Long-Term Variations of Airborne Cadmium ( Cd ) Concentrations in Major Urban Areas of Korea between 1991 and 2010

The long-term (1991–2010) monitoring data for cadmium (Cd) collected from seven major cities in Korea were evaluated to assess spatial-temporal trends in urban areas and help improve air quality control. The results show that the level of Cd pollution in the different cities, as distinguished by industrial activity, was the highest in Ulsan (9.21 ± 6.29 ng/m) and Incheon (6.25 ± 6.05 ng/m). During the study period, Cd concentrations showed significant reductions in all monitoring cities, especially in the more industrialized ones (e.g., a nearly six-fold reduction in Ulsan over 13 years), while in other areas this reduction was more moderate (about 1.7-fold). In addition, Cd concentrations generally peaked in the winter, with the exception of the more industrialized areas. Although Cd levels in Ulsan and Incheon exceeded the general guidelines of 5 ng/m (e.g., those of the US EPA) until recently (e.g., 2004), its values in most urban areas in Korea have been continuously decreasing over the 20-year period examined in this work. Therefore, greater efforts are necessary to more effectively reduce Cd released from the prominent source activities in the urban environments.


INTRODUCTION
Like most airborne metallic pollutants, cadmium (Cd) is released from both natural and anthropogenic sources and present in a particle bound form in ambient air.The dominant fraction of airborne Cd generally comes from anthropogenic processes such as non-ferrous metal production, the mining and refining of zinc (and lead) ores (Bernard, 2008), the combustion of fossil fuels, and the operation of waste incinerators and power plants (USEPA, 2000;ATSDR, 2008;Avino et al., 2008;UNEP, 2010).The use of Cd can be expanded further to a stabilizer in the production of batteries, pigments, polymers, and plating (USEPA, 2000;ATSDR, 2008;Avino et al., 2008;UNEP, 2010).In our everyday activities, the intake of Cd can also occur through such activities as direct or indirect exposure to tobacco smoke (ATSDR, 2008).In contrast, terrestrial dust, wild fires, and volcanic activity are classified as the main sources of natural Cd (Avino et al., 2008;Hieu and Lee, 2010).
Cadmium (Cd) has drawn most environmental concern with respect to significant ecological risk due to high toxicity and mobility (Bi et al., 2006;Kim et al., 2003Kim et al., , 2009;;Ma et al., 2012;Ray et al., 2012).The US EPA (2011) has classified airborne cadmium as a probable human carcinogen (Group B1) listed substance (with limited evidence of carcinogenicity).Cadmium, if exposed chronically, may cause proteinuria and/or damage lungs due to bronchial and pulmonary irritation (USEPA, 2000;Glorennec et al., 2005;UNEP, 2010;USEPA, 2011).Furthermore, it can be transferred to food chain due to its high potential for accumulation in various parts of the ecosystem (soil, sediment, water, and plants) (Bernard, 2008;Aas et al., 2009).Because of great efforts with the support of advanced control technology, a constant reduction in the concentration levels of many toxic metal pollutants is observed in most of world including North America and Western Europe (e.g., Harmens et al., 2007).
To assess the long-term variability of airborne Cd levels in urban areas, we analyzed the airborne Cd concentration data acquired from the seven major cities of Korea over a 20-year period (1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010) by the Korean Ministry of Environment (KMOE).As the long-term monitoring of Cd was made at each of the seven major cities in Korea, these Cd data can be examined to assess its spatiotemporal variability in a number of respects.The results of our analysis will thus help us learn more about the fundamental characteristics of Cd in the urban environment and develop strategies to control its pollution.In our previous study, the data sets of Cd collected during the earlier period of this study (1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004) had already been evaluated (Kim, 2007).Hence, the results of this study will offer us an opportunity to assess the temporal trends of Cd over an extended period up to 2010.

MATERIALS AND METHODS
In this study, we analyzed the Cd data measured from 7 major cities (Seoul, Busan, Daegu, Incheon, Gwangju, Daejeon, and Ulsan) in Korea between 1991 and 2010 (Fig. 1).These seven major cities are well-known for their high population densities, large urban residential areas, and major industrial (and transportation) activities.The overview of the size, population, temperature, and humidity in the major monitoring areas are shown in Table 1.
To assess the extent of metal pollution in the urban atmosphere, concentrations of hazardous metal species have been measured routinely from major urban areas of Korea with large populations (e.g., above a half million) or strong industrial activities.As part of such efforts, Cd concentration data measured from seven major cities in Korea between 1991-2010 were evaluated to account for the basic status of Cd pollution over an extended period.Seoul had the largest number of monitoring stations totaling 13 in 1998 and was followed by Daejeon (7), Busan (5), Daegu (5), Gwangju (4), Ulsan (4), and Incheon (3).At each individual station, the collection and analysis of particle samples were made by following the basic analytical procedures recommended by the air quality measurement method of the KMOE.According to this method, the amounts of all metals are determined from total suspended particle (TSP) samples collected by high-volume samplers with the aid of flame atomic absorption spectrometry (FAAS) (e.g., Kim et al., 2004;Mutlu et al., 2012).The Cd concentrations of the extracted solutions were determined in reference to a standard solution of trace metals certified by the US National Institute of Science and Technology (US NIST).
These original metal concentration data are then examined further by the quality assurance (QA) procedures of KMOE and stored in its data management network system.As the part of QA procedure in the analysis of Cd (and other metals), the reliability of the experimental method was examined at least on a once-per-year basis.The environmental behavior of airborne Cd measured for the entire 20-year study period (1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010) was evaluated after grouping the datasets into two separate terms, period I (1991-1997) and period II (1998II ( -2010)), considering the difference in data acquisition patterns between the two study periods.In period I, the Cd data are available only as the annual average for each city.However, in period II, all measurement data were recorded consistently at monthly intervals at each station.As such, this period-based grouping scheme was used to evaluate the temporal patterns of Cd.
To make a parallel comparison of the data between the two periods and to analyze temporal trends over an extended period of 20 years, the monthly data in period II were first grouped together in an annual category to yield the respective annual mean values for each city.In addition, the seasonal patterns of Cd from seven individual cities were also evaluated by focusing mainly on the monthly Cd data obtained during period II.
As shown in Fig. 2, Cd values in most cities were much higher in period I than in period II.Interestingly, cities with low Cd levels (e.g., Busan, Daegu, Gwangju, and Daejeon) generally exhibited very little changes between the two periods.Ulsan and Incheon showed the highest Cd levels of all cities, and their annual means during period I generally exceeded the guideline value set by the European Union (EU) (5 ng/m 3 ).Such exceedance patterns from those 2 cities were also observed in the early part of period II.This guidance value was fixed by the Ministry of Environment, Heritage and Local Government in the EU (EU, 2009).They standardized the values of metals (including Cd) in ambient air to be observed in every European country (EU, 2009).As such, adequate information on the concentrations of metals has been made available to the public.
In Table 2, the Cd data measured at all the different cities are compared between the two time periods of I (1991-1997) and II (1998II ( -2010)), as explained above.Although temporal patterns of Cd changed considerably between the two periods, the highest values were seen consistently from Ulsan and Incheon, regardless of period.In period I, the mean Cd values peaked in Ulsan (13.14 ± 5.53) and Incheon (8.53 ± 4.98).However, these values dropped to half of those levels during period II.As a result, the concentration levels of Cd during period II fell below European Union (EU) guideline value of 5 ng/m 3 in most cities except for Ulsan and Incheon.Such a reduction in Cd levels is likely to have been achieved by the combined effects of several factors including the adoption of new emission regulations, new manufacturing methods, and recycling of waste in the industries (e.g., Mutlu et al., 2012).However, Cd levels in many countries still do not meet such guidelines despite considerable reductions achieved over the years (Mutlu et al., 2012).For instance, relatively high concentrations of Cd were measured in some countries such as Nanling Mountains and the Pearl River Delta (urban zone), Southern China: 4.6-21.6ng/m 3 during 2002-2003 (Lee et al., 2005); Cartagena (urban zone), Spain: 2.9-8.9 ng/m 3 during 1991-1998(Moreno-Grau et al., 2000); and central Taiwan (urban area): 4.3 ng/m 3 during 2001-2002 (Fang et al., 2003).

Seasonal Distribution of Airborne Cd
As the Cd data in period II were recorded on monthly basis, the results can be used to explore the seasonal patterns of Cd.Many studies have shown that the distribution of particulate matter (PM) is affected greatly by the seasonality of the meteorological conditions (Kim et al., 2003(Kim et al., , 2004;;Kim, 2007).In Fig. 3, seasonal patterns of Cd in each city have also been evaluated (1) directly by the seasonal mean values and (2) after the normalization with respect to the overall mean values.To facilitate the comparison of Cd levels between seasons, the mean for each season at a given city was divided by the overall seasonal mean value of that city.
The seasonal trends of Cd data were similar in most target cities.Like the general pattern of PM data, the lowest Cd concentration occurred in summer, while the highest did so in winter.In contrast, the mean Cd values of Incheon were slightly higher in fall than in winter.Lim et al. (2010) reported that the PM concentration (including Cd) at a Daejeon industrial area increased in winter rather than other seasons, while observing a different seasonal trend in other industrial areas.As such, temporal patterns of Cd were often distinguishable between the most cities and ones with strong industrial activities (e.g., Ulsan).Although the data from these industrial (and port) areas like Ulsan tend to peak in summer (Basha et al., 2007), the Cd values in Busan remained almost constant across seasons, showing little variation.The patterns observed in this study are slightly different from Mutlu et al. (2012) who disclosed relatively weak seasonal variation of Cd in most urban cities in Korea from 2004 to 2010.The observed differences may be explained by the selection of a more wider time band in this study (1998 to 2010) and slight differences in areal coverage (e.g., a total of 41 monitoring sites in this study vs. 30 monitoring sites for Mutlu et al. (2012)).
Seasonal patterns of Cd were also investigated at a roadside in Hong Kong from 2004-2005(Cheng et al., 2006;;Shi et al., 2012).According to these authors, the pattern of PM (including Cd) value increased dramatically in the winter season followed by spring and autumn, while the Cd values in summer showed a noticeable decreasing trend.However, there was very little deviation of PM values observed among the seasons (especially in summer and fall) as opposed to location.This similar trend was also observed in different countries in both industrial and residential areas.For instance, Cd values exhibited a greater increase in winter season than those of other seasons in Qingdao, China during 2001 to 2002 (Hao et al., 2007), and Kolkata, India during 2003 to 2004 (Karar and Gupta, 2006).On the other hand, a different trend was observed with the highest values in fall and lowest in winter during 2006 to 2007 in Washington DC, USA (Melaku et al., 2008).

Factors Controlling the Distribution of Cd
To investigate the factors controlling the behavior of Cd in the urban environment, a correlation analysis was conducted using the monthly Cd datasets measured between different cities in period II (e. g., the maximum of 156 monthly data points) for 13 years (1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010).This correlation analysis was conducted in two different respects: (1) correlations of Cd data between different cities (Table 3(A)) and ( 2) correlations between Cd and other metal species for each city (Table 3(B)).In Table 3(A), the results of correlation analysis were highly comparable to show many matching pairs between different cities with statistical significance (P < 0.01).For instance, seven out of all 21 matching pairs showed the strongest correlations (P < 0.01), while there was only one pair with moderately strong correlation (P < 0.05) ( strongest correlation (p < 0.01) of Cd was seen consistently with Pb and Ni from all seven cities.Although such strong correlations were also observed from Cu and Mn (from 6 cities), it was less abundant in the case of Cr (4 cities) and Fe (3 cities).This result thus suggests that the source properties of Cd should be tightly bound with other metals, although the level of such compatibilities may vary depending on the paired metal specie.It was reported that industrial and traffic emissions were the major contributors to the local airborne Cd (Basha et al., 2010;Fang et al., 2010).Thus, the high Cd concentrations in Ulsan and Incheon were likely to reflect the combined effects of local traffic and industrial emissions.Wu et al. (2007) also reported that vehicle exhaust was the main source of airborne metallic substances, including Cd.In contrast, average TSP concentrations decreased from the urban and industrial zones to residential areas.The same behavior is observed for Cd, Zn, and Pb, but not for Cu, which has a relatively short residence time (Morawska et al., 2001;Espinosa et al., 2002).It is interesting to note that Fe consistently maintained inverse correlations with Cd, regardless of statistical significance.According to Lee and Park (2010), Cd showed an inverse correlation with Fe in Ulsan during misty days in 2007.Similarly, Fe consistently showed inverse correlations with Pb and Cr.The observed patterns may suggest that a major crustal component like Fe exhibits behavior different from those made of an anthropogenic origin (like Cd).It is well known that Cd, Pb, and Cr generally come from various industrial activities (Pakkanen et al., 2001;Lim et al., 2010).One of the most important sources of heavy metals (including Cd) in the urban environment is also road traffic (Chandler, 1996;Ayras and Kashulina, 2000).Other contributors include waste incinerators, coal fired power plants, geogenic dust, and construction debris (Chandler, 1996;Ayras and Kashulina, 2000).The abundance and type of man-made activities can thus be a key factor leading to the alteration of Cd levels in urban and industrial area.

Long Term Variations in Cd Concentrations in Major Cities
Because the measurements of Cd had been made continuously in seven metropolitan cities in Korea over the past 20-year period, comparison of its annual mean values can be made to assess its long-term trend (Fig. 4).To conduct a long-term analysis, annual mean values of Cd obtained from all seven cities were examined by a simple linear regression analysis.The use of such simplified approach has commonly been made to provide a quick view of the general trend of many airborne pollutants (e.g., Hillery et al., 1997).If the long-term trends of Cd concentrations are compared between cities, Cd concentrations in major metropolitan cities experienced gradual reductions throughout the past decades.
A linear regression analysis of Cd was conducted to describe long-term trends of annual Cd data on a statistical basis (Fig. 5).The most significant trend (P < 10 -4 ) was seen in Seoul during the entire period (1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010), while the trends of Daegu and Gwangju exhibited the lowest significance (P < 0.8, and P < 0.04, respectively).On the other hand, Busan, Ulsan, Incheon, and Daejeon showed a moderately enhanced significance (P < 0.01).As such, most cities showed apparent reductions over the entire study period.A similar reduction pattern (~80%) was also observed in Europe during 1955-2005(Pacyna et al., 2009).In case of France, a reduction of 82% was seen from 1990 to 2006 (CITEPA, 2009).In addition, such a declining trend (~66%) was also seen in North-western Mediterranean, France from 1998 to 2008 (Heimbürger et al., 2010).In Spain, the decreasing trend of Cd was 78% during 2001 to 2005 (Vicente et al., 2012) In industry, Cd can cause acute and chronic intoxications via inhalation and ingestion.Due to the long biologic halflife of Cd (10 to 40 years) in kidney (EC, 2007), it can exhibit a salient toxicological property in human body (Bernard, 2008).As a very small amount of Cd can cause greater deleterious effects than other toxic metals (Järup et al., 1998;Bernard, 2004;Nordberg et al., 2007), the US Environmental Protection Agency (USEPA) Integrated Risk Information System (IRIS) has recommended an inhalation unit risk (IUR) of Cd (as probable carcinogen) to be 1.8 ng/m 3 (Greene and Morris, 2006;Hieu and Lee, 2010).Accordingly, the results obtained in this study indicate some contrasting patterns between the city types.In 2010, Cd levels in most of the metropolitan cities (Seoul, Busan, Daegu, Gwangju, and Daejeon) of Korea remained in a relatively constant range of 0.68-1.38ng/m 3 .On the other hand, Cd levels in Ulsan and Incheon exhibited much higher values, above 2 ng/m 3 in recent years.The results in these industrialized cities can hence be of concern if one considers some of the health hazard guidelines.

CONCLUSIONS
In this study, Cd data were analyzed to evaluate its spatiotemporal variations in the major cities of Korea.The Cd concentrations measured at seven major cities were compared at different temporal intervals: (a) over the entire period of 20 years, (b) between years, and (c) between seasons.The Cd levels in highly industrialized areas exhibited the highest mean at 9.21 ng/m 3 (Ulsan) and 6.25 ng/m 3 (Incheon) for the 20 year period, while those of other urban areas showed moderately reduced values between 1.40-2.77ng/m 3 .The results suggest that the Cd concentration levels measured from the seven cities can be distinguished to a certain degree by the level of industrial activities.
To examine the temporal patterns of Cd, the data were analyzed between different seasons and across the entire study period.The seasonal patterns showed higher Cd concentration levels in winter in most cities.However, the Cd values at the highly industrialized city of Ulsan tend to peak in summer.When the mean Cd concentrations were analyzed on long-term basis, most cities consistently exhibited decreasing trends with statistical significance.
The results of this study confirm a constant reduction in 1 9 9 2 1 9 9 3 1 9 9 4 1 9 9 5 1 9 9 6 1 9 9 7 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 Cd values in all seven major cities during most of the study period.This indicates that the efforts of the Korean government to control anthropogenic emissions of Cd have been effective in a broad sense.However, Cd concentrations in certain cities such as Ulsan and Incheon still remain high enough, exceeding ~2 ng/m 3 .Now, it is well known that Cd-containing products (batteries, alloys, coatings, pigments, etc.) are recyclable to a large extent and that the costs involved in such processes are in the economically acceptable range.The future studies are thus needed to develop more effective control methods and to tighten the individual anthropogenic emission sources of Cd along with other hazardous metal species.If such efforts are made successfully, the present level of Cd pollution observed in many industrialized areas can be reduced further.

Fig. 1 .
Fig. 1.Geographical locations of the seven metropolitan cities for the monitoring of airborne Cd concentration levels in air in Korea.

Fig. 3 .
Fig. 3. Plot of seasonal mean values of Cd between seven cities in Korea during period II (1998-2010).

(
A) The two cities with the maximum Cd concentration levels (Ulsan and Incheon) The annual Cd values of the other five cities (Seoul and others)

Fig. 4 .
Fig. 4. Annual trend of Cd concentrations measured from all seven cities during the entire study period (1991-2010).

Table 1 .
Basic information of the seven major metropolitan cities in Korea.
a Korean Statistical Information Service (Korean) as of 2 June 2010.b Korea Meteorological Administration, 2010

Table 2 .
Comparison of the annual mean concentrations of Cd in the seven major cities in Korea throughout the entire study period 1991-2010.
Kim (2007)ean value.bSDand CV denote standard deviation and coefficient of variation, respectively.cPeriodI: same as previous study ofKim (2007).d Period II is evaluated newly in the present study.e Results of t-test to check for the statistical significance of concentration differences between the study period of I and II are given as the probability (P) values.

Table 3
(A)).The Cd data in Seoul generally showed fairly strong correlations (p < 0.01) with those of Busan, Daegue, and Daejeon.Likewise, the results of Daejeon showed

Table 3 (
B), the correlation analysis was also made to check the relationship between Cd and other metals in each city.The results indicate that there are abundant cases of significant correlations (p < 0.01) between most metals.The

Table 3 .
The results of correlation analysis using monthly Cd concentration data obtained from major urban areas in Korea during periodII (1998II ( -2010)).(A)Correlation of Cd data between different cities Correlation is significant at the 0.01 level (2-tailed).* Correlation is significant at the 0.05 level (2-tailed).
(B) Correlations between Cd and other metal species in each city