Introduction

In an urban area such as Seoul, there are many potential sources leading to groundwater contamination, such as domestic effluents, septic tanks, leaky sewage systems, gasoline stations, leachate from waste disposal sites, and spillage from industrial sites (Barber et al. 1996; Subbarao et al. 1996; Eiswirth and Hotzl 1997; Sharp 1997; Trauth and Xanthopoulos 1997; Niemczynowicz 1999; Lerner 2002). The groundwater quality in urban area primarily depends on land use. Groundwater quality in developed urban settings, compared to those in less-developed settings, are characterized by increased concentrations of major and/or minor elements and the changes in oxidation–reduction condition (Cain et al. 1989; Anderson 1993; Eckhardt and Stackelderg 1995; Squillace and Price 1996; Barrett et al. 1999; Trojan et al. 2003; Choi et al. 2005).

A myriad of groundwater pollution sources exist in Seoul and are considered to be closely related to land use (Lee PK et al. 2001; Lee et al. 2003; Kim 2004; Choi et al. 2005). Thus, a good understanding of the control of groundwater contamination through land use will be most important and effective for the sustainable management of groundwater in Seoul. Choi et al. (2005) examined the distribution of major cations and anions in Seoul groundwater in relation to land use and found that total dissolved solids (TDS) can be used as an effective parameter for evaluating anthropogenic contamination. Little is known, however, about the relationships among groundwater quality, anthropogenic contamination sources, and land use in Seoul. Furthermore, contamination of Seoul groundwater by either volatile organic compounds (VOCs) or trace metals has not been investigated in relation to land use. The present work was initiated to elucidate the impacts of land use on groundwater contamination by trace metals and VOCs. However, these types of studies have many difficulties and limitations in ensuring whether a sample represents the intended land use and in working with statistically non-normally distributed data for concentrations of trace metals and VOCs, mainly due to their low detection frequency (Barringer et al. 1990; Eckhardt and Stackelderg 1995; Hudak and Blanchard 1997; Trojan et al. 2003). Non-parametric statistical analysis was used in this study to test the hypothesis that the concentrations of trace metals and VOCs are significantly affected by land use. The results of this study can be effectively used to set up a guide and policy for sustainable management and protection of groundwater in urban areas in Seoul.

Study area

Physical setting and groundwater use

Seoul Metropolitan City (the capital of South Korea) is located in the mid-west of the Korean Peninsula. Seoul currently covers an area of 605.7 km2. It measures 37 km in an East–West direction and 30 km in a North–South direction. It has undergone significant and steady expansion since 1913 (36 km2), as shown in Fig. 1a. Concurrently, the population increased enormously from 1.5 million in 1949 to circa 10 million in 2001. The current population density amounts to about 17,000 people/km2 (SMC 2001a). As a result of the rapid urbanization and population increase, land-use characteristics in Seoul has significantly changed: for example, large expansion of urban centers, progressive retreat of agricultural fields, increase in the number of high traffic roads, etc. Water demand has also significantly increased.

Fig. 1
figure 1

Maps showing the enlargement of Seoul City from 1913 to 2001 (a) and geologic setting (b)

Full size image

A monsoonal climate with an average annual temperature and precipitation of 11.8°C and 1,300 mm, respectively, is characteristic of Seoul (KMA 2002). About 70% of total annual precipitation falls typically as downpours in the summer months between June and September. The surface geology of the Seoul area consists of thin (<20 m) soils and alluviums and underlying crystalline rocks comprising granite and gneiss with minor schist (Chae et al. 2004; Fig. 1b). The gneiss of Precambrian age consists mainly of banded biotite gneiss and migmatitic gneiss. Granites of Jurassic age are the most widespread and characteristically crop out at high mountains surrounding Seoul. Alluvium occurs along the Han River and its tributaries and mainly comprises sand and silt with high permeability.

The municipal water supply in Seoul largely depends on surface water (especially, the Han River). The daily average of public water supply amounts to about 4 million m3/day, corresponding to a water supply per person of about 400 l/day. Groundwater occupies circa 10% of the water supply. About 15,000 wells were reported to be in use in 1999, yielding urban groundwater for domestic (75.7%), agricultural (14.5%), industrial (3.9%), and drinking (5.9%) purposes. The use of groundwater for drinking is continuously and rapidly increasing. The current annual yield of groundwater in Seoul is considered to amount to 23% of the estimated sustainable yield (SDI and SMC 2001; Lee et al. 2003). However, in addition to progressively increasing anthropogenic contamination, the local increase in groundwater pumping is considered as a major cause of progressive deterioration of the groundwater quantity and quality. Owing to the thin nature of soil and alluvium, groundwater in Seoul has largely been developed from fractured aquifers in crystalline rocks.

Land use

For this study, the land use in Seoul City was classified into five categories: less-developed, residential, agricultural, traffic, and industrial (Yun et al. 2000; SMC 2001b; Lee et al. 2003). The results of land-use classification are shown in Table 1 and Fig. 2. The ‘less-developed‘ areas occupy about 37% of the total area. This includes grasslands, forests (including Green Belts), legally inaccessible areas, rivers, streams, and wetlands. Most of the mountain forests are in the northern and southern parts of Seoul (Fig. 2). The ‘residential’ areas (about 37%) encompass detached houses, apartment houses, and traditional homes, and they also include the commercial and business areas, and mixed residential and business areas. The ‘agricultural’ areas (about 5%) are preferentially located near the western, eastern, southeastern city boundaries, where suburban agricultural activities for cropping of rice and flowers is still performed on open fields and in green houses. The ‘traffic’ areas occupy about 10% of the total area and include railroads, roads (dominantly, paved), airport and related facilities. More than 2 million vehicles are recently running on the roads, with an annual oil consumption of more than 80 million barrels/day (SMC 2001). Locally there is a very high traffic density (more than 12,000 vehicles/day) in the central downtown area and in main trunk roads, forming a potential source of groundwater pollution in Seoul (Choi et al. 2005). The ‘industrial’ areas occupy about 10% of the total area and are ubiquitous in Seoul (Table 1; Fig. 2). In such areas complex and varied factories and facilities have been located since 1970s, including small-scale light industries (e.g., textile industry, manufacture of plastic material, printing), machinery and non-ferrous metal processing industries, high-tech electronics, and urban infrastructure facilities (e.g., sewage treatment facilities, waste landfill areas, power plants, waste incineration facilities). Particularly in the southwestern part of Seoul, more than 1,000 factories are currently being operated within each administrative district. The industrial activity in Seoul has been considered as an important cause of groundwater contamination (Lee PK et al. 2001; Choi et al. 2005).

Table 1 Land use characteristics of Seoul Metropolitan City, Korea (after SMC 2001b)
Full size table
Fig. 2
figure 2

Land-use map of Seoul, Korea (modified after SMC 2001b)

Full size image

Materials and methods

Sampling and chemical analysis

Groundwater samples for the present study were collected during 2000 to 2001 from 57 pre-existing wells currently in use. The localities of sampling sites were chosen on the basis of land-use characteristics on both regional and local scales (Fig. 3). The average depth of chosen wells was 94 m below the land surface, thus indicating an environment of deep-fractured aquifers. All the sampling and analytical procedures followed the standard method described by APHA et al. (1992). Water samples were collected using suction pumps after purging at least three to four well volumes. Table 2 summarizes the instrumental methods used for the determining trace metals and VOCs and their detection limits.

Fig. 3
figure 3

Localities of urban groundwater sampling in Seoul, Korea

Full size image
Table 2 Analytical methods and their detection limits used in this study
Full size table

Samples for trace metal analysis were immediately filtered through 0.45 μm cellulose membranes and were acidified to pH<2 by adding several drops of ultra-pure nitric acid. They were kept at 4°C before the chemical analysis using ICP-MS (Perkin-Elmer SCIEX ELAN 6000). The standard solutions used for trace metal analysis were the NIST SRM 1643d standard (Trace Elements in Water) and Perkin Elmer Pure Atomic Spectroscopy standards. The percentage relative standard deviation (RSD) values determined from repeated analyses of standards and duplicate or triplicate samples were less than 5%.

Groundwater samples for the determination of VOCs were separately collected in 40 ml glass bottles with a tight Teflon cap. Special care was taken in order to minimize the evaporation of VOCs during sampling and storage. The samples were kept at 4°C under acidic (pH<2) conditions by adding HCl. Sixty-nine compounds belonging to VOCs were analyzed at the Korea Institute of Science and Technology (KIST) using a Gas Chromatographer (model HP5890) directly interfaced with a HP5970 mass selective detector (MSD) (Lee KJ et al. 2001). Before the GC-MSD analysis, extraction and pre-concentration of water samples were performed with a Tekmar LSC 3000 sample concentrator and an ALS 2016 purge-and-trap. The purge-and-trap was connected with GC by inserting a capillary (about 20 cm long) between the GC injection port and the purge-and-trap module. The conditions used for the purge-and-trap were: purge flow=40 ml/min (35°C, 99.9999% He), dry purge flow=20 ml/min (99.9999% He), sample volume=5 ml, desorption temperature=225°C (trapping T=225°C), desorption time=1 min, cold trap temperature=−150°C. All chromatograms were obtained in the selective ion monitoring (SIM) mode. The conditions of the GC/MSD were as follows: ultra-2 column (cross-linked 5% phenylmethylsilicon, 50 m×0.2 mm I.D.×0.33 μm film thickness); the He carrier gas set at 0.48 ml/min; 1/100 split ratio; injection port temperature=200°C; transfer line temperature=250°C. The standard solution used for VOCs determination was the Supel Co Stock Standard that was diluted with pure methanol solution. The analyses of field and laboratory duplicates showed the RSD values of <10%.

Statistical analysis

Non-parametric statistical analysis (Helsel and Hirsch 1992; Hudak and Blanchard 1997; Trojan et al. 2003) was performed to verify if significant statistical differences were present among land-use categories in terms of the concentrations of trace metals and VOCs. The Kruskal–Wallis test in the software package SPSS 10.0 was used because only five land-use groups with relatively small numbers of samples (7–18 for each group) were compared and the distribution of the measured data showed a non-normal distribution. For the test, analytical data (absolute concentrations) was ranked from 1 (smallest) to N (largest). These ranks, R ij , were then used for computation. Within each group, the average group rank, \(\ifmmode\expandafter\bar\else\expandafter\=\fi{R}_{j},\) was computed by the following equation:

$$\ifmmode\expandafter\bar\else\expandafter\=\fi{R}_{j} = \frac{{{\left[ {{\sum\limits_{i = 1}^{n_{j}} {R_{{ij}}}}} \right]}}}{{n_{j}}}.$$
(1)

The \(\ifmmode\expandafter\bar\else\expandafter\=\fi{R}_{j}\) was then compared to the overall average rank, \(\ifmmode\expandafter\bar\else\expandafter\=\fi{R}= (N+1)/2,\) squaring and weighting by sample size, to form the test statistic K:

$$K = \frac{{12}}{{N(N + 1)}} \times {\sum\limits_{j = 1}^k {n_{j} } }{\left[ {\ifmmode\expandafter\bar\else\expandafter\=\fi{R}_{j} - \frac{{N + 1}}{2}} \right]}^{2}. $$
(2)

The null hypothesis that the concentration of individual chemicals is not significantly different according to land use was tested using the Kruskal–Wallis (K–W) test; if the K value was larger than χ2 1-α,(k-1), the 1-α quantile of a chi-square distribution with (k-1) degrees of freedom, then the null hypothesis was rejected.

Results and discussion

Trace metals

Trace metals naturally occur in groundwater and their background concentrations are usually around at least two orders of magnitude. Under certain land-use conditions, however, trace metals may occur in groundwater at higher concentrations. Table 3 summarizes the concentrations of trace metals (Cd, Cu, Pb, Zn, Cr, Fe, Mn, As) in relation to land use in Seoul. Comparison of the data with the US EPA’s maximum contaminant levels (MCLs) and Korea Drinking Water Standards (KDWSs) showed that most of the considered metals did not exceed the regulation levels. However, some localities showed exceeding levels of Fe (two sites in the traffic areas and three sites in the industrial areas), Mn (two sites in the residential areas, one site in the traffic areas, and three sites in the industrial area), and Cu (one site in the traffic areas). The concentrations of most trace metals were remarkably low in the less-developed areas (Fig. 4). On the other hand, agricultural areas are characterized by high concentrations of Zn (average 229 μg/l); traffic areas by Cu (average 186 μg/l); and industrial areas by significant enrichments of most metals such as Fe (average 2,276 μg/l), Mn (average 248 μg/l), As (average 4.21 μg/l), Cr (average 4.15 μg/l), Pb (average 3.20 μg/l), and Cd (average 0.09 μg/l) (Fig. 4).

Table 3 Basic statistics of the concentrations of trace metals in urban groundwater (N=57), Seoul, Korea
Full size table
Fig. 4
figure 4

Box plots of trace metal concentrations in urban groundwater, Seoul, Korea

Full size image

The Kolmogorov–Smirnov (K–S) test was performed on the distribution of trace metals to determine whether parametric or non-parametric tests might be successfully employed (Table 4). Because all the P values were smaller than the significance level of 0.05 (Table 4), the distribution of all the trace metals was interpreted as non-normal. Therefore, a non-parametric test was used to investigate the effect of land use on the distribution of trace metals. The K–W test was applied to the proposed null hypothesis (i.e., no significant change of the concentrations of individual chemicals according to land use). The results showed that P values of six metals (Fe, Mn, As, Cr, Pb, Cd) are smaller than the significance level of 0.05 (Table 5), which indicates that at the 95% confidence level, the distribution of those six metals in Seoul groundwater is significantly changed with the land use. Therefore, it is suggested that anthropogenic contamination with respect to trace metals such as Fe, Mn, As, Cr, Pb and Cd has been proceeding in the industrial, traffic, and residential areas of Seoul.

Table 4 Results of the Kolmogorov–Smirnov (K–S) test for normality of the distribution of trace metals in urban groundwater (N=57), Seoul, Korea
Full size table
Table 5 Results of the Kruskal–Wallis test for the difference in distribution of groundwater trace metals according to land use, Seoul, Korea
Full size table

Volatile organic compounds

Many kinds of VOCs may pose a potential risk to the environment because of their toxicity (e.g., pesticides and herbicides) and carcinogenicity (e.g., chloroform and TCE), and their potential adverse effects of reactions occurring within the natural system, such as oxygen consumption. As summarized in Appendix, a total of 69 compounds of VOCs were analyzed in this study. These are 47 species of chlorinated compounds (28 halogenated alkanes, 10 halogenated alkenes, 9 halogenated aromatics), 18 species of gasoline-related compounds (3 aromatic hydrocarbons, 14 alkyl benzenes, 1 ether) and 4 other species. Among 69 chemicals analyzed, 19 species were detected in Seoul groundwater. The basic statistics on the concentrations of detected 19 species are summarized in Table 6 according to the land use types in Seoul. More detailed data, together with detailed trace metal data, are available upon request to the corresponding author (S.-T. Yun).

Table 6 Basic statistics of the concentrations (unit: μg/l) of VOCs in urban groundwater (N=57), Seoul, Korea
Full size table

As shown in Table 6 and Fig. 5, VOCs belonging to the chlorinated hydrocarbons were most frequently detected. A total of 16 chlorinated compounds were detected in a few localities, which include 12 species of halogenated alkanes [methylene chloride, chloroform, carbon tetrachloride, bromodichloromethane, dibromochloromethane, bromoform, 1,1-dichloroethane (1,1-DCA), 1,1,1-trichloroethane (1,1,1-TCA), 1,2-dichloroethane (1,2-DCA), 1,2-dichloropropane, 1,1,2-trichloroethane (1,1,2-TCA), 1,1,2-trichloro-1,2,2-trifluoroethane (CFC113)] and 4 species of halogenated alkenes [cis-1,2-dichloroethene (c-DCE), trichloroethene (TCE), tetrachloroethene (PCE), 1,1-dichloroethene (1,1-DCE)]. Among 18 analyzed species of gasoline-related compounds, only 2 species of alkyl benzenes (toluene, tert-butylbenzene) and 1 species of ethers [methyl tertiary butyl ether (MTBE)] were detected. On the other hand, halogenated aromatics and aromatic hydrocarbons were not detected at all. The kinds and concentrations of the VOCs detected in this study are similar with those reported for shallow urban groundwaters across the USA (Lopes and Bender 1998).

Fig. 5
figure 5

Box plots of VOCs concentrations in urban groundwater, Seoul, Korea

Full size image

The detection frequency, as represented by the percentage of the samples with detection among 57 studied samples, was highest for toluene (86%), followed by chloroform (63%), methylene chloride (58%), TCE (40%), bromodichloromethane (32%), PCE and 1,1,1-TCA (each 25%), 1,1-DCA (18%), c-DCE (16%), MTBE (13%), tert-butylbenzene (12%), carbon tetrachloride (9%), and others (each <7%) (Fig. 5). These species were also documented as frequently encountered VOCs in shallow urban groundwaters (Lopes and Bender 1998). A comparison of data with the US EPA and Korean standards for drinking water (Table 6) showed that some compounds from a few localities exceeded the levels, such as TCE in six sampling sites (one site in the residential areas, one in the traffic areas, and four in the industrial areas), PCE in eight sites (one site in the residential areas, four in the traffic areas, and three in the industrial areas), 1,2-DCA in one site of the industrial areas, and 1,2-dichloropropane in one site of the industrial areas.

In relation to land use, more than one compound of VOCs was detected in all samples from the traffic, industrial, and residential areas. Table 7 shows the total numbers and kinds of VOCs detected in relation to land use. The detection frequency was obviously minimal in the less-developed areas (i.e., totaling 19 species in 10 samples, corresponding to a detection probability of circa 3% [=19 species/(10 samples×69 species/sample)×100%], while groundwaters in the industrial areas showed the most frequent detection (circa 9%). The summation of the concentrations of detected VOCs also showed that the industrial areas have the highest concentrations (average 25 μg/l) with a wide range [not detected (<0.1) to 1,429.9 μg/l]. In the less-developed areas only three compounds (i.e., methylene chloride, toluene, chloroform) were detected with very low concentrations; the summation of the 3 species was not detected (<0.1) to 2.3 μg/l (average 0.6 μg/l) (Table 6). This suggests that urban groundwater contamination by VOCs appears to be significantly low in the less-developed areas, compared to areas of other land uses.

Table 7 Total numbers of the detection of VOCs in groundwaters (N=57) from each land-use category, Seoul, Korea
Full size table

The present study shows that anthropogenic contamination especially by TCE and PCE is significant in urban groundwater. The very high concentration of TCE was observed in the industrial areas (average 309.1 μg/l), followed by the traffic (average 13.3 μg/l), residential (average 9.5 μg/l), agricultural (average 4.5 μg/l), and less-developed (not detected, <0.2 μg/l) areas. Likewise, the average concentration of PCE was significantly higher in the industrial areas (63.4 μg/l) than the residential (6.8 μg/l), traffic (6.0 μg/l), and less-developed and agricultural areas (not detected, <0.5 μg/l) (Table 6; Fig. 5). The measured concentrations of TCE and PCE frequently exceeded the US EPAs and Korean Drinking Water Standards in many of the industrialized and traffic areas. This indicates that TCE and PCE may cause serious impacts on ecosystems and human health and should therefore be carefully monitored in Seoul.

A non-parametric statistical test was conducted to confirm the dependence of VOCs contamination upon land use. The K–S test showed that all the P values for detected VOCs were smaller than the significance level of 0.05 (Table 8), suggesting that the distribution of detected species can be considered to be non-normal. Sequentially, the K–W test was performed as a non-parametric statistical analysis to compare the distribution of VOCs among different land-use types. The results showed that for seven compounds (toluene, TCE, PCE, chloroform, carbon tetrachloride, bromodichloromethane, CFC113) among 19 detected species, the P values are smaller than the significance level (0.05) (Table 9). This may indicate that the occurrence of these compounds closely depends upon the land use. Accordingly, those seven compounds may reflect the progress of anthropogenic contamination and can be effectively used as indicators to identify the degree of anthropogenic contamination in relation to land use. In comparison, six trace metals (Fe, Mn, As, Cr, Pb, Cd) with the P values of <0.05 (Table 5) can also be recommended as indicators of anthropogenic contamination in Seoul groundwater. Furthermore, three compounds (TCE, PCE, chloroform) among the seven species are likely the best parameters in evaluating the effect of land use on groundwater quality in Seoul because they occur frequently and show relatively high concentrations.

Table 8 Results of the Kolmogorov–Smirnov test for the normality of the distribution of VOCs in groundwater (N=57), Seoul, Korea
Full size table
Table 9 Results of the Kruskal–Wallis test for the difference in distribution of groundwater VOCs according to land use, Seoul, Korea
Full size table

Summary and conclusions

The present study was performed to investigate the contamination of Seoul groundwater by trace metals and VOCs, focusing on the relationship between anthropogenic contamination and land use. The land use was classified into five categories: less-developed, residential, agricultural, traffic, and industrial. For research purposes, a total of 57 groundwater samples were collected from preexisting wells in use (average 94 m deep) and they were analyzed for eight trace metals (Fe, Mn, As, Cr, Pb, Zn, Cd, Cu) by ICP-MS and 69 VOCs by GC-MSD. Among the 69 compounds of VOCs, only 19 species were detected. The water quality data were statistically analyzed using the K–W test (a non-parametric statistical tool) to understand land-use control on groundwater contamination.

This study verified that the concentrations and spatial distribution of some trace metals (especially, Fe, Mn, As, Cr, Pb, Cd) and VOCs (especially, TCE, PCE and chloroform; and potentially toluene, carbon tetrachloride, bromodichloromethane, and CFC113) are significantly influenced by the land use in Seoul. The groundwaters in the less-developed areas are the least contaminated, as reflected by very low concentrations (if detected) of trace metals and VOCs. On the contrary, groundwaters in the industrial areas are significantly contaminated by some trace metals and VOCS. Though the current concentration levels for the examined trace metals and VOCs, for the most part, do not exceed the drinking water standards suggested by US EPA and Korea Ministry of Environment, PCE (eight sites), TCE (six sites), Mn (six sites), Fe (five sites), Cu (one site), 1,2-DCA (one site), and 1,2-dichloropropane (one site) showed the local contamination to an undrinkable level. In particular, groundwaters in the industrial areas have significantly higher concentrations of most of the trace metals and detected VOCs. The contamination by TCE and PCE is noticeable and locally serious in the industrial and traffic areas and therefore should be carefully monitored with special attention.

The results of this study show that land use is an important factor in urban groundwater quality. In particular, the concentrations of some trace metals (Fe, Mn, As, Cr, Pb, Cd) and VOCs (TCE, PCE, chloroform) are significantly controlled by land use. Those chemicals can therefore be used as reliable indicators on the progress of anthropogenic contamination in relation to land use. This study also suggests that careful and regular monitoring of groundwater for trace metals and VOCs is required in Seoul to safeguard the public water supply. In addition, monitoring works should be carefully designed to reflect the change in land-use patterns.