Changes in NO2 Concentration from Major Cities and Provinces in Korea: A Case Study from 1998 to 2003

1 Department of Earth & Environmental Sciences, Sejong University, Seoul, Korea * Corresponding author address: Prof. Ki-Hyun Kim, Department of Earth & Environmental Sciences, Sejong University, Seoul, Korea; E-mail: khkim@sejong.ac.kr In this paper, the spatio-temporal distribution characteristics of nitrogen dioxide ( NO2) were investigated using data obtained routinely from air quality monitoring stations located in seven major cities and nine provinces in Korea for the period 1998 to 2003. The results indicate that annual trends in NO2 concentrations generally reflect changes in environmental conditions, while exhibiting cyclic and systematic patterns across seasons. Its wintertime concentrations were most prominent, with concentrations decreasing gradually across spring, fall, and summer. If concentration patterns are examined among different cities and/or provinces, the highest mean values were found from the Capital city, Seoul (34.7 ppb) and the surrounding province, Gyunggi (30.2 ppb). In contrast, remarkably reduced NO2 concentrations were seen in such regions as the remote island, Jeju and Jeonnam province with mean values of 17.5 and 16.5 ppb, respectively. The overall results of our study indicate that there are strong geographical gradients in NO2 distributions to exhibit strongly polluted patterns consistently in densely populated urban areas (e.g., major city), compared to large rural areas (e.g., province). (


INTRODUCTION
A broad spectrum of pollutants, which include sulfur and nitrogen oxides, are released into the atmospheric environment as a result of both natural and man-made activities.Pollutant emissions adversely impact air quality and consequently human health.Policymakers and scientists thus pay a great deal of attention to the various health effects of air pollution, such as those which occur due to chronic exposure to ambient air pollution; for example, epidemiological studies have pointed out specific concerns about traffic pollution.Here accounts of notable health effects such as cardiopulmonary mortality among those residing near major road areas have been reported (e.g., Hoek et al. 2002).
In environments affected by high traffic loadings, pollutant emission patterns need to be described precisely as direct emissions of airborne pollutants such as nitrogen oxides can have a large effect on the total pollution budget (Stedman et al. 2001).Concentrations of NO 2 and other vehicular pollutants in the atmospheric environment have increased substantially with vehicular populations.However, vehicular populations alone do not govern the overall extent of such pollution; for example, NO 2 concentrations have been shown to be affected by a number of parameters including vehicle type, fuel used, and vehicle operating conditions (Carslaw et al. 2005).Maximum NO 2 emissions in Korea are seen mainly in big cities and over industrial areas, where NO 2 emissions have been found to exceed 10,000 equivalents (eq) ha yr NO 2 is released into the atmosphere directly as a product of combustion processes (Silibello et al. 1998) or as a result of conversion through oxidation of NO by ozone (O 3 ) or oxygen ( O 2 ) (Mackenzie et al. 1995;Soltic and Weilenmann 2003).Problems caused by air pollution due to such photochemical oxidants have been recorded in many of previous studies (e.g., Lee et al. 1996).The environmental fate of ozone (O 3 ) and nitrogen dioxide ( NO 2 ) are important, due to their potential to adversely impact surrounding ecological systems.Because of their chemical coupling, the concentration levels of O 3 and NO 2 are often inextricably linked to each other (Mazzeo et al. 2005).Hence, a better knowledge of these chemicals is a prerequisite to providing more efficient methods for their control.In this study, we investigate NO 2 measurement data collected from seven major cities and nine provinces in Korea during the period 1998 to 2003.In order to provide a better description of NO 2 behavior, those NO 2 data were examined on the basis of diverse temporal and spatial criteria.To explain factors affecting the distribution of NO 2 across different cities or provinces, correlation analysis was also conducted.

MATERIALS AND METHODS
Concentrations of NO 2 were determined from air quality monitoring stations dispersed across a total of 16 districts (seven major cities and nine provinces) in Korea from 1998 to 2003 (Fig. 1).To estimate NO 2 pollution levels in each district, concentration levels of criteria pollutants reported by the Korean Ministry of the Environment (KMOE) were used (Table 1).The original concentration data were quality-controlled and stored in a data management network system operated by the KMOE (refer to regulation criteria in Table 1).The initial datasets recorded routinely at hourly intervals from each monitoring station were stored in the KMOE data management system after being converted into monthly intervals.The QA/QC for these data have been reported in the KMOE annual reports.
The number of individual air quality monitoring (AQM) stations in Korea increased gradually each year throughout the study period, i.e., from 127 (1998) to 190 (2003), as shown Fig. 1.A map of S. Korea is drawn to show all districts for air quality monitoring in Korea.Short name of each district are given in the parenthesis.For all provinces, ''' symbols are also added next to the two letters to make easy comparison with the short names of major cities.
in Table 2.For the purpose of classification, all of the NO 2 data collected from those individual stations were grouped into the 16 major administrative districts.The geographical locations of these major cities and provinces in Korea are described in Fig. 1.After being grouped, the NO 2 datasets for all individual stations were combined together to derive values for each respective district.Due to this combining process, each of these 16 districts has a total of 72 monthly NO 2 concentration values to represent its distribution patterns for the entire study period.In the course of this study, NO 2 values converted by the above procedure were exam-Table 1.The types of criteria pollutants regulated by the Korea Ministry of the Environment (KMOE).(Exceedance criteria set in 1995 are given along with each respective measurement method).
ined further at various temporal scales (e.g., seasonal and inter-annual trends) for each individual district.In addition, using monthly mean datasets for all different districts, correlation patterns were also analyzed and evaluated.

Spatial Distribution Patterns of NO 2 Data
The mean, min., and max.values for each specific month throughout the study period were computed for each individual district after pooling and combining monthly datasets from the sum of 127 to 190 individual stations; in the case of Seoul, the number of individual stations varied from 27 (1998 -2002) to 31 (2003) (Table 2).In order to examine spatial factors on NO 2 distribution patterns, basic statistical parameters derived from the above procedure were further evaluated for each district (Table 3).For the description of NO 2 distribution patterns for each district, we use the short symbols presented in Table 2 (e.g., Seoul to SL) throughout this text.As seen in Table 3, the highest mean of NO 2 was found in SL (mean ± SD : 34.7 ± 6.38 ppb, N = 72) followed by GG (30.2 ± 5.88 ppb, N = 72); GG province, while surrounding the capital city of SL, has a number of large industrial areas (including the Ansan industrial area).In contrast, the lowest mean NO 2 level was seen in such provinces as JN (16.5 ± 3.43 ppb, N = 72) and JJ province (17.5 ± 6.79 ppb, N = 72).If the results of the two data groups (i.e., cities and provinces) are compared, moderate differences in NO 2 concentration levels are indicated.The mean values for major cities increased in the following order: US, GJ, DJ, BS, IC, DG, and SL with values ranging from 19.6 to 34.7 ppb.The results for the provinces showed a minimum value in JN and then increased in the following order: JJ, GW, CN, JB, GN, CB, GB, and GG with values ranging from 16.5 to 30.2 ppb.
To make a meaningful comparison of NO 2 spatial distribution patterns, the NO 2 data for each district were compared in terms of different statistical terms.To explain how those parameters differ between the districts, basic statistical parameters (mean, min., and max.) obtained for each district each month [throughout the study period (N = 72)] were re-summed for the derivation of their 2 nd -stage average values, as shown in Fig. 2. The results of the majorcities group indicate that the highest maximum value occurred at SL with 51.0 ppb, and the Table 2.The number of total individual stations for the comparison of NO 2 distribution patterns on a district basis (seven major cites and nine provinces).
lowest minimum at IC with 9.4 ppb.In the case of the provinces, the highest maximum value was found from GG with 48.7 ppb, and the lowest minimum from JN at 8.9 ppb.
To explain NO 2 occurrence patterns for each individual district, frequency distribution patterns for NO 2 occurrence were examined by sorting and classifying monthly datasets for each district.When the results of seven major cities are examined, maximum frequency normally occurs near 20 ~ 30 ppb.However, in the case of SL, it was much higher with a value of 40 ppb.In the case of the nine provinces, the maximum frequency was generally seen near 20 ~ 25 ppb.This frequency analysis clearly indicates that NO 2 concentrations in major cities occur at values that are notably higher than those of the provinces.To further describe NO 2 distributions for different districts, monthly mean NO 2 concentrations were plotted as a function of time (Fig. 3).The results show that the largest mean value was found at SL (48.1 ppb), while the lowest mean was at JJ (5 ppb).The significantly enhanced NO 2 levels in the capital SL and the surrounding GG province may also be related to the distinctively high population and traffic density in that region of the peninsula (Tables 4, 5).This is easily understood, Table 3.A statistical summary of NO 2 measurement data from seven major cities and nine provinces in Korea from 1998 -2003.
considering the high number of vehicles in these densely populated regions leading to significantly enhanced pollutant emissions (such as NO 2 ).In general, vehicular pollution from both diesel and petrol motors remains a major problem, especially in the regions surrounding provincial capitals where the numbers of automobiles are highest (Goyal and Sidthartha 2003).
Consideration for emission factors and vehicle activity has also demonstrated that the principal reasons for higher NO 2 / NO X ratios in city center areas are the high proportion of diesel vehicles, high flows of buses and taxis, and low vehicle speeds caused by traffic congestion (Carslaw et al. 2005).It is thus reasonable to conclude that the differences in observed NO 2 spatial distribution patterns (e.g., between cities and provinces) are due primarily to anthropogenic activity levels.

Seasonal Patterns
To examine the effects of temporal factors on NO 2 distribution patterns, the data from all 16 districts in Korea (seven cities and nine provinces) were compared for each season.As seen in Fig. 4, the seasonal mean values of most districts generally show a consistent and systematic pattern.Generally, they exhibit concentration changes in the following descending order: winter, spring, fall, and summer.However, on certain occasions (SL, IC, and GJ), the springtime concentrations tend to be the highest of all seasons.The most common pattern for most districts thus can be characterized as a notable increase during winter (and spring) and a significant drop in summer (and fall).Although there tends to be a moderate difference in NO 2 distributions between spring and fall, those seen between winter and summer are quite dramatic.For example, the wintertime mean value at SL was 38.2 ppb, while its summertime counterpart showed a 24% reduction at 28.9 ppb.There are many previous studies showing similar temporal patterns with high NO 2 concentrations being found during winter and lower values during summer (Hargreaves et al. 2005, Gupta et al. 2003).Observed seasonality of NO 2 distribution is attributable to a number of factors including higher fuel usage for heating (Gupta et al. 2003).If the results among seven major cities are compared, the highest seasonal mean value was recorded in SL during spring at 39.3 ppb (winter value: 38.2 ppb).On the other hand, the lowest value was seen in DJ during the summer (13.4 ppb).If the results are compared among the nine provinces, the highest and lowest seasonal mean concentrations were found in GG (winter 34.3 ppb) and JJ (summer 10.4 ppb), respectively.The winter (or spring) highs could also be linked to such factors as increased central heating.Moreover, the high winter concentrations of NO 2 could be enhanced further by a reduction in photochemical reactions in which NO 2 and hydroxyl (OH) radicals combine to form nitric acid ( HNO 3 ) (Derwent et al. 1995).In terms of summertime variability, strong rains can contribute to lower NO 2 concentrations.Such seasonal trends have been demonstrated previously, in particular the efficient removal of NO 3 − from the atmosphere was principally attributed to summer rains by Hong et al. (2002).In addition, the mean mixing height layer, which is reduced to its minimal level in winter (or spring), can also be a very important factor, as it controls the transport pattern of NO 2 through the seasons (Shahgedanova et al. 1999).

Annual Patterns
As NO 2 measurements were made continuously over several years, these data can be used to evaluate trends over a relatively long-term period.As seen in Fig. 5, the results of longterm analysis indicate that there are some variations in NO 2 distribution patterns throughout the five-year period for all 16 districts.The highest mean values were typically found at big cities and densely populated urban areas (e.g., SL and GG).On the other hand, the lowest values were seen most frequently from those with wide rural areas or where vehicular activities and industrialization are low (such as JJ and JN).When long-term NO 2 distribution patterns among the seven major cities are compared, the most prominent pattern exists at SL where a gradual increase from 30.1 (1998) to 38.5 ppb ( 2003) is observed.On the other hand, the NO 2 distributions from the six remaining cities tend to fluctuate, with mean values ranging from 16.1 (GJ) to 29.9 ppb (IC).Although we were not able to obtain entire annual datasets for detailed analysis of vehicular registration patterns (Table 5), those available for the year 2003 indicate that patterns for SL and GG area are not easy to predict.While there exist some reports regarding efforts to control NO 2 emissions (e.g., Dixon et al. 2001), patterns in SL suggest that many control policies developed in Korea may not be effective in reducing NO 2 concentration levels, especially in large urban areas.
The results of this comparison indicate the possibility that fast growth in vehicular population, prominent in large urban areas such as SL, may have led to a corresponding increase in vehicular emissions of NO 2 .There has been a marked increase in the number of vehicle registrations in SL, and this corresponds to increased population (Fig. 6 and Table 5).Such factors linked together can promote increased consumption of fossil fuels (Fig. 7), leading to enhanced NO 2 emissions.Except for the case of GG, with an increasing trend in annual NO 2 levels [27.2 (1998) to 32.7 ppb ( 2003)], the remaining provinces generally exhibit moderate changes with mean values of 10.5 (JJ) to 24.3 ppb (GN) during the entire study period.
Another readily observable trend is that concentrations in major cities are systematically higher than those of the provinces.This observation supports previous finding that gradual increases in NO 2 emissions have coincided with increased urbanization and industrialization in areas (Jo et al. 2000).

Factors affecting NO 2 distribution
To study factors affecting distribution characteristics of NO 2 , we investigated spatial relationships among different districts using monthly NO 2 datasets.The relationships among all districts were checked after dividing them into three data groups A (city -to -city), B (province -to -province), and C (city -to -province) (Table 6).The strengths of correlations among the matching district pairs were estimated arbitrarily by dividing all results into five different classes based on the magnitude of probability (P) such as: ( 1 7).Results for data group A indicate that for a total of 21 correlation cases, SL generally exhibits the highest correlations with the remaining six cities, especially with IC (class IV).It is in fact found that the closer the distance between the sites, the stronger the correlation is.It should also be noted that the strongly correlated pairs are generally found between sites of high NO 2 concentrations.In the case of group B, the strongly correlated pairs tend to be seen between GN and many other provinces.In this group, the weakest correlations are generally found where the NO 2 concentrations are low, such as JJ and JB.In the case of group C, with a total of 63 matching cases, it was seen that 3 cases of the strongest correlation (class IV) exist, while 10 cases do not belong to any class (no class).The weakly correlated pairs in this C group are found most abundantly from the pairs between JJ and the seven cities.The results of this correlation analysis thus indicate that there are many cases of strongly correlated pairs from all data groups, while the patterns tend to be distinguished by such factors as the level of man-made activities for a given city (or province).

CONCLUSIONS
In this study, spatio-temporal distribution characteristics of NO 2 were investigated using the datasets measured routinely from air quality monitoring stations located throughout the 16 districts of Korea during 1998 to 2003.The spatial distribution patterns for the study period were generally characterized by the highest NO 2 concentration being at the capital, SL (city), and the lowest concentration being at the remote islands area of JN (province).If the distribution patterns are compared among different districts, the occurrence pattern of NO 2 datasets in the cities peaked near 20 ~ 30 ppb (with the prominent case of SL at 40 ppb), while those of the provinces were generally at 20 ~ 25 ppb range.Moreover, examination of seasonal patterns indicates that the highest mean values were typically found during wintertime followed by spring, fall, and summer.The cause of these seasonal patterns can be attributed to such factors as the combined effect of fossil fuel consumption patterns and meteorological conditions.Examinations of the annual patterns also indicate that high NO 2 concentrations typically occur at big cities and their vicinities, where rapid urbanization and high industrialization exist.
The results of correlation analysis among all 16 districts show that stronger correlations predominate in areas of higher NO 2 concentration.In addition, the distance between districts can also act as an important factor in determining the strength of correlation.The overall results of this analysis thus clearly indicate that enhanced man-made activities in more urbanized re-c.Group C (city -to -province) gions contribute significantly to changes in NO 2 levels on the Korean peninsula.Emission standards as they relate to pollution sources need to be elaborated on in order to further control pollutants such as NO 2 .Tactics such as placing emission standards on newly manufactured vehicles, in-use vehicles, and fuel production standards may be the most effective methods for efficiently reducing vehicle exhaust air pollution.

Fig. 2 .
Fig. 2. Comparison of NO 2 concentrations among seven major cities and nine provinces in Korea for the period covering 1998 to 2003.The secondstage mean values of three basic statistical terms (mean, min., and max.) were calculated using the monthly datasets for each of all 16 districts investigated in this study.

Fig. 3 .
Fig. 3. Comparison of the monthly mean concentrations of NO 2 (ppb) from major cities and provinces in Korea.

Fig. 4 .
Fig. 4. Comparison of seasonal mean concentrations of NO 2 (ppb) measured from both major cities and provinces.

Fig. 5 .Fig. 7 .
Fig. 5. Comparison of the yearly mean concentrations of NO 2 (ppb) from the major cities and provinces.

Table 6 .
Results of correlation analysis on NO 2 concentration datasets: their relationships are compared after being grouped into three different categories: (A) city -to -city, (B) province -to -province, (C) city -to -province.a. Group A (city -to -city) b.Group B (province -to -province)

Table 5
. Total motor vehicle registration of districts in 2003 in Korea.

Table 7 .
Evaluation of correlation analysis results for all NO 2 concentration data sets.All data were collected from seven cities and nine provinces in Korea during the study period.Refer to Table6for the original results of the correlation analysis.