Source Analysis of Trace Metal Pollution Received at Harbor, Airport and Farmland Locations in Central Taiwan

Measured concentrations of five metallic elements (Cr, Cu, Zn, Cd and Pb) during August 2013 to July 2014 at three sites in central Taiwan were analyzed for identifying source regions and seasonal and site variations. Air mass backward trajectory analysis suggested that pollution sources were mainly from southern Taiwan and Philippine seashore areas during the spring and summer, from Korea, Japan and Hainan during the fall, and from northern China and Hong Kong during the winter. Biomass burning activities in south East Asia in spring were believed to have contributed a significant amount to the measured high elements concentrations. Two-way Analysis of variance was conducted to identify the significance of differences in measured elements concentrations between element species, season and sites. Significant differences between elements were identified at all the three sites with exceptions for Cr and Pb at the Airport site. Season differences were significant at two of the three sites; but site to site differences were found to be insignificant. The elements pollution levels in central Taiwan were in medium to high ranges in East Asia. Keyword: Total suspended particulates; Backward trajectory analysis; Two-way ANOVA; Temporal and spatial variations.


INTRODUCTION
Long-range transport of dust aerosol has severe impact on the atmospheric environment over vast areas (Huang et al., 2013b).Resuspended desert dust is an important air pollutant in both local and transboundary levels (Dobrzhinsky et al., 2012).Atmospheric particulate matter (PM) characterizes the atmospheric air quality (Vernile et al., 2013).Urban atmosphere is subject to large inputs of anthropogenic contaminants arising from both stationary (power plants, industries, incinerators, and residential heating) and mobile sources (road traffic) (Pacyna, 1984, Sweet et al., 1993, Sullivan and Woods, 2000, Melaku et al., 2008).Heavy metals and metalloids could enter various environmental media, including soil, dust, drinking water, ambient air and food (Granero andDomingo, 2002, Nadal et al., 2005), and they may also enter the human body through dermal contact, inhalation and ingestion exposure from environmental media.The major elements (Al, Ca, Fe, Mg, Na and Mn) were mainly distributed in coarse fractions, while toxic elements, such as As, Cd, Pb and Zn, were significantly enriched in the fine fractions (Hu et al., 2013).Trace metals of primary concern for human health and the natural environment include arsenic, cadmium, chromium, copper, lead, mercury, nickel, selenium, vanadium, and zinc (Dore et al., 2014).Ni and V presented moderate or heavy contamination, and were influenced by both crustal materials and anthropogenic sources (Li et al., 2013).The organic forms of these metals, in particular, are readily taken up and absorbed by biota, and can be accumulated in food chains, thus imposing health risks to human and wildlife.
Ca, Cr, Ni, Pb and Zn metals in TSP samples have different pollution levels.The spatial and temporal variations are also different for different heavy metals.There exist strong correlations between various metal species, e.g., chromium and nickel, cadmium and chromium, cadmium and nickel, cadmium and zinc, chromium and zinc (Zhao and Zhao, 2012).The above results were consistent with those of our previous studies (Fang et al., 2004(Fang et al., , 2006(Fang et al., , 2014)).Road traffic used to be the primary source for Pb in Taiwan, and now the iron and steel industry has become the main emission source since the introduction of lead-free fuel in the 1990s.Vanadium is produced almost exclusively from oil combustion, with international shipping making a major contribution to emissions (Wang et al., 2013).A likely additional contribution to atmospheric heavy metal concentrations is the wind driven resuspension of surface dusts still present in the environment from the legacy of much higher historic emissions (Dore et al., 2014).
The purposes of this study are to identify source regions of pollutants arriving at central Taiwan and their seasonal patterns using backward trajectory analysis, and to test mean concentration variance differences for metallic elements (Cr, Cu, Zn, Cd and Pb) among the four seasons (Spring, Summer, Fall and Winter) through ANOVA analysis.

Sampling Sites
Three sampling sites located on the west side of Taichung City in central Taiwan were chosen in this study (Fig. 1).Taichung is the largest city in central Taiwan with a population of 2,701,661 during 2013-14.The first sampling site -Shalu farmland site (24°12'54.9"N,120°33'34.5"E)was surrounded with about 2 acres of vegetable farmlands.No major industrial emissions existed in the nearby areas, although a temple is located in this area and some mobile emissions are expected during traffic hours.The second sampling site -Taichung Airport (TA) site was set on the roof of a three-story building inside the Gung-Ming junior high school (24°14'59.82"N,120°35'56.45"E).The sampling site is about 1 km south of Taichung Airport (TA), where about 25 airplanes took off and landed on daily basis.The site is about 10 km from the Taiwan Strait and the Taiwan II Highway is just nearby.The sampling was conducted at 10-15 m above the surface.The third sampling site -Taichung Harbor (TH) site was set on the roof (10 m above the ground) of a four-story building in the Taichung Harbor (24.293152, 120.520736), which is located on the west coast of central Taiwan.Taichung Harbor is an artificial harbor, has a total of 83 ports and occupies about 37,600 acres.The sampling site was 100 m away from the ocean water.The Taichung Thermal Power Plant (TTPP) has been developed on 281 hectares land located along the west coast of the sampling site.TTPP mainly burn coals and is the largest thermal power plant in Southeast Asia producing around 4400 MW electricity per day.The distances between these three sites are within 11 km.

PS-1 Sampler
PS-1 sampler is a complete air sampling system designed to simultaneously collect suspended airborne particles.It collects total suspended particulate matter with the maximum collection size of appropriately 100 µm (Graseby-Andersen, GMW High Volume Air Sampler).The flow rate was adjusted to 200 L min -1 in this study.The quartz filter (diameter 102 mm) with pore size of 3 µm was used to filter suspended particles in the study.The filters were first conditioned for 24 hours under an electric chamber at humidity of 35 ± 5% and temperature of 25 ± 5°C prior to both on and off weighing.Filters were placed in a sealed CD box during transport and storage processes.The sampling device and procedures are the same as the previous study (Fang et al., 2012;Huang et al., 2013a).

Dry Deposition Plate
A dry deposition plate (DDP) was used, which is comprised of a smooth, horizontal surrogate surface that provides a lower-bound estimate of dry deposition flux.The DDP consists of a smooth surface plate made of polyvinyl chloride (PVC) that measured 21.5 cm long, 8.0 cm wide and 0.8 cm thick.The DDP also contains a sharp leading edge that points into the prevailing wind.All filters were maintained in a condition of 35 ± 5% relative humidity and a temperature of 25 ± 5°C for over 24 hours.Prior to sampling use, all filters were weighed to 0.0001 gramsignificant digits (Chu et al., 2008).

Chemical Analysis
The samples were placed in an oven one night before being weighed.A quarter of the filter was cut and selected before the digestion process.The filters were cut into thin pieces and added into a Teflon cup. 3 mL of hydrochloric acid (HCl) and 9 mL of nitrate (HNO3) were mixed and then added to this cup.After that, the samples were heated at 500°C on the hot plate for two hours.After digestion on the hot plate the samples will then be filtered.After filtration, 0.2% of HNO 3 was then added to the sample solution, up to 100 ml.solution.Before ICP-AES analysis, the samples were kept at 40°C in the refrigerator.Metallic elements Cr, Zn, Cu, Cd and Pb were analyzed in this study.ICP-AES using a Perkin Elmer N0800540 Plasma Emission Spectrometer was applied to analyze the metallic elements.The analyzed conditions were set at a 30 sec delay and with an argon gas plasma flow rate at 15 L min -1 .The nebulizer flow rate was set at 0.65 L min -1 and the sample flow rate was set at 1.5 mL min -1 .The analysis procedure is the same as in a previous study (Fang et al. 2012).

Analysis of Variance (ANOVA)
A two-way ANOVA aims to compare the effects of several levels of two factors in a factorial experiment with two-way layout.It is widely used in experimental sciences, e.g., biology, psychology, physics, among others.There is a vast literature concerned with two-way ANOVA using classical F-tests (Fujikoshi, 1993).Heteroscedastic twoway ANOVA is frequently encountered in real data analysis.In the literature, classical F-tests are often blindly employed although they are often biased even for moderate heteroscedasticity.To overcome this problem, several approximate tests have been proposed in the literature (Zhang, 2012).The literature observed in ANOVA that some asymptotic procedures and the generalized F-test perform satisfactorily for a small number of treatments and/or moderate to large samples.For one-way ANOVA, a parametric bootstrap (PB) approach has been proposed as a solution.The PB approach has been applied to solve a number of problems when conventional methods are difficult to apply or fail to provide exact solutions (Krishnamoorthy et al., 2007;Xu et al., 2013).
The basic method of two-way ANOVA is as follows:

Efficiency Test
The mean value of sum of squares of deviations should be calculated, as the first step By then the efficiency test should be followed by the following equations: where the ratio, F(c), follows F(γ) distribution with ((1 -α; c -1, (c -1)(γ -1)) degree of freedom (dof).For a given significant level, α, if factor X has a significant effect on test results, otherwise, has no significant effect.This method can easily apply to factor Y.

Meteorological Data
Meteorological variables, including wind speed, wind direction, temperature, and humidity at the Long-Jing (coastal regions) sampling site in central Taiwan were monitored at 24 m above the ground by using a Model 525 Watchdog weather station (Spectrum Technologies, Inc., USA) (Fang et al., 2012).

RESULTS AND DISCUSSION
Table 1 presents the atmospheric meteorological conditions at Harbor, Airport and Farmland sampling sites in the year of 2013-2014.The average temperature was 22.9°C, 23.0°C and 23.4°C, respectively, and the average relative humidity was 74.8%, 74.1% and 72.6%, respectively, at Airport, Harbor and Farmland.The average wind direction was from east southeast, southeast, and southeast and the average wind speed was 3.45 cm s -1 , 3.45 cm s -1 and 3.45 cm s -1 , respectively.
Figs. 2 to 4 show the backward trajectories of air masses arriving at the three sites at heights of 100, 250 and 500 meters during the four seasons.Air masses arriving at Harbor, Airport and Farmland sites were mainly from southwest, northeast and northwest, respectively, in autumn (Fig. 2), from northwest, southeast and northwest in winter (Fig. 3), from west, southwest and northwest in spring (Fig. 4), and from southwest, southeast and southwest in summer (Fig. 5).
The three sites are within a small region and have similar long-range transport sources and are thus discussed together here.During the spring, air masses arriving at the three monitoring sites were mainly from southern Taiwan and Philippine seashore areas.Due to the extensive biomass burning activities in Southeast Asia, metallic pollutants concentrations were higher in this season than in other seasons.During the summer, air masses sites were mainly originated from south Asia and Philippine Sea seashore.During the fall season, the prevailing winds direction was from the southwest, implying pollutants transport from Korea, Japan and islands in the East China Sea to central Taiwan.Back trajectory analysis suggests that pollutants from Hainan of mainland China and Mongolia affected the morning site.During the winter, the northeast monsoon brought pollutants produced from industrial emissions in north China and seashore areas to central Taiwan.Back trajectory analysis also suggests pollutants sources from inland China and Hong Kong affecting central Taiwan.
Annual mean concentrations of metallic elements Cr, Cu, Zn, Cd and Pb at the three sites were on the order of 170, 80, 120, 20 and 100 (ng m -3 ), respectively.Seasonal mean concentrations of all the elements and at all the three sites are shown in Fig. 6.There were no consistent seasonal patterns between different elements or at different sites.For example, Cr had much higher concentrations in summer than in other seasons at the Airport and Harbor sites, but this was not the case at the Farmland site.In comparison, Cu had the highest season average concentrations in winter at the Harbor and Farmland sites, but not at the Airport site.However, if averaging the three sites together, the highest seasonal average concentrations occurred in summer for Cr and Cu and in winter for Zn, Cd and Pb.
Tables 2 and 3 show the two-way ANOVA analysis results at the three sampling sites during the year of 2013-2014.Taking the Airport site as an example (Table 2), the first null hypothesis was that there were no significant differences in the mean values between the five metallic elements.The calculated Fc value was 3.48 at this site, greater than the F (F (0.95, 4, 12) = 3.26) value from the distribution table.Therefore, the null hypothesis needs to be rejected.The second null hypothesis was that there were no significant differences in the mean concentrations between the differences seasons.The calculated Fr value was 0.024, smaller than the F (F (0.95, 3, 12) = 3.49) value from the distribution table.Therefore, the null hypothesis is accepted.Thus, we can conclude that at this site there were significant differences in metallic elements mean concentrations, but there were no significant differences in the mean concentrations between the different seasons.At the other two sites, both null hypotheses are accepted, implying significant differences in mean concentrations between the elements and between the seasons.
In ANOVA analysis conducted for annual average connotations (Table 3), the first null hypothesis was that there were no significant differences in the mean values between the five metallic elements and the second null hypothesis was that there were no significant differences in the means concentrations between the different sampling sites.The calculated Fc value was 13.38 greater than the F (F (0.95, 4, 12) = 3.88) value from the distribution table, and the calculated Fr value was 0.093, smaller than the F (F (0.95, 2, 12) = 4.46) value from the distribution table.The first null hypothesis is rejected and the second is accepted.In other words, significant differences in the mean concentrations existed between the five metallic elements, but not between the three sampling sites.
To verify the above conclusions and to gain further insights, t-tests were also conducted for various scenarios.
To investigate site differences, 15 cases (5 element species × 3 pair of the sites) were conducted.For Cr cases  647.Thus, no significant differences in mean Zn, Cd and Pb concentrations were found between the three sites.In summary, the above T tests analysis indicated that there were no significant differences for any metallic element between the different sampling sites except for Cr between Airport site and the other two sites.
To investigate elements differences, 30 cases (10 pairs of the elements species × 3 sites) were conducted.Significant differences were found for mean metallic element concentrations at all the sampling sites for all metallic elements in mean concentrations comparisons.The only exceptions are for Cr and Pb at the Airport site.Cr and Pb displayed insignificant differences in mean concentrations comparisons at the Airport site.To investigate seasonal differences, 30 cases (10 elements species × 3 pairs of the seasons) were conducted for each site.At the airport site, no significant seasonal differences were found in most cases except for Cr (spring vs. fall) and Pb (spring vs. fall, fall vs. winter).At the Harbor site, no significant seasonal differences were found in most cases except for Cr (spring vs. summer).And at the Farmland site, no significant differences in most cases except for Pb in all the seasons.Similar ttests were also conducted for each season.Both significant and non-significant differences were found in elements mean concentrations in each individual season, as can also be seen in Fig. 6.
Data collected in this study are compared with literature reported values for the period of 2006-2013 as shown in Table 4.The measured concentrations of most elements were in the upper range of reported values and were much higher than values reported in mainland China, other countries in  (Giorio et al., 2013) Nice (France) 2009 --94.5 ± 22.9 64.8 ± 15.2 0.331±0.15511.4± 5.19 (Fabretti et al.,2009) Northern Belgium (Belgium)  (Perrone et al., 2013)

Fig. 1 .
Fig. 1. geographical location for Taichung Airport, Taichung Harbor and Sha lu farmland characteristic sampling sites in central Taiwan.
(a) Airport vs. Harbor, (b) Airport vs. farmland, and (c) Harbor vs. farmland, T statistic values of -2.136, -1.001 and -0.581 were obtained, respectively, suggesting significant differences in the sample population means in cases (a) and (b), but not in case (c).Similarly, T statistic values of -0.695, -1

Table 1 .
Meteorological conditions at the three sampling sites during the sampling period.

Table 2 .
Two-way ANOVA analysis results for metallic elements concentrations during the four seasons at the three sites (Airport, Harbor and Farmland).

Table 3 .
Two-way ANOVA analysis results for annual average metallic elements concentrations at the three sites (Airport, Harbor and Farmland).

Table 4 .
Comparisons of atmospheric metallic elements average concentrations at various sampling sites during 2006-2015.