Seasonal and Diurnal Trend of Carbon Dioxide in a Mountainous Site in Seoul , Korea

ABSTRACT In this research, the environmental behavior of car-bon dioxide(CO 2 ) was investigated in a mountainoussite in the proximity of a highly industrialized mega-city, Seoul, Korea. The concentration data of CO 2 monitored routinely at hourly intervals at Mt. Gwan-Ak(GA), Seoul, Korea throughout 2009 were analyz-ed in several respects. The mean CO 2 value was 405±12.1ppm(median=403ppm) with a range of 344to 508ppm(N=8548). The analysis of its seasonaltrend indicated that the CO 2 levels peaked in the win-ter but reached a minimum in fall. If the short-termtrend is analyzed, the CO 2 values generally peakedduring daytime along with the presence of twoshoulders; this is suspected to be indicative of strongman-made effects(e.g., traffic activities). It is seenthat the general patterns of CO 2 distribution in thisstudy area are highly comparable to thosetypicallyfound in urban areas with strong signals of anthro-pogenic activities. Key words: Carbon dioxide, Mountain, Continuousmonitoring, Urban area, Anthropogenic


INTRODUCTION
Carbon dioxide (CO 2 ) is considered a trace gas constituting about 0.038% of the earth's atmosphere (Williams, 2009).Currently about 57% of man-made CO 2 emissions are known to be removed by the biosphere and oceans (Canadell et al., 2007).However, because of various human activities (e.g., deforestation, combustion of fossil fuels, power generation, etc), the concentration of atmospheric CO 2 has been increasing gradually through the years (Colombo et al., 2000;Denning et al., 1995).Keeling (1960) initiated continuous monitoring of atmospheric CO 2 at Mt. Mauna Loa, Hawaii, US since 1958.From that time on, monitoring of CO 2 has been conducted routinely in many background areas (e.g., South Pole).At present, CO 2 concentration at Mauna Loa is 392 ppm (by volume) (NOAA, 2010).This level of change corresponds to an increase of about 40% since the beginning of the industrialization age.As greenhouse gases (like CO 2 ) are identified to play a big role in climate change (both regionally and globally), its control is a matter of concern in many countries and societies.
According to the Fourth Assessment Report by the Intergovernmental Panel on Climate Change (IPCC, 2007), global surface temperature increased by 0.74± 0.18� C during the 20 th century with much of the warming centering over the last 30 years.This warming will cause significant changes in ecosystem such as the reduction in the area of ice cover and the rise of sea level among other impacts (NASNAEIMNRC, 2008a, b).In the next 100 years, the temperature is likely to increase at least 1.1� C and possibly over 6� C, if the current trend of global warming continues (NASNAEIMNRC, 2008a, b).The cause of increases in temperature has been attributed to the rise in concentration levels of greenhouse gases.
As the concentration of CO 2 varies in diverse temporal scales (Keeling et al., 1995(Keeling et al., , 1984)), so does its spatial scales between urban and rural areas (Nemitz et al., 2002) and between indoor and outdoor environments (Kovesi et al., 2007).In South Korea, the rapid industrial growth, accompanied by socioeconomic change, has brought an immense rise in CO 2 levels to become the 9 th ranked country of the global fossilfuel consumption, i.e., nearly 130 million metric tons of carbon in 2006 (Boden et al., 2009).Accordingly, South Korea needs to reduce its CO 2 emission by 55% to reach the average value for the world per capita.The statistics from the Carbon Dioxide Information Analysis Center (CDIAC) also indicate that South Korea experienced a phenomenal growth in fossil-fuel CO 2 emissions with a mean growth rate of 11.5% from 1946 -1997 (Boden et al., 2009).As sustainable development has become a prior obligation in the 21 st century,

Seoul
was investigated using its hourly measurement data collected at the Mt.Gwan-Ak (GA) monitoring station (elevation: 632 m) in Seoul, Korea during the year 2009.Using these hourly measurement data, we first examined the temporal distribution of CO 2 in both diurnal and seasonal scale.These data sets were then analyzed in relation with the relevant environmental parameters to describe the fundamental factors controlling its distribution in a number of aspects.The results of this study will provide valuable insights into the factors governing the environmental behavior of CO 2 in response to various source/sink processes in the urban environment.

1 Site Description
As the capital of South Korea, Seoul has a carbon footprint of 1.59 metric tones per person (Sovacool and Brown, 2010).The Seoul metropolitan area has 10 million inhabitants, while occupying only 0.6 percent of the country's land area.Nonetheless, the city with 25 districts produces 21 percent of its GDP, mainly in business and financial sectors with technology firms and banking giants such as Samsung, Hyundai, Kia, and LG (Sovacool and Brown, 2010).More than 80% of the total energy used in Seoul comes from fossil fuels, mostly coal, petroleum, and natural gas (Sovacool and Brown, 2010).The city also maintains a large number of water and waste treatment plants along with landfills that can release fairly large quantities of greenhouse gases (Jo et al., 2008;Kaneko and Dhakal, 2008;Jo, 2002).
In this study, the distribution pattern of carbon dioxide was investigated using the data sets collected from an air quality monitoring station on the top of Mt.GA (37� 26′44′′N, and 126� 57′49′′E), a relatively small mountain located at the southern district of Seoul, South Korea with the total area of 19,226,942 m 2 (Fig. 1).Mt. GA is situated to cover 4 districts in terms of the administrative zones, (1) 11,412,035 m 2 (59.4%) in Gwanak-gu, (2) 2,120,595 m 2 (11%) in Geumcheongu, and (3) 5,694,312 m 2 (29.6%) in both Gwacheon City and Anyang City of Gyeonggi-do.Our target study area of Mt.GA belongs to a temperate climate zone with the mean temperature of 9.53±10.4�C (seasonal means of -3.2 (winter) to 20.3� C (summer)).Likewise, the UV radiation also exhibited the seasonal mean value (mWcm -2 ) of 0.07 (winter) to 0.45 (summer).Relative humidity was significantly lower in spring (58.6%) than other seasons.Examination of the wind rose pattern indicated that during most of the time, winds were blown from WWN fol-lowed by westerlies.It is steep topographically with a ravine developed in all directions.Only a few types of needle-leaf trees (e.g., pine trees) are grown wildly while a variety of falling broadleaved trees (e.g., black oaks) can be found frequently (Gwanak-Gu, 2010).

2 Data Collection
From the air quality monitoring site, the hourly CO 2 data were collected continuously to cover a whole year from January 1 st to December 31 st 2009 using a CO 2 / CH 4 /H 2 O analyzer (Picarro G1301, US).An automated air pollution monitoring system (Thermo, USA) is located at a height of 620 meters above mean sea level.The vertical height of sampling inlet from the monitoring station is 14 m, while the station itself is aloft 9 m above the soil.A list of the criteria pollutants (CH 4 , NO x , O 3 , SO 2 , and particulate matters) were also monitored concurrently along with the meteorological parameters (air temperature, UV, humidity, wind speed, etc).The analysis of CO 2 data and all the relevant parameters can help us properly evaluate the influence of various factors and processes affecting the behavior of CO 2 in the study area.

3 QA/QC Section
The instrument used to measure concentration of CO 2 is Picarro G1301 analyzer (Picarro, US).The analyzer is based on Picarro's unique Wavelength-Scanned Cavity Ring Down Spectroscopy (WS-CRDS), a time based measurement technique based on a near infrared laser.This CRDS technique is a highly precise method allowing to measure a spectral signature of the target molecule (Picarro, 2010).It is a real time, trace gas monitor capable of measuring gases at partsper-billion (ppbv) sensitivity.The instrument is capable of measuring CO 2 in the range 0-1,000 ppmv and CH 4 in the range 0-20 ppmv.By following the procedure of Busch and Busch (1997), the measurement precision was assessed by taking a spectral scan at every 5 min with the 380 ppm CO 2 standard at room temperature.The relative standard deviation was then estimated as 0.04%.

1 The Basic Aspects of CO 2 Distribution in
the Study Area In this study, CO 2 concentration data were collected from an air quality monitoring station located on the top of Mt.GA, in the southern district of Seoul, South Korea for a one year period (January to December 2009).To explore the overall trend of CO 2 , its hourly data were at first plotted as a function of time (Julian day) (Fig. 2(a)).The mean hourly concentration of CO 2 for the entire study period was 405±12.1 ppm (median= =403 ppm) with a range of 344 to 508 ppm (N= =8,548).The maximum hourly concentrations of CO 2 (508 ppm) was determined at 1 am on December 3 rd , while its minimum value (344 ppm) at 7 am on March 8 th .This observed annual concentration of CO 2 in Mt.GA was 4.55% higher than the average value of Mauna Loa (387.35 ppm) for the year 2009.The trend in CO 2 at Mauna Loa can be one of the most representative sites to predict the global trend.

2 Seasonal Distribution of CO 2
As shown in Table 2, the CO 2 data divided into each season show the maximum value in winter (410± 13.4 ppm) followed by spring (407±9.56ppm), summer (402±11.3ppm) and fall (399±11 ppm).To accurately describe the temporal trend of CO 2 over a one year period, some parameters derived on daily basis (average, minimum, and maximum) were also plotted in Fig. 2(b).It is found that differences in CO 2 levels between winter (the highest) and spring (the next) are statistically significant (P⁄0.05).The relative enhancement in wintertime CO 2 levels can be basically sought from the combined effect of such factors as the increasing consumption rate of fossil fuel (e.g., house heating), reduced photosynthesis, and more stable atmospheric conditions (Henninger and Kuttler, 2010).Our findings of relative enhancement in CO 2 during the wintertime comply well with many previous studies conducted in the background as well as in the urban region in the northern hemisphere (Henninger and Kuttler, 2010;Miyaoka et al., 2007;Pataki et al., 2003;Aikawa et al., 1995;Woodwell et al., 1978;Bolin and Keeling, 1963).It is however interesting to note that the lowest seasonal mean took place in fall.This pattern is quite unique if one considers the fact that the minimum CO 2 values were typically seen during summer in most of  the previous studies (Pataki et al., 2003;Woodwell et al., 1978;Bolin and Keeling, 1963).Summer minimum of CO 2 concentration was also observed in urban area of Nagoya and Sapporo in the neighboring country Japan as opposed to our finding of minimum concentration in fall (Miyaoka et al., 2007;Aikawa et al., 1995).Henninger and Kuttler (2010) also found the lowest concentration of CO 2 in summer in Essan, a typical urban conurbation city of Germany.In compliance with seasonal trend, comparison of monthly mean values of CO 2 shows it's minimum and maximum in October and February, respectively.In contrast, minimum and maximum values in most of the previous studies were found most commonly in August and April, respectively (e.g., Nakazawa et al., 1992).

3 Diurnal Variation in Carbon Dioxide
Levels To understand the short-term variability of CO 2 , the mean hourly CO 2 values were examined over a diurnal cycle for both seasonal group and a whole year period (Fig. 3).For all data groups, a clear cycle is observed  with the maximum occurring in the late morning (around 11 am), while the minimum in the early morning (between 5 and 8 am).In addition, the presence of three shoulders throughout the year is also found along with the clear diurnal cycle of CO 2 .This should be considered to reflect a dynamic nature of its temporal variabilities.One shoulder near midnight (around 12 am) is likely to be caused by the combined effects of respiration from soils and/or by the living organism.The other two shoulders observed earlier (9 am to 2 pm) and in the evening (6 to 10 pm) should be associated with traffic activities.The timing for these late peaks matches with the peak traffic hours in Seoul.These diurnal patterns of the CO 2 data thus clearly indicate the significance of anthropogenic activities (especially emission from automobiles) to a large extent; even at a mountainous site that is little distant from the highly urbanized sector of Seoul, one of the most populated megacities.It was also found that CO 2 concentration is immensely dependent on traffic activity in Denmark and France, respectively (Soegaard and Moller-Jensen, 2003; Widory and Javoy, 2003).
In Mexico, Velasco et al. (2005) found the CO 2 concentration to be directly related to vehicular traffic because the transportation sector accounts for approximately 60% of emission burden.In this study, the analysis of diurnal trend generally indicates that minimum and maximum values of CO 2 concentrations are in early morning and afternoon, respectively (Fig. 3).However, this pattern is completely opposite to what was found both in background as well as in urban areas in a number of previous studies (Anthwal et al., 2010;Henninger and Kuttler, 2010;Miyaoka et al., 2007;Idso et al., 2002;Aikawa et al., 1995).The authors commonly found its maxi-mum in early morning and minimum in daytime.The unique diurnal trend in this study characterized by the least values in the early morning should be accounted for by the combined effect of several factors (e.g., the nighttime respiration of living organisms and soil layer emission).In contrast, a daytime minimum is suspected to be caused by the photosynthetic activities (Nasrallah et al., 2003;Baez et al., 1988;Spittlehouse and Ripley, 1977) and the expansion of the mixing height (Aikawa et al., 1995).In addition, the appearance of CO 2 peaks near busy traffic hours, as seen in this study, should comply with those in other urban areas under strong anthropogenic activities (such as vehicular emission, and burning of fossil fuels) (Gratani and Varone, 2005;Idso et al., 2002;Takahashi et al., 2002;Aikawa et al., 1995).Many previous studies of CO 2 reported that its concentrations in many urban areas (including Phoenix (Arizona, USA), Essen (Germany), Kuwait city (Kuwait), Taiwan, and Rome) are controlled by vehicular emission to a degree (Henninger and Kuttler, 2010;Lu et al., 2007;Gratani and Varone, 2005;Nasrallah et al., 2003;Idso et al., 2002).As such, the observed diurnal trend of CO 2 in the study area should be accounted for by the typical activities in urban areas.

4 Factors Affecting the Distribution of CO 2
To examine the factors controlling the distribution of CO 2 in the studied mountain area, Pearson correlation analysis was conducted between CO 2 and the environmental parameters determined concurrently (CH 4 , PM 1 , PM 2.5 , PM 10 , NO, SO 2 , and O 3 ) (Table 3a).As one of the most simplified approaches, the daily mean data for all variables were derived initially and used to assess the possible relationship between different parameters.It is now perceived that heating for industry, private automobiles, and landfill can play big roles in carbon emission in Seoul (Sovacool and Brown, 2010).As expected, a strong correlation is observed between major pollutants (like NO x (NO 2 and NO) and SO 2 ).In addition, the relationship between CO 2 and CH 4 data is also evident with a correlation coefficient of 0.37 (p= =3.71E-13, N= =361).The particulate matter (PM) concentration also exhibited a good correlation with the CO 2 data with similar correlation coefficient (r) values between different particles of 0.441 (PM 1 ), 0.491 (PM 2.5 ), and 0.393 (PM 10 ).In contrast, CO 2 data maintained an inverse correlation with that of O 3 without meaningful significance (p= =0.46).The highly strong correlation between CO 2 and other air quality indices suggest that there should be a close relation between the primarily measured trace elements.
All meteorological parameters (e.g., UV radiation, humidity, temperature, and wind speed) showed in- verse correlations with the CO 2 data (Table 3b).Except UV radiation, all of those meteorological parameters are not statistically significant (at p= =0.05 level).However, the CO 2 data are also found to be little affected by such factors as wind speed and wind direction (Figs. 4 and 5).This observation thus signifies the trend that the increases in many meteorological variables are associated with the reduction of CO 2 levels.
The distribution of carbon dioxide in the study area can be affected not only by local sources but also by long range transport from the distant source areas.It was demonstrated previously that the distributions of relatively long lived species (e.g., atmospheric Hg and particulate matters) are influenced by long range transport from surrounding areas both in and out of Korea (Nguyen et al., 2010;Nguyen et al., 2009).It was also found that concentrations of NO and SO 2 in Korea are largely dependent on long range transport between Korea and East China (Shim and Park, 2004;Park and Cho, 1998).Although the main source of energy is petroleum in Japan and Korea, it is still coal for China (more than 75% of total energy source) (Hayakawa, 2009).As such, there is high possibility that the CO 2 concentration in the study area can be affected by long range transport from neighboring China.Future studies may be able to collect more direct evidence of such possibility.

5 Comparison with Previous Studies
To estimate the status of CO 2 pollution in our study area, our data were first compared with those measured previously from other sites on the Korean peninsula.This comparison was extended further to cover some mountainous sites around the world and some background areas.In case of the Korean peninsula, one may refer to the continuous measurements made in Gosan, Jeju (1990Jeju ( -2000) ) (Oh et al., 2001) or in An-Myeon Island (Climate Change Information Center, 2008).The site at Gosan, Jeju is on a small hill (71.2 m) above sea level near the ocean coast (33� 17′N and 126� 10′E).As seen in Fig. 6a, comparison of our data with those of two previous studies showed large differences.The average CO 2 value in Jeju (1990Jeju ( -2000) ) appears to be fairly low (363.8ppm), whereas that for An-Myeon Island (2008) was 391.4 ppm.Because the data are taken between different years, some uncertainties in CO 2 data are expected.However, if we predict the value of Jeju by following its rate of increase (1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000), the value for the year 2009 is 392.7 ppm which is comparable with those of An Myon Island and Mt.GA.
The average annual CO 2 data at our study site (405 ppm) can also be compared with those values taken from World Meteorological Organization (WMO) sites (Fig. 6b).For this comparison, the 2007 annual data from WMO global atmosphere watch (World Data Centre for Greenhouse Gases (WDCGG)) were used.The CO 2 data at Mt. GA are 2.86 to 6.26% higher than all the available WMO data from various mountainous sites, eg, Sonnblick, Austria (381.1 ppm), Deuselbach, Germany (386.1 ppm), Sary Taukum, Kazakhstan (385.3 ppm), Ullan Uul, Mongolia (383.6 ppm), Mt.Kenya, Kenya (379.63 ppm), Srinagar-Garhwal, India (393.4 ppm: Anthwal et al., 2010).We cannot directly balance the difference due to the time gap between our and previous studies.However, as most of these sites are distant from cities, the concentration levels of CO 2 in those sites are lower than our data measured in the urban area.Because Mt.GA is surrounded by a densely populated urban area, its CO 2 level may reflect the man-made activities, especially traffic activities surrounding the mountainous site.
Our CO 2 data were compared further with those derived from many urban areas around the world.It is observed that our data were comparable to those in other urban areas like Rome, Essan (Germany), and Phoenix (AZ, USA).The urban area of Essen, Germany exhibited a wintertime mean value of 415 ppm which is slightly higher than our wintertime data (410 ppm).Likewise, the minimum CO 2 value in our study (mean= =399 ppm) was slightly higher than that of Essan, in the summer of 2004 (mean= =393 ppm) (Henninger and Kuttler, 2010).In contrast, our yearly mean concentration (405±12.1 ppm) was much lower than those of urban areas in Rome, Italy where strong correlation was formed between traffic density and CO 2 (mean yearly value= =477±30 ppm) (Gratani and Varone, 2005).Unlike other urban areas, the CO 2 data measured from Phoenix, AZ, USA was moderately lower (390.2±0.2 ppm) than other areas.Thus, we can see that the CO 2 concentrations in Mt.GA are generally comparable with those obtained from the common urban areas rather than the mountainous (or background) sites around the world.

CONCLUSIONS
In this study, concentration of CO 2 was measured continuously at Mt. GA air quality monitoring station in Seoul, Korea in the year 2009.To describe the basic features of CO 2 distribution, we investigated the factors affecting the environmental behavior of CO 2 in a number of respects.The concentration of CO 2 in the study area averaged as 405±12.1 ppm with its peak occurrence in winter (410 ppm) followed by spring (407 ppm), summer (402 ppm), and fall (399 ppm).The occurrence of winter maximum is now explained by the combined effects of several factors (fossil fuel consumption, reduced plant activity with lower photosynthesis, and stable atmospheric conditions).According to the analysis of diurnal variation, its concentration was the highest during daytime and lowest in the early morning.This trend thus does not comply with those typically observed from clean background areas in which low concentrations are maintained during daytime.
To assess the fundamental factors affecting the environmental behavior of CO 2 , we analyzed relationship between CO 2 and the basic environmental parameters measured concurrently.The CO 2 data generally exhibited strong correlations with most air pollutants, with an exception of O 3 .In contrast, many meteorological parameters tended to exhibit strong inverse correlations with CO 2 , while CO 2 data are affected less significantly by wind speed than others.To learn more about the status of CO 2 pollution in the study area, our results were compared with those determined in various locations in the world.The results of the comparative analysis suggest that the CO 2 levels in this study area are affected fairly sensitively by man-made processes, especially traffic activities in the surrounding areas.

Fig. 2 .
Fig. 2. Plot of daily mean concentration of CO 2 in Mt.GA, Seoul, Korea in 2009.(a) Temporal changes of CO 2 at Mt. GA using hourly data collected throughout 2009; (b) Comparison of daily parameters of CO 2 data in 2007.

Fig. 4 .
Fig. 4. A plot of CO 2 data in relation to wind direction in Mt.GA.

Table 2 .
The mean concentrations of CO 2 and its relative amplitude between seasons.

Table 1 .
Statistical summary of CO 2 and the relevant environmental parameters measured at Mt. Gwan-ak (GA) in 2009.

Table 3 .
The results of correlation analysis between CO 2 and the basic environmental parameters