Continuous Observation of the Mass and Chemical Composition of PM2.5 using an Automatic Analyzer in Kumamoto, Japan

Due to economic growth in China, emissions of gaseous components from factories and automobiles have been increasing, which has resulted in severe air pollution. During the winter and spring seasons, Japan, which is on the leeward side of the Asian continent, is on the receiving end of this increasingly problematic transboundary air pollution. In this study, the mass concentration and chemical components of the particulate PM2.5 were continuously observed using an automatic analyzer at Kumamoto on the west coast of Japan from October 2014 to March 2015. A greater number of high PM2.5 days were observed in winter than in autumn. This seasonal change in concentrations was believed to be due to transboundary air pollution traveling from the Asian continent due to seasonal monsoons. The analysis of the chemical composition of PM2.5 supported this idea. The factors leading to high PM2.5 concentrations were investigated and categorized into transboundary air pollution, local air pollution, and volcanic activity based on the analysis of sulfate (SO4) and sulfur dioxide (SO2) concentrations and model simulations. The average concentration of chemical components showed that local air pollution also influenced air quality in Kumamoto.


INTRODUCTION
Due to the economic development in East Asia, particularly China, the emissions of gaseous components from factories and automobiles have been increasing and have resulted in severe air pollution problems (Ohara et al., 2007;Lu et al., 2010;Kurokawa et al., 2013).In Beijing, the number of days with haze has been increasing from the early 1990s (Zhao et al., 2011).This increase is due to the aerosols generated by the reaction and coagulation of gases emitted from factories and automobiles.Many studies have investigated air pollution by focusing on concentrations, composition, distribution, and sources of aerosols in large cities like Beijing (Okuda et al., 2004;Sun et al., 2004;Wang et al., 2005;Duan et al., 2012;Zhao et al., 2012;Gao et al., 2015) or Shanghai (Lin et al., 2014;Wang et al., 2015).Authors of these papers observed the concentration of sulphate or trace metals in PMs at Chinese cities.And, high concentration of sulphate or metals were observed especially in winter season.
On the other hand, in Japan, the transboundary air pollution transported from these regions in East Asia has become a problem from early 1990's (for example, Husar et al., 2001;Akimoto, 2003).To date, high concentrations of particulate matter (PM) have been observed on the west coast of Japan, such as in the Kyushu region (for example, Uno et al., 2001).This pollution was suspected to be partly due to air pollution traveling from the Asian continent owing to the seasonal monsoons.To investigate the origin of air pollution in the Kyushu region, many studies have been performed using various methods (Takami et al., 2005(Takami et al., , 2007;;Kaneyasu et al., 2010;Takami et al., 2013).Kaneyasu et al. (2010) investigated the air mass including high concentrations of PM 2.5 originating on the Asian continent through observation at Fukuoka and Fukue Island, Japan, and Jeju Island, Korea (Kaneyasu et al., 2010).Takami et al. (2005) analyzed the chemical composition of ambient aerosols using an aerosol mass spectrometer (AMS) at Fukue Island, and they investigated whether PM, including sulfates, originated from the Asian continent by long-range transport and whether the organic components in PM were well oxygenated (Takami et al., 2005(Takami et al., , 2007)).In addition, when individual particle analysis was performed with a time-of-flight secondary ion mass spectrometer, it was shown that black carbon (BC)containing particles were relatively more abundant on Fukue Island than in Tokyo (Takami et al., 2013).PM 2.5 is suspected to be harmful to human health.Especially from 2013, PM 2.5 and its effect for human health has become highly public interest in Japan.Some epidemiologic studies have investigated the relationship between PM and several diseases (Dockery et al., 1993;Kamouchi et al., 2012;Yoda et al., 2015).Kamouchi et al. studied the association between Asian dust, particularly the sand dust from the deserts of China or Mongolia, and the incidence of ischemic stroke.Yoda et al. investigated the effects of PM 2.5 on asthma sufferers.
Recently, researchers have begun to investigate the relationship between PM concentrations and the occurrence of acute myocardial infarction in Kumamoto, Japan.In this study, the authors report the observation results of the mass concentration and chemical composition of PM 2.5 using an automatic analyzer at Kumamoto from autumn 2014 to spring 2015.The factors leading to high concentrations of PM 2.5 and the effects on the chemical composition of PM 2.5 due to transboundary air pollution and local air pollution are also discussed.

Continuous Observation
An automatic PM 2.5 analyzer (ACSA-14: Kimoto Electric Co., Ltd., Japan) was set a top a nine-story building on the Kurokami campus of Kumamoto University in Kumamoto (N 32.8, E 130.7) (Moreno et al., 2012(Moreno et al., , 2013)).This PM 2.5 analyzer is capable of measuring not only the mass concentration of PM 2.5 but also some chemical components in PM 2.5 .Samples were collected hourly on a tape-type filter after the automatic separation of PM 2.5 and PM 10 by a virtual impactor.Optical black carbon (OBC), nitrate ions (NO 3 -), water soluble organic compounds (WSOC), and sulfate ions (SO 4 2-) were observed in this study.The mass concentration of PM 2.5 was measured by the absorption of beta radiation.The mass concentration of OBC was measured by the near-infrared (IR) scattering method.Then, the filtered sample were extracted by the (NH 4 ) 2 SO 4 solution, and introduced into optical cell for optical measurements.The mass concentrations of NO 3 -and WSOC were measured by the ultraviolet (UV) absorption method.The concentration of SO 4 2-was observed by the turbidimetry method with barium chloride (BaCl 2 ).The observation period was October 2014 to March 2015.

Comparison with Filter Sampling
To check the performance of the automatic analyzer, ACSA14, we compared the results of filter sampling data.The filter sampling were performed by a cascade impactor, the Nanosampler (Kanomax, Japan), set adjacent to the ACSA-14.The Nanosampler was pumped at 40 L min -1 by a diaphragm pump (DA-60S: ULVAC KIKO Inc., Japan), and PM was collected on the filter and sampled for one day.The mass and chemical composition of PM 2.5 were analyzed.SO 4 2-in the PM 2.5 was analyzed by ion chromatography (Shimadzu, Japan).This filter sampling was performed from Dec. 17-21, 2014, andfrom Mar. 13-17, 2015.Hourly data of the mass concentration of SO 4 2-obtained by the ACSA-14 were averaged for the time period in which the samples were collected by the Nanosampler, so as to compare to the mass concentration obtained by the Nanosampler.Fig. 1 shows the scatter plot of the mass concentration of SO 4 2-obtained by the Nanosampler compared to those of the ACSA-14.The data observed by the two different methods almost matched.Thus, the mass concentration of SO 4 2-measured by the ACSA-14 was reliable for relative concentration changes.However, the intercept of the plot was about -0.9, and, therefore, the absolute concentration had an error of about 25 % because the average concentration of SO 4 2-was about 3.8 µg m -3 during the observation period.In a previous study using an old version of ACSA (ACSA-08: Kimoto Electric Co., Ltd., Japan), the mass concentration of SO 4 2-was about 40% of the concentration observed by the filter sampling method (Saito et al., 2012).This difference resulted from an insufficient generation of barium sulfate (BaSO 4 ) in the pretreatment.The concentration of SO 4 2-was observed as the concentration of BaSO 4 by the turbidimetry method with BaCl 2 .Therefore, the insufficient generation of BaSO 4 resulted in a lower concentration of SO 4 2-than expected.This problem was solved by mixing in polyvinylpyrrolidone to aid in the generation of BaSO 4 in the subsequent version of the analyzer (ACSA-12: Kimoto Electric Co., Ltd., Japan).Using ACSA-12, the mass concentration of SO 4 2-could be observed quantitatively (Uno et al., 2016).In this study using the current version of the analyzer, ACSA-14, the same pretreatment was used for the improvement of SO 4 2quantitative analysis, which resulted in the good correlation between the ACSA and ion chromatography data (Fig. 1).

Comparing with Calculated Data
Some calculated data were used to estimate the origin of high concentrations of PM 2.5 .For air mass transport, the NOAA Hybrid Single Particle Lagrangian Integrated Trajectory Model (HYSPLIT) (Air Resources Laboratory, Accessed on 9 February 2016), one of the backward trajectory calculation methods, was used.As for the spatial distribution of SO 4 2-, the results of the Chemical Weather Forecasting System (CFORS) (CFORS, Accessed on 9 February 2016) were used.The CFORS is a regional-scale multi-tracer chemical transport model, and the algorithm and application are described in detail by Uno et al. (2003Uno et al. ( , 2004)).

Continuous Observation
The results of the observations by the ACSA-14 from October 2014 to March 2015 are shown in Fig. 2. In this paper, time periods for each season were specified Oct. and Nov. for autumn, from Dec. to Feb. for winter and Mar. for spring.In Fig. 2(a), the mass concentration of PM 2.5 is generally higher in winter and spring than in autumn.In addition, some events of high PM 2.5 observations over 30 µg m -3 occurred in autumn.In Fig. 2(b), the mass concentration of OBC in PM 2.5 is slightly higher in autumn and winter than in spring.In Figs.2(c) and 2(d), the concentrations of NO 3 -and WSOC in PM 2.5 are clearly higher in winter than in autumn.In Fig. 2(e), the concentration of SO 4 2-is also higher in winter, and some peak concentrations appear at the same position as the high PM 2.5 in autumn.SO 4 2-in PM 2.5 is one of the markers of transboundary air pollution, as a large amount of SO 2 is emitted from China (Kurokawa et al., 2013).Sulfuric acid generated from gaseous SO 2 by a photochemical reaction and/or heterogeneous reaction in droplets, form nanoparticles, which become PM 2.5 .Especially in winter, one of the main reasons for high concentrations of PM 2.5 is the combustion of coal that contains sulfur for the heating of homes in the northeast area of China (Kurokawa et al., 2013).It was assumed that the west coast of Japan is affected by this air pollution because of seasonal monsoons from the Asian continent.High concentrations of SO 4 2-detected in this study confirmed that such transboundary transported particles reached Kumamoto, Japan.

Factors Leading to High PM 2.5 Concentrations
Factors leading to high PM 2.5 concentrations were investigated through the composition of PM 2.5 and simulations.Fig. 3(a) shows the mass concentration of PM 2.5 and SO 4 2-in PM 2.5 as measured by the ACSA-14 from Oct. 16-20, 2014.On the 16th, 18th, and 19th, the concentrations of PM 2.5 exceeded 30 µg m -3 .On the 16th and 18th, the concentrations of SO 4 2-were highest.One possibility for the high concentration of SO 4 2-in PM 2.5 is that the PM was transported from the Asian continent (Takami et al., 2005(Takami et al., , 2007;;Kaneyasu et al., 2010).Another possibility for the high concentration of SO 4 2-in PM 2.5 is the effect of volcanic activity (Katsuno et al., 2002).In the Kyushu region, there are several active volcanoes, such as Mt.Aso and Mt.Sakurajima.To investigate the effects of volcanic activity, the concentration of gaseous SO 2 officially reported in the website (AEROS, Accessed on February 2016) was checked.Fig. 3(b) shows the concentration of gaseous SO 2 at the Kyomachi station, the nearest official monitoring station, 2 km west to our monitoring point at Kumamoto University.On Oct. 16, low level of gaseous SO 2 was observed, but on the 18th, a peak concentration of SO 2 was recorded.Thus, the high concentration of SO 4 2on Oct. 18 was caused not only by transboundary air pollution but also by volcanic activity.Chatani et al. (2011) showed the main source of SO 2 in Japan on 2005 was volcanic activity.And now, it assumed that the ratio of the effect of volcanic activity became higher because anthropogenic emission is reduced by the effect of laws in Japan, for example, Air Pollution Control Act, and volcanic activity become more active for example Mt.Aso or Mt.Sakurajima.Therefore, the emission of high concentration SO 2 means the effect of volcanic activity.On Oct. 19, a relatively low concentration of SO 4 2-was observed.Therefore, the levels were not highly affected by transboundary air pollution and volcanic activity.Local air pollution was the main cause of the high concentration of PM 2.5 on Oct. 19.
The mass concentrations of PM 2.5 and SO 4 2-in PM 2.5 measured from Dec. 17-22, 2014, are shown in Fig. 4(a).The highest concentrations of PM 2.5 were observed on the 19th and 20th.The concentrations of SO 4 2-were also high on these two days.These high concentrations of SO 4 2-were caused by transboundary air pollution or volcanic activity.The gaseous SO 2 concentration change observed in the same period at the Kyomachi station is shown in Fig. 4(b).On the 19th, a peak concentration of SO 2 was observed.This high concentration of SO 2 was caused by volcanic activity, same as observed on Oct. 18, 2014.
To confirm the causes of the high PM 2.5 concentrations shown by the experimental results described above, model simulations were conducted.First, backward trajectory analysis was performed.This phenomenon is consistent with the observation that the high concentration of PM 2.5 on Oct. 18 was mainly caused by transboundary air pollution.On the other hand, on Oct. 19 and Dec. 19, air mass was transported from within Japan.Thus, the high concentrations of PM 2.5 on these days were not caused by transboundary air pollution.On Dec. 20, air mass was transported from within Japan in the morning, then, due to changing meteorological conditions, air mass was transported from the Asian continent in the afternoon.The peak concentration of PM 2.5 began in the afternoon, and therefore the results of the backward trajectory analysis were consistent with the observation results.To investigate the spatial distribution of SO   high SO 4 2 concentrations had not yet reached Japan.The simulation results on Dec. 19 are consistent with the observation that high concentrations of PM 2.5 were caused by volcanic activity, because in the simulation, the effect of the latest volcanic activity in 2014 at Mt. Aso, Kumamoto, was not considered, and, therefore, the SO 4 2-caused by volcanic activity was not reproduced in the simulation in Fig. 6(c) (Uno et al., 2003).
Consequently, the factors leading to high concentrations of PM 2.5 were classified as transboundary air pollution, , and the data of gaseous SO 2 and NO x observed at the Kyomachi station are shown in Figs.7(e) and 7(f).In Fig. 7(a), the concentration of OBC is high from 8:00 a.m. to 10:00 a.m., becomes low in the afternoon, and becomes high again in the evening after 5:00 p.m.The same variability is seen in the case of gaseous NO x in Fig. 7(f).The periods of high concentrations of these components matched the morning commute time when people travel by various types of vehicles from home to office, and, therefore, the main origin of these components was local vehicle exhaust.The concentration variation of NO 3 -in PM 2.5 is expected to have the same variability of NO x , because NO x is converted to NO 3 -and it is expected that the concentration of NO 3 -in PM 2.5 depends on the concentration of NO x .Yet in Fig. 7(b), the concentration variation of NO 3 -appears mostly flat for the entire day, and does not perform similarly to the concentration variation of NO x .This difference in concentration changes is caused by the reactivity of NO x (Mulcahy and Smith, 1971;Westenbery and Haas, 1972;Anderson et al., 1974).The reaction rate of NO x to nitric acid is very slow due to the low photochemical reaction in winter; therefore, the concentration variation of NO x was not reflected in the concentration variation of NO 3 -.The concentration variations of WSOC, SO 4 2-, and SO 2 were almost flat because the main origin of these components was transboundary air pollution, except for occasional volcanic activity.As a result, at Kumamoto, the concentration of PM 2.5 depended on both the transboundary air pollution and the local air pollution.
Fig. 8 shows the histogram of the PM 2.5 concentrations from October 2014 to March 2015.The highest concentration frequency was in the range of 15 to 20 µg m -3 .In addition, a high concentration above 35 µg m -3 was observed for only five days in the observation period.Thus, PM 2.5 pollution is not severe in Kumamoto.However, there were observed events of high concentrations of PM 2.5 including high concentrations of SO 4 2-.Transboundary air pollution was highly related to these events.Therefore, to reduce PM 2.5 concentrations, it is necessary to take measures to reduce transboundary air pollution from Asian continent.In addition, it is important to control local emissions from automobile exhaust or fuel combustion.

CONCLUSIONS
In Kumamoto in western Japan, the mass concentration and chemical components of PM 2.5 were continuously observed using an automatic analyzer from October 2014 to March 2015.High concentrations of PM 2.5 were observed more often in winter than in autumn due to transboundary air pollution from the Asian continent due to seasonal monsoons.The chemical composition of PM 2.5 supported these observation results.The factors leading to high PM 2.5 concentrations were investigated and categorized into transboundary air pollution, local air pollution, and  2-and SO 2 concentrations as well as model simulations.The average concentrations of chemical components showed that local air pollution also influenced the air quality in Kumamoto.Therefore, to reduce PM 2.5 concentrations, it is necessary to take measures to reduce transboundary air pollution from China, and to control the local emissions from automobile exhaust or fuel combustion of power plants.

Fig. 1 .
Fig. 1.Comparison of the mass concentration of SO 4 2-measured by the Nanosampler and the ACSA-14.

Fig. 4 .
Fig. 4. Continuous observation of the concentrations of (a) PM 2.5 and SO 4 2-in PM 2.5 measured by the ACSA-14 and (b) SO 2 at the Kyomachi station from Dec. 17-22, 2014.