ranging from 0 to 100 µm), PM 10 (ranging from 0 to10 µm) and PM 2.5 (ranging from 0 to 2.5 µm) according to the aerodynamic diameters (Lu et al., 2016;Tang et al., are extremely dangerous chemicals, and their main routes of enter the human body are ingestion, inhalation and skin contact (Chen et al., 2010), Thereby causing damage to the immune system and even leading to cancer (Lin et al., 2010;Chi et al., 2011).The main ways in which PCDD/Fs are released into the environment are the combustion process and some industrial activities such as metal smelting processes and waste incineration (Wang et al., 2003;Hsieh et al., 2009;Chuang et al., 2010Chuang et al., , 2011)).PCDD/Fs are complex mixtures of different congeners, and 17 of these have been proved to be very toxic, with the 2,3,7,8 positions attached by chlorine atoms being the most toxic, where the toxicities are estimated by the toxicity equivalents (TEQ) (Cheruiyot, et al., 2016).After being discharged from the combustion equipment, PCDD/Fs are distributed in the gas phase and in the particle phase in the atmosphere (Chen et al., 2011a).Many studies have shown that the gas-particle partitioning of PCDD/Fs is determined by vapor pressure and ambient temperature, as well as other parameters (Wu et al., 2009a;Wang et al., 2010;Cheruiyot et al., 2015).As the temperature rises, more PCDD/Fs volatilizes into gas (Oh et al., 2001).Chemical and photochemical reactions can degrade PCDD/Fs, but their removal depends mainly on atmospheric deposition (Chi et al., 2009;Wu et al., 2009a;Huang et al., 2011a;Mi et al., 2012).
Dry deposition is one of the major routes by which air pollutants enter the ecosystem (Mi et al., 2012).Dry deposition of PCDD/Fs is the sum of the gas phase sedimentation and particle phase sedimentation, while the methods by which dry deposition occurs includes turbulent diffusion, sedimentation, inertial forces, electrical migration and diffusion electrophoresis (Zhu et al., 2017).The factors governing the dry deposition process include wind, temperature, humidity, and the size and shape of the particles, as well as the surface characteristics of both the particles and sink media (Chandra et al., 2015).
This study investigated the PM 2.5 concentrations, PM 2.5 /PM 10 ratios, PCDD/F concentrations, gas-particle partitioning, PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content, and dry deposition of total PCDD/Fs-WHO 2005 -TEQ.In addition, a sensitivity analysis of both PM 2.5 -bound content and dry deposition of total PCDD/Fs-WHO 2005 -TEQ was also conducted and discussed.

METHODS
Two cities, Xiamen (24°23′N, 117°53′E) and Zhangzhou (23°48′N, 117°34′E) in Fujian province, China, were evaluated in this study.Monthly average concentrations of PM 2.5 and PM 10 from two urban air quality monitoring stations and statistical yearbooks, as well as monthly temperatures and precipitation.The total PCDD/F concentration was simulated using a regression analysis of the PM 10 concentration.Tang et al. (2017) reported that there is a high correlation between PM 10 values and total PCDD/F mass concentrations (Wang et al., 2018).included the following two regression equations: (1) Y 1 , Y 2 : total PCDD/F concentration (pg m -3 ); x: PM 10 concentration in ambient air (µg m -3 ).The total PCDD/F concentration value is the average of Y 1 and Y 2 .

Gas-Particle Partitioning
The PCDD/F concentrations in the gas and particle phases, respectively, were calculated using a gas-particle partitioning model, as in Eq. (3) (Yamasaki et al., 1982;Pankow, 1987;Pankow andBidleman, 1991, 1992): (3) K p : temperature-dependent partitioning constant (m 3 µg -1 ); TSP: concentration of total suspended particulate matter, which was multiplied by the PM 10 concentration with 1.24 (µg m -3 ); F: concentration of the compounds of interest bound to particles (pg m -3 ); A: gaseous concentration of the compound of interest (pg m -3 ).Plotting log K p against the logarithm of the subcooled liquid vapor pressure, P L 0 , gives: Complete datasets on the gas-particle partitioning of PCDD/Fs in Taiwan have been reported (Chao et al., 2004), with the values m r = -1.29 and b r = -7.2 and with R 2 = 0.94.These values were used in this study to establish the partitioning constant (K p ) for PCDD/Fs.

Atmospheric Dry Deposition of PCDD/Fs
The atmospheric dry deposition flux of PCDD/Fs is a combination of both gas-and particle-phase fluxes, which are given by: where, F d,t is the dry deposition flux of total PCDD/Fs contributed by both the gas-and particle-phases, and, F d,g and F d,p are the PCDD/Fs dry deposition flux contributed by the gas and particle phases, respectively.C p and C g are the calculated concentrations of PCDD/Fs in the particle and gas phases.C T is the measured concentration of total PCDD/Fs in the ambient air; V d,T is the dry deposition velocity of total PCDD/Fs, and V d,g and V d,p are the dry deposition velocities of the gas-and particle-phases, respectively.
In this study, the mean dry deposition velocity of total PCDD/Fs (V d,T = 0.42 cm s -1 ) was as proposed by Shih et al. (2006).Due to the lack of measured data for PCDD/Fs, a selected value (0.010 cm s -1 ) for the gas-phase PAH dry deposition velocity, as proposed by Sheu et al. (1996) and used by Lee et al. (1996) was used in this study.

PM 2.5 Concentration
The PM 2.5 concentration not only has a significant correlation with air visibility and human health, but can also reflect the PCDD/F concentration in a given region.The monthly average PM 2.5 concentrations in the ambient air of Xiamen and Zhangzhou in 2015 and 2017 are shown in Figs.1(a), 1(b) and 1(c), respectively.In the case of Xiamen, the three-year data indicated that the lowest concentration of PM 2.5 occurred in 2017, and the monthly   average PM 2.5 concentration ranged from 13.0 to 39.0 µg m -3 and with an average of 26.4 µg m -3 .During 2016, the range was between 14.0 and 41.0 µg m -3 and with an average of 27.4 µg m -3 .In 2015, the monthly average PM 2.5 concentration ranged between 20.0 and 49.0 µg m -3 , with an average of 29.1 µg m -3 .Comparing the annual average PM 2.5 concentration, we can infer that the highest concentration occurred in 2015, and the lowest occurred in 2017.Compared with the average concentrations from 2015 to 2017, PM 2.5 concentration was reduced by approximately 10.2%.In general, Xiamen's three-year average PM 2.5 concentrations ranged from 13.0 to 49.0 µg m -3 , with an average of 27.6 µg m -3 .The concentration of PM 2.5 was less than that of cities in central and northern China such as Handan and Bengbu (Wang et al., 2018;Zhao et al., 2018).This is because the economic pillar of Xiamen is mainly two major industries -electronics and machinery, followed by a third service-based industry, which does not emit very much pollution.Xiamen itself is surrounded by the sea, and its forest coverage is high (45%), which also resulted in a lower PM 2.5 concentration.It can also be seen that even if the air quality in Xiamen significantly improves, the PM 2.5 concentration in Xiamen will still be higher than the WHO air quality regulated standard (10 µg m -3 ).Therefore there are still gaps from the international standards, and efforts should still be made to improve the air quality.Zhangzhou (Figs. 1(a), 1(b) and 1(c)), is a coastal city, and its main economic pillar is the cultivation and processing of agricultural products.Therefore, the pollution emitted into the atmosphere is much less than that of the industrial cities in central and northeastern China (Handan and Bengbu).The average monthly PM 2.5 concentration ranges between 21.0 and 48.0 µg m -3 and averaged 35.1 µg m -3 during 2017; in 2016, it ranged from 23.0 to 48.0 µg m -3 , with an average value of 32.9 µg m -3 ; in 2015, the average PM 2.5 concentration ranged from 21.0 to 54.0 µg m -3 , with an average of 33.6 µg m -3 .These results show that from 2015 to 2016, the annual average concentration of PM 2.5 decreased from 33.6 µg m -3 to 32.9 µg m -3 , which was a decrease of approximately 2.1%.From 2016 to 2017, the annual average concentration of PM 2.5 increased from 32.9 µg m -3 to 35.1 µg m -3 , an increase of approximately 6.7%.In general, the concentration of PM 2.5 within the three years for Zhangzhou was between 21.0 and 54.0 µg m -3 , with an average of 33.9 µg m -3 .
As for seasonal variations, we define January, February and December as winter and define March, April and May as spring.June, July, August are defined as summer, and September, October, November are defined as fall.In Xiamen, in 2015, the average PM 2.5 concentrations in spring, summer, fall and winter were 30.3, 21.0, 26.3 and  38.7 µg m -3 , and those in 2016 were 32.0, 17.3, 25.0 and  35.3 µg m -3 ; in 2017 they were 31.0,15.7, 24.0 and 35.0 µg m -3 .In Zhangzhou, in 2015, the average PM 2.5 concentrations in spring, summer, fall, and winter were 36.0,21.3, 32.3 and 44.7 µg m -3 ; those in 2016 were 37.7, 25.3, 26.3, and 42.3 µg m -3 , and the average PM 2.5 concentrations in 2017 were 37.7, 23.3, 34.3 and 45.0 µg m -3 .This demonstrates that PM 2.5 concentration changes with the seasons, where the PM 2.5 concentrations are highest in winter and are the lowest in summer.In Xiamen, the average PM 2.5 concentration over the three years in summer was 18.3 µg m -3 , and that in winter was 36.3 µg m -3 .The values in winter were 2.0 times higher than those in summer.As for Zhangzhou, the average value over the three years of summer was 23.3 µg m -3 , which it was 44.7% lower than that in winter (44.0 µg m -3 ).
There are two main reasons for changes of in PM 2.5 concentration: pollution source emissions and meteorological conditions.Due to the effects of these two conditions, PM 2.5 concentrations have obvious seasonal variations.In winter, due to the obvious cooling of the nighttime ground radiation, the "lower atmospheric temperature" is prone to a "inversion," and the air's horizontal and vertical exchange and circulation capacities are weakened.The pollutants emitted in the air are confined to the shallow atmosphere and gradually gather into the earth.In summer, there are frequent cyclones, so, water vapor transport is good, which is conducive to precipitation.Frequent rainfall and windy weather are conducive to the diffusion and removal of PM 2.5 .Therefore, the PM 2.5 concentration in the summer is the lowest.Xiamen and Zhangzhou are both coastal cities, and there is not much industrial pollution.So, the atmospheric environment in these two cities is better than that of cities in central and northern China.

PM 2.5 /PM 10 Ratio
The PM 2.5 /PM 10 ratio can reflect the contribution of fine particulates to ambient air pollutants, thus reflecting air pollution.The monthly mean values of PM 2.5 /PM 10 in the ambient air around Xiamen and Zhangzhou are shown in Figs.2(a), 2(b), and 2(c).
The ratio of PM 2.5 /PM 10 in Xiamen each month in 2015 was between 0.48 and 0.70, with an average of 0.58.In 2016, the PM 2.5 /PM 10 ratio was between 0.43 and 0.69, with an average of 0.56.The PM 2.5 /PM 10 ratio for each   month in 2017 was between 0.48 and 0.65.The average value was 0.54.The value of PM 2.5 /PM 10 per month in Zhangzhou in 2015 was between 0.47 and 0.77, with an average of 0.59.In 2016, the monthly PM 2.5 /PM 10 ratio was between 0.41 and 0.61, with an average of 0.49.The PM 2.5 /PM 10 ratio each month in 2017 was between 0.43 and 0.72, and the average value was 0.58.In general, the PM 2.5 /PM 10 ratio in Xiamen during the three years under consideration ranged from 0.43 to 0.70, with an average of 0.56.The three-year PM 2.5 /PM 10 ratio in Zhangzhou was between 0.43 and 0.77 with an average of 0.55.In Xiamen, the three months with the highest PM 2.5 /PM 10 ratios in 2015 were January, February, and March.Their averages were 0.68, 0.70, and 0.63, respectively.In 2016, the months with the highest average PM 2.5 /PM 10 ratio were 0.64 in January and February and 0.69 in March; the highest average for the three-month period for the PM 2.5 /PM 10 ratio in 2017 was 0.65 in January; in February, it was 0.64, and in September, it was 0.58.However, in 2015, the lowest average PM 2.5 /PM 10 concentration ratio for the three months was April, and in November and December, the values were 0.53, 0.48, and 0.52, respectively; in 2016, the three lowest average PM 2.5 /PM 10 months were June (0.44),July (0.43) and October (0.51).In 2017, the lowest four months average PM 2.5 /PM 10 ratios were 0.49 and 0.48 in July, and 0.50 in October and November.
In Zhangzhou, the three months with the highest PM 2.5 /PM 10 ratio in 2015 were January (0.75), February (0.77) and March (0.67);In 2016, the highest three-month PM 2.5 /PM 10 ratio average was 0.61 in January and 0.54 in February and March; in 2017, the three months with the highest PM 2.5 /PM 10 ratios were January, February, and June, for which the averages were 0.69, 0.72, 0.66, respectively.In contrast, the three months with the lowest PM 2.5 /PM 10 concentration ratios in 2015 were June (0.48),July (0.49), and August (0.47).In 2016, the three months with the lowest average PM 2.5 /PM 10 concentration ratios were May, August, and October, with values of 0.42, 0.41, and 0.41, respectively.The average value of PM 2.5 /PM 10 for the lowest three months of 2017 were 0.49 in July, 0.46 in September and 0.43 in October.
In general, the increase in the PM 2.5 /PM 10 ratio was found to be accompanied by higher PM 2.5 concentrations.This conclusion mainly proves that PM 2.5 is the main component in atmospheric particles, and previous research (Tang et al., 2017) has also concurred with this result.These atmospheric particles are mainly attributed to the conversion of gas particles, undergoing coagulation and flocculation processes, and agglomerating PM 2.5 .

Total PCDD/Fs-WHO 2005 -TEQ Concentrations in Ambient Air
The      These results indicate that the lowest PCDD/Fs-WHO 2005 -TEQ concentrations in both cities occurred in summer, which means that the level of particulate matter will affect total PCDD/Fs-WHO 2005 -TEQ concentrations.Therefore, higher PCDD/Fs-WHO 2005 -TEQ concentrations are always accompanied by higher concentrations of particulate matter.Therefore, controlling the source of PM emissions subsequently leads to a reduction in the level of environmental dioxins.

Gas-Particle Partitioning of PCDD/Fs
The gas-particle partitioning of PCDD/Fs plays an important role in dry and wet atmospheric deposition.Several factors are important here, including ambient temperature, PCDD/F concentration, atmospheric pressure, and atmospheric particulate concentration (Hoff et al., 1996).The calculation of the gas-particle partitioning is based on the meteorological data and the Eqs.(3), ( 4) and (5), as well as the seasonal total PCDD/Fs-WHO 2005 -TEQ in the atmosphere of Xiamen and Zhangzhou during the period from 2015-2017, for which the gas-particle partitioning is shown in Figs.4(a), 4(b) and 4(c).
In Zhangzhou, the average seasonal temperatures in spring, summer, autumn, and winter were 22.3, 29.7, 25.2 and 15.7°C, respectively; in 2016, the seasonal mean temperatures in spring, summer, autumn, and winter were    PCDD/Fs in summer was higher than that in winter, which may have been due to the fact that lower molecular weight PCDD/Fs usually have higher vapor pressure.With an increase in the ambient temperature, the proportion of PCDD/Fs in the gas phase also increases.The temperature decreases, and some PCDD/Fs are exchanged and transferred to the particle phase.As a result, lower molecular weight PCDD/Fs mainly exist in the gas phase, and the fraction of gas phase PCDD/Fs increases with increases temperature.
The results showed that the average of PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content in summer was lower than the other three seasons.This is because the average ambient temperature in summer is higher than the other three seasons, and more PCDD/Fs combine with the particles and evaporate into the gas phase, so the PM 2.5bound total PCDD/Fs-WHO 2005 -TEQ contents is decreases.

Dry Deposition
Dry deposition fluxes of PCDD/Fs were calculated based on Eqs. ( 6) and ( 7).The terms C g and C p were determined based on the gas-particle partitioning shown in the previous section, after when the unknown V d,p could be calculated, and the monthly dry deposition fluxes in In Xiamen, the average dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ in 2015 ranged between 214 and 443 pg WHO 2005 -TEQ m -2 month -1 , and the annual dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ was 3914 pg WHO 2005 -TEQ m -2 year -1 , which was about 2.96 times lower than that of the same year in Handan (11590 pg WHO 2005 -TEQ m -2 year -1 ) (Zhao et al., 2018)    As for Zhangzhou, in 2015, the monthly average dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ ranged from 273 to 443 pg WHO 2005 -TEQ m -2 month -1 , and the annual dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ was 4168 pg WHO 2005 -TEQ m -2 year -1 , which about 56.9% lower than that of in Kaifeng in the same year (9680 pg WHO 2005 -TEQ m -2 year -1 ) (Zhao et al., 2018)

Sensitivity Analysis
As sensitivity analysis can provide a basis for confirming some important parameters for PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ contents.For example, in this study, it was taken into account that ambient temperature, PM 2.5 concentration, PM 10 concentration, and total PCDD/F mass concentration may affect PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ contents.In the case of Xiamen, sensitivity analyses were carried out depending on the ambient air temperature = 22.3°C, PM 10 = 50.0µg m -3 , PM 2.5 =27.0 µg m -3 , and total-PCDD/F mass concentration = 0.68 pg m -3 .The parametric sensitivity for the atmospheric PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content in Xiamen and Zhangzhou are shown in Figs.7 and 8.The parametric sensitivity for the dry deposition fluxes of total PCDD/Fs-WHO 2005 -TEQ in Xiamen and Zhangzhou are shown in Figs. 9 and 10.P: initial value of parameters; ΔP: increase or reduction in parameters; S: predicted value in each of the parameters at the initial value; ΔS: response in each of the parameters.
The sensitivity analysis demonstrated that the most two sensitive parameters for atmospheric PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ contents were total PCDD/F mass concentration and PM 10 .In regard to the total PCDD/F mass concentration parameter, when ΔP/P was changed from 0% to +60%, ΔS/S responded from 0% to +31%.However, when ΔP/P was changed from +60% to +100%, ΔS/S responded from +31.2% to 6.82%.This may be because PCDD/F mass concentrations are the root cause of total PCDD/Fs-WHO 2005 -TEQ, so a change in PCDD/F mass concentration has a significant effect on PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ contents.When the impact of the rate of change reaches its peak, the impact on PM 2.5bound total PCDD/Fs-WHO 2005 -TEQ decreases accordingly.
It can also be seen in Fig. 7, that influence of the PM 10 concentration on the PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content is more complicated than is the case for parameters.When ΔP/P was changed from 0% to +20%, ΔS/S responded from 0% to +27%.When ΔP/P was changed from +20% to +40%, ΔS/S decreased from +27% to -0.63%.The sensitivity analysis indicated that air temperature has an effect on atmospheric PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ contents, where ΔP/P was increased from 0% to +40%, ΔS/S responded from 0% to -96%.However, a decrease in air temperature has an effect on the atmospheric PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content.When ΔP/P was decreased from 0% to -40%, ΔS/S also decreased 0% from -27%.Temperature influenced the atmospheric PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content by altering the gas-particle partitioning of PCDD/Fs, where the high molecular weight PCDD/Fs make a significant contribution to the total PCDD/F mass concentration and mainly exist in the particle phase.When the air temperature was increasing, as more particles bound to PCDD/Fs evaporated to the gas phase, PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content was significantly reduced.When the temperature is relatively low, PCDD/Fs mostly remain in the particle phase, and when the temperature reduced, the rest of the gas phase PCDD/Fs change into the particle phase.
The decrease in the PM 2.5 concentration was negatively correlated with atmospheric PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ contents, and there are two stages for the change of the effect of PM 2.5 concentration: when ΔP/P changed from -60% to +40%, ΔS/S becomes -100% to +17%, but when ΔP/P changed from +40% to +100%, the ΔS/S response was from +17% to -34%.The effect of PM 2.5 concentration on the PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content was mainly represented by particlebound PCDD/F.The lower PM 2.5 concentration means that the atmospheric stability will be better, which is conducive to the diffusion and migration of air pollutants, so the PCDD/F also has a significant decline.The PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content is the total PCDD/Fs-WHO 2005 -TEQ divided by PM 2.5 concentration, PM 2.5bound total PCDD/Fs-WHO 2005 -TEQ content also decreased slowly.Since PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content is the ratio of the total PCDD/Fs-WHO 2005 -TEQ concentration calculated by dividing by PM 2.5 , the higher PM 2.5 concentration is always accompanied by lower air temperature.The ratio of PM 2.5 /PM 10 is high, which means that the rate of increase of PM 2.5 concentration is greater than that of PM 10 , but also higher than that of PM 10 .The increase in the concentrations of PCDD/F and the rate of PM 2.5 concentration is greater than that of total PCDD/Fs-WHO 2005 -TEQ concentration, therefore PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content decreases when the PM 2.5 concentration continues to rise.
In Zhangzhou, sensitivity analyses based on the values of ambient air temperature = 23.1°C,PM 10 = 58.0µg m -3 , PM 2.5 = 32.0 µg m -3 and total-PCDD/F mass concentration = 0.75 pg m -3 .The parametric sensitivity for the atmospheric PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ contents of Zhangzhou is shown in Fig. 8.The sensitivity analysis demonstrated that the most sensitive parameters for atmospheric PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ contents was PM 10 .In regard to the parameter of total PCDD/Fs concentration, when ΔP/P was increased from 0% to +40%, ΔS/S responded from 0% to +41%.This may be because PCDD/F are the root cause of total PCDD/Fs-WHO 2005 -TEQ, so the change in PCDD/Fs mass concentration has a significant effect on PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ contents.
It can also be seen in Fig. 8, that influence of the PM 10 concentration on the PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content is very close to the total PCDD/F concentration.When ΔP/P was changed from 0% to +60%, ΔS/S responded from 0% to +47.5%.This reveals that the PM 10 concentration also has a close relationship with the total PCDD/F concentration.Their fitting curves basically coincide.
The sensitivity analysis indicated that air temperature has an effect on atmospheric PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ contents and can be divided into two stages, when ΔP/P was reduced from -100% to -40%, ΔS/S reflected from -78.7% to 18.2%.However, when ΔP/P was  increased from -40% from 40%, ΔS/S decreased 18.2% to -84.4%.The temperature influenced the atmospheric PM 2.5bound total PCDD/Fs-WHO 2005 -TEQ contents by changing the gas-particle partitioning of PCDD/Fs, the high molecular weight PCDD/F contributes a lot to the total PCDD/F mass concentration, mainly in the particle phase.When the air temperature was increasing, as more particles bound to PCDD/Fs evaporated to the gas phase, PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content was significantly reduced.The decrease in the PM 2.5 concentrations was negatively correlated or positively correlated with the atmospheric PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content, when ΔP/P was decreased from 0% to -40%, ΔS/S reflected from 0% to -65.8%.There are two stages can be divided in influence of increasing PM 2.5 : when ΔP/P was increased from 0% to +40%, ΔS/S reflected from 0% to +22.6%, respectively, but when ΔP/P was increased from +40% to +100%, ΔS/S reflected from +20% to +10% conversely.The results of the sensitivity analysis showed that the sensitivity of atmospheric PM 2.5 -bound total PCDD/Fs- WHO 2005 -TEQ contents to total PCDD/F mass concentration and PM 10 concentration was very close and high, followed by air temperature and PM 2.5 concentration.In Xiamen, PM 2.5 and PCDD/F mass concentration have similar sensitivity for the dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ.The influence of PCDD/F mass concentration can be divided into two parts: When ΔP/P was decreased from 0% to -60%, ΔS/S also decreased from 0% to -77%; when ΔP/P was increased from 0% to +60%, ΔS/S reflected from 0% to +20.8%, but ΔP/P was increased from +60% to +80%, ΔS/S decreased from 20.8% to 15.3%, respectively.As for PCDD/F mass concentration, when ΔP/P was increased 0% to +60%, ΔS/S reflected from 0% to +22.7%, ΔP/P was increased +60% to +100%, ΔS/S reduced from +22.7% to +5.09%.The parameter of the air temperature has a significant effect on dry deposition flux, can be divide into two stages: when ΔP/P was increased -80% to -20%, ΔS/S also increased from -55.6% to +6.7%, when ΔP/P was altering -20% to +40%, ΔS/S reflected from +6.7% to -54.2%.When the temperature is below 17.0°C, this parameter has a positive influence on dry deposition flux, when the temperature is higher than 17.0°C, the air temperature has negative correlation with the dry deposition of total PCDD/Fs-WHO 2005 -TEQ.As for PM 10 , it can be seen that the dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ is most sensitive to PM 10 concentration.
In the case of Zhangzhou, according to Fig. 8, the sensitivity analysis indicated that the effects of PM 10 and PCDD/F mass concentration on dry deposition fluxes are very similar, both of them have a significant, positive effect on the dry deposition fluxes of total PCDD/Fs-WHO 2005 -TEQ.The parameter of PM 2.5 concentration effect on dry deposition fluxes can be divided into two parts: When ΔP/P was increased -40% to 0%, +40% to +80%, ΔS/S reflected from -63.8% to -14%, -2.3% to +82.8%; when ΔP/P was increased 0% to +40%, ΔS/S decreased from 0% to -2.3%, respectively.As for the effects of air temperature on dry deposition is very like to the city of Xiamen, there are two parts to analyze: when ΔP/P was increased -80% to -20%, ΔS/S reflected from -95.1% to +5%; when ΔP/P was increased -20% to +40%, ΔS/S responded from +5% to -67.6%, respectively.
The above analysis indicates that the parameters most sensitive to the dry deposition of total PCDD/Fs-WHO 2005 -TEQ are atmospheric PM 10 concentration and PCDD/F mass concentration, followed by PM 2.5 concentration and air temperature.

CONCLUSION
1.The PM 2.5 concentration in Xiamen ranged between 13.0 and 49.0 µg m -3 in the three-year period under investigation, with an average value of 27.6 µg m -3 .The concentration of PM 2.5 in Zhangzhou during the same period was 21.0-54.0µg m -3 , with an average of 33.9 µg m -3 .The concentration of PM 2.5 in Xiamen was lower than that in Zhangzhou.Since the two cities are close to the sea, and their heavy industry pollution is relatively low, their PM 2.5 concentrations are lower than most cities in China.About the seasonal variations, in Xiamen, the average value of the PM 2.5 concentration over the three years in the summer was 18.3 µg m -3 , and the average in winter was 36.3 µg m -3 .The values in winter were 2.0 times higher than those in summer.
In Zhangzhou, the average value over the three years in summer was 23.3 µg m -3 , which was 44.7% lower than that in winter (44.0 µg m -3 ). 2. The PM 2.5 /PM 10 ratio in the three years from 2015 to 2017 in Xiamen was in the range of 0.43-0.70,with an average of 0.56.The PM 2.5 /PM 10 ratio in the three years of Zhangzhou was between 0.43-0.77with an average of 0.55.The high PM 2.5 /PM 10 ratio is highly correlated with the higher PM 2.5 concentration.This shows that PM 2.5 is the main part of atmospheric particles.3.In Xiamen, the average PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content during the summer in the study period was 0.131 ng-WHO 2005 -TEQ g -1 , which was lower than the value of PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content for the other three seasons; in Zhangzhou, the mean value of PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content in the summer was also significantly lower than in the other three seasons.This is due to the fact that summer temperatures are higher and in summer more particles bound PCDD/Fs into vapor phase and PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content was reduced.4. In terms of the gas partitioning of PCDD/Fs, for Xiamen, the three-year average proportion of gas phase total PCDD/Fs-WHO 2005 -TEQ concentration in summer (70.0%) is much higher than the other three seasons.In Zhangzhou, the gas phase fraction in summer (66.7%) was the highest among the four seasons.In general, the most relevant parameter in the fraction of gas phase total PCDD/Fs-WHO 2005 -TEQ is ambient temperature.5.In 2015, the annual dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ in Xiamen was 3914 pg WHO 2005 -TEQ m -2 year -1 , and in Zhangzhou was 4168 pg WHO 2005 -TEQ m -2 year -1 .In 2016, the annual dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ in Xiamen (3791 pg WHO 2005 -TEQ m -2 year -1 ) was about 23.5% lower than that of Zhangzhou (4889 pg WHO 2005 -TEQ m -2 years -1 ).In 2017, the annual dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ in Xiamen was 3886 pg WHO 2005 -TEQ m -2 year -1 , and that in Zhangzhou was 4538 pg WHO 2005 -TEQ m -2 year -1 .It can be seen that the dry deposition flux in summer is lower than in the other three seasons.This is because higher temperatures cause more PCDD/Fs to evaporate into the gas phase during the summer.6.For sensitivity analysis, we can get such a result that whether it is atmospheric PM 2.5 -bound total PCDD/Fs-WHO 2005 -TEQ content or dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ are most sensitive to total PCDD/F mass concentration and PM 10 concentration, then atmospheric temperature and PM 2.5 concentration.
Fig. 4(a).Seasonal variations of gas-particle partition of total-PCDD/Fs-WHO 2005 -TEQ in the ambient air in Xiamen and Zhangzhou during 2015.

Fig. 4
Fig. 4(b).Seasonal variations of gas-particle partition of total-PCDD/Fs-WHO 2005 -TEQ in the ambient air in Xiamen and Zhangzhou during 2016.
Fig. 4(c).Seasonal variations of gas-particle partition of total-PCDD/Fs-WHO 2005 -TEQ in the ambient air in Xiamen and Zhangzhou during 2017.
. The maximum monthly average dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ of 443 pg WHO 2005 -TEQ m -2 month -1 occurred in January, and the lowest value 214 pg WHO 2005 -TEQ m -2 Month -1 , occurred in July and August, which two times lower than the highest value.In 2016, the monthly average dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ ranged between 186 and 434 pg WHO 2005 -TEQ m -2 month -1 , and the annual dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ was 3791 pg WHO 2005 -TEQ m -2 years -1 .The dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ in March of 434 pg WHO 2005 -TEQ m -2 month -1 was the maximum monthly average dry deposition flux for the year, for which the lowest level occurred in July (186 pg WHO 2005 -TEQ m -2 month -1 ), which was about 57.1% lower than that in March.During 2017, the monthly average dry deposition flux of total PCDD/Fs-WHO 2005 -TEQ ranged from 158 to 527 pg WHO 2005 -TEQ m -2 month -1 , and the 0