Research Paper
Earozoru Kenkyu, 30（2） , 126–133（2015）doi: 10.11203/jar.30.126
Chemical Characteristics of Airborne Particulate Matter during the Winter Season in Ulaanbaatar Masataka NISHIKAWA 1,2*, Ichiro MATSUI 1, Ikuko MORI 3, Dashdondog BATDORJ 4, Enkhmaa SARANGEREL 4, Kaoru OHNISHI 1, Atsushi SHIMIZU 1 and Nobuo SUGIMOTO 1 Received 18 December 2014 Accepted 22 April 2015 Abstract Atmospheric pollution by airborne particulate matter (PM) in Ulaanbaatar, particularly in the winter season, is very serious problem in Mongolia. The chemical compositions of PM samples were measured in Ulaanbaatar in the winters of 2012 and 2013. The concentration trends for both PM2.5 and PM10 in the city center were similar and the PM2.5/PM10 ratios were about 0.8. Arsenic, Pb and Zn were weighted toward the fine size range, whereas Ca, Fe and Mn dominated the coarse size range. The PM ratios for acidic ions, the ions assumed to be derived from coal combustion, were about 0.8, except for fluoride. The average winter concentrations for PM10 samples collected simultaneously at three separate locations in the city were 167 μg/m3 for the city center, 422 μg/m3 for an area dominated by Gers housing (traditional Mongolian yurt dwellings houses) and 155 μg/m3 for an area with traffic congestion. Total carbon represented the largest component of PM, accounting for approximately half of the total. The organic carbon (OC) concentration of PM10 samples in the Gers area was significantly higher than that of other areas. Within the five fractions of OC, the OC1 fraction with gasification at low temperature (120°C) in the Gers area was consistently higher than in other areas. Keywords : Airborne Particulate Matter, Organic Carbon, Chemical Composition, Coal Combustion, Ulaanbaatar.
1.
of PM10 in the world. Atmospheric pollution in winter is
Introduction
particularly severe and, consequently, various environmental
The Mongolian capital, Ulaanbaatar, is located in a basin
improvement strategies 2,3） have been planned as evidenced
surrounded by mountains, which can result in the formation
by several foreign-aids budgets including JICA (Japan Inter-
of an inversion layer. Consequently, atmospheric pollutants
national Cooperation Agency). It has been reported 2）that the
generated at ground level do not readily disperse resulting in
emission sources for PM in Ulaanbaatar are uncomplicated.
a build-up of concentrations. Due to this topographic situa-
The primary sources are particles and dusts emitted from
tion, Ulaanbaatar was listed by WHO in a 2011 report 1） as
coal combustion, which is the main source of heat. PMs from
the largest city with the highest annual average concentration
other sources include fine particles from vehicle exhausts, which have increased rapidly of late and soil dusts that are disturbed by passage of vehicles along roads. Chemical
National Institute for Environmental Studies 16-2, Onogawa, Tsukuba, Ibaraki 305-8506 2 Tokyo University of Science 12-1, Ichigaya-Funakawara, Shinjuku, Tokyo 162-0826 3 Overseas Environmental Cooperation Center, Japan 3-25-33, Nishi-Shinbashi, Minato, Tokyo 105-0003 4 National Agency for Meteorology and Environment Monitoring (NAMEM) Juulchny, Guidamj-5, Ulaanbaatar-46, 210646, Mongolia * Corresponding Author. E-mail:
[email protected] (M. Nishikawa) 1
126
analysis of PM10 samples collected simultaneously from three sites̶a Gers area where many domestic stoves are in use, an area where traffic congestion is severe and the city center̶ was carried out and an attempt was made to elucidate the chemical compositional characteristics of the samples from each area. The results may provide useful scientific data to aid in the amelioration of urban airborne pollution due to PM10 and PM2.5 in Ulaanbaatar. （46）
エアロゾル研究
2.
scattering method: KOSA Monitor, TOA-DKK Co., Tokyo,
Experimental Methods
Japan), MOR (meteorological observation range; visibility
Multiple low volume samplers (Model 2000-FRM; R&P,
sensor of forward scatter method: PWD, VAISARA OYJ.,
part of Thermo Fisher, NY, USA) fitted with quartz fiber
Helsinki, Finland), and wind (ultrasonic method: PGWS-100,
filters (2500 QAT-UP; Pall Science Co., MI, USA) were
Hampshire, U.K.) on the rooftop of NAMEM building at site
used for sampling at three locations: the city center (site 1),
1 have been carried out from 2008.
a Gers area where many domestic stoves are in use (site 2)
3.
and an area where traffic congestion is severe (site 3). The locations of the sampling sites are shown in Fig. 1. Sampling
3.1
Results and Discussion
Daily PM variations in winter
for PM10 and PM2.5 was performed on a daily basis (23 hours
Fig. 2 shows an example of winter monitoring of PM2.5,
from around midday local time (starting between 11 am
visibility and wind velocity at site 1 (City center). Visibility
and 1 pm)). Key elements of the sampling protocol were as
measurements are extremely sensitive and highly responsive
follows:
with respect to time resolution, although factors affecting
1) Simultaneous and parallel sampling of PM2.5 and PM10
visibility such as rain, snow, fog and PM cannot be distin-
in the city center (site 1) was carried out on a continu-
guished. They can be employed as reference data for judging
ous daily basis over the period January 19–26, 2012.
PM pollution when poor visibility occurs in Ulaanbaatar.
2) Simultaneous and parallel sampling of PM10 was under-
A cyclical variation, whereby visibility worsens at night,
taken at three sites (sites 1, 2, and 3) a total of eight
is clearly apparent and PM2.5 varies in a similar manner.
times, at intervals of around 10 days, between Decem-
In general, PM including PM2.5 and PM10 also varied on a
ber 2012 and February 2013. These three sites were
daily cyclical basis following similar patterns to variations
considered to exhibit typical characteristics for their lo-
in visibility. This phenomenon is conspicuous when wind
calities in Ulaanbaatar. Sampling was also performed
speed is low, regardless of wind direction, and where charac-
three times during the summer (May to June 2013) to
teristically high concentrations of PM pollutants are formed
allow comparison with winter sampling.
under an inversion layer 4）. We have come to understand that
The collected samples were each divided into four equal
such daily variations are a typical phenomenon associated
portions: in one portion the ionic components were ana-
with winter conditions based on the findings from continuous
lyzed and in another two portions various trace elements
measurements at site 1 over many years. The daily fluctua-
were determined according to a published method 4）. For
tions in PM concentrations tend to be small around midday
the remaining portion, organic carbon (OC) and elemental
and, therefore, it was decided to commence daily sampling
carbon (EC) were determined using a thermal/optical carbon
of PM by a low volume sampler during the period 11 am to 1
analyzer (Model, 2001; DRI Co., NV, USA) with the sep-
pm.
aration parameters set according to the IMPROVE method
3.2
of PM10 and PM2.5. A high correlation was observed for
In addition, continuous hourly observations of PM2.5 (light
Fig. 1 Vol. 30 No. 2（2015）
PM2.5 and PM10 in central Ulaanbaatar
Fig. 3a shows measurements from simultaneous sampling
conditions .
5）
Location map of PM sampling sites in Ulaanbaatar. （47）
127
Fig. 2
Monitoring results (hourly basis) for PM10 and PM2.5, and visibility and wind speed at the City center site (site 1) in December 2012.
Fig. 3a Average concentration changes (daily basis) for PM10 and PM2.5 at site 1 for the period January 19–26, 2012.
Fig. 3b Relationship of hourly concentrations for PM10 and PM2.5 at site 1 for the period January 10–31, 2012.
concentration variations of PM2.5 and PM10 (r＝0.88). Fig. 3b shows the correlation of hourly observations between PM2.5 and PM10 in January, 2012. The slope of PM2.5/PM10 gave a k value equal to 0.77 and the correlation factor r was 0.99. On this basis of good correlation between PM2.5 and PM10, it was unlikely that the emission sources for the two particle classes were different. Concerning the fluctuations of PM2.5 and PM10 in the wintertime, Ulaanbaatar has been reported 6） to exhibit similar patterns on a daily basis, giving maximum peaks at midnight and in the morning, and minimum peaks Fig. 4
for a few hours from midday. Fig. 4 shows the average PM2.5/PM10 concentration ratios (PM ratio) for each component and their respective standard deviations. The components that had PM ratios notably smaller than the average PM ratio of 0.8 (±0.07) had 128
（48）
PM2.5/PM10 concentration ratios of chemical components based on daily sampling at site 1. Vertical bars for each component represent the standard deviation range (n＝7). Chain line represents the average values for the PM2.5/PM10 mass concentration ratio. エアロゾル研究
Fig. 5 Average concentration changes (daily basis) for PM10 at site 1, site 2 and site 3.
exceptionally high values for the coarse particles ( ＞ PM2.5).
emissions in the Gers area is the coal and firewood used in
The concentration ratio of PM2.5/PM10 for F ions, derived
domestic stoves.
, was similar to those ratios (0.3–0.4)
The following chemical constituents all exhibited strong
for Ca, Fe, and Mn which are considered to originate from
correlation (r ＞ 0.8 ) with variation in PM10 concentrations
soil particles existing mainly in the coarse particle size range.
for each area:
The characteristics for the ratio for F ion contrast with the
Site 1: OC, EC, K＋, NH4＋, F-, SO42-, NO3- (7 components)
PM ratios for ionic components (NO3 , SO4 , Cl ), which
Site 2: OC, EC, K＋, NH4＋, F-, SO42-, Zn, Pb (8 components)
are around 0.8, these components also originating from
Site 3: OC, EC, K＋, Mg2＋, NH4＋, F-, SO42- (7 components)
from coal combustion
4,7）
-
2-
-
combustion sources. Further, the ratios for trace elements (Zn,
When the source of PM10 emission source is uncom-
As, Pb) are greater than 0.8, indicating such elements have
plicated, the aforementioned components that had strong
a tendency to be present on particles of extremely fine size
correlation with PM10 concentrations exhibited relatively
range confirming such elements were derived from human
small relative standard deviations (RSDs) compared with
activities.
the RSD values for the PM10 concentrations. In other words,
3.3
Site-specific PM10 concentrations and composition
it is important to document which chemical components in
ratios
PM10 samples are stable and of major concentration in order
As discussed in the previous section, emission sources
to infer the extent of pollution at each site. Based on the
for PM10 and PM2.5 must be of similar origin given that there
samples collected in the winter period (as shown in Table 1),
was a strong correlation of the two categories of PM at
Table 2 summarizes the respective average concentration
Ulaanbaatar. It is hoped that by elucidating the results for the
ratios, the standard deviations and the RSDs for PM and the
distribution of PM10 in the city, this will aid in identifying the
chemical constituents. For all three sites, the concentration
distinguishing features and emission sources of PM2.5.
ratios for total carbon (sum of OC and EC), NH4＋ and SO42-
Fig. 5 shows the variation in PM10 concentrations for
were largest and, moreover, the RSDs were smaller than
samples collected simultaneously at the three locations, while
the RSD for the PM10 concentrations, indicating that these
Table 1 summarizes the respective chemical compositions.
components behaved similarly in the urban atmosphere. As
All chemical components represented in Table 1 for each
to the high correlation (r ＞ 0.94) for NH4＋ and SO42- (Table 1)
site in the winter (heating period) had significantly higher
at all sites, it is most likely that discrete particles of (NH4)2SO4
concentrations than in the summer (non-heating period).
were present in the urban atmosphere.
These large concentrations differences for ionic components,
Total carbon is the largest component of the winter PM10
and NH4 , the heavy metals Zn and Pb,
particles accounting for 40–50% of the total. Fig. 6 summa-
and OC and EC, must be caused by the combustion of coal
rizes the variations in the OC and EC concentrations. The
as has been reported 4）. The average PM10 concentrations for
total carbon concentration at site 2 in winter was significantly
all samples collected in the winter were as follows: site 1,
higher than at the other sites (1, 3); this result implies the
167 μg/m3; site 2, 422 μg/m3; site 3, 155 μg/m3. The average
existence of a major emission source in the vicinity of site
concentration at site 2 was more than double that at the other
2. Furthermore, the concentration variations at sites 1 and
two sites. Fig. 5 shows that such disparate concentrations
3 were very similar, which implies the absence of major
are not evident in the comparative samples for the summer
emission sources of total carbon in their immediate vicinities.
period. Besides traditional domestic stoves, we are unable
Significance tests conducted for the OC composition ratios
to identify any other PM emission sources in the Gers area.
of all winter samples show that the differences in OC com-
On this basis, therefore, it is concluded that the source of PM
position ratios between site 2 and site 1, and between site 2
such as F , SO4 -
2-
Vol. 30 No. 2（2015）
＋
（49）
129
Table 1 Chemical components in PM10 samples at each site for the period 17 December 2012–17 February in the winter season, and the period 18 May–29 June 2013 in the summer season Sampling Date
Sample
PM10 OC EC Ca2＋ K＋ Mg2＋ Na＋ NH4＋ FClSO42- NO3Zn Pb Fe (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (ng/m3) (ng/m3) (ng/m3)
Sample 1 2012/12/17 Site 1
225
64
18
1.93
0.44
0.18
0.17
10.8
0.21
1.49
22.3
7.83
158
42.0
1220
2012/12/17 Site 2
502
200
34
2.28
0.70
0.21
0.24
17.9
0.39
3.48
28.1
5.55
259
82.7
1020
2012/12/17 Site 3
181
65
18
2.06
0.41
0.18
0.21
9.78
0.20
1.22
21.4
7.96
111
34.4
1300
Sample 2 2012/12/25 Site 1
no sample
2012/12/25 Site 2
518
210
31
1.68
0.65
0.19
0.25
15.9
0.50
4.41
26.9
5.82
232
74.1
945
2012/12/25 Site 3
189
77
17
1.36
0.30
0.17
0.59
7.24
0.21
2.15
14.2
4.31
128
29.8
995
189
63
15
0.99
0.30
0.15
0.44
8.03
0.19
1.61
14.9
4.96
112
26.2
773
2013/1/3 Site 2
500
230
43
1.38
0.70
0.19
0.30
18.4
0.51
4.53
29.0
6.24
319
87.7
822
2013/1/3 Site 3
141
56
14
1.02
0.27
0.15
0.43
7.12
0.19
1.44
13.3
4.86
90.0
24.4
841
244
79
19
1.86
0.37
0.20
0.18
9.68
0.25
1.62
19.1
6.90
142
38.9
1520
2013/1/12 Site 2
478
200
30
1.75
0.70
0.20
0.27
16.2
0.36
3.66
31.0
8.95
252
83.1
1130
2013/1/12 Site 3
218
69
21
1.54
0.33
0.19
0.20
9.23
0.26
1.66
17.3
7.21
122
35.5
1530
125
37
10
1.59
0.21
0.17
0.98
4.47
0.14
2.35
9.68
4.48
142
57.1
1290
2013/1/22 Site 2
408
180
23
2.14
0.55
0.20
0.77
12.3
0.27
4.76
23.7
7.96
265
73.6
1530
2013/1/22 Site 3
121
37
12
1.62
0.20
0.16
1.14
4.42
0.12
2.48
9.51
4.34
142
33.7
1380
2013/2/2 Site 1
160
49
12
1.61
0.24
0.18
1.22
6.51
0.22
2.82
12.9
7.26
130
25.2
1740
2013/2/2 Site 2
409
170
22
1.80
0.62
0.19
0.70
16.3
0.29
4.60
31.0
9.96
198
62.7
1100
2013/2/2 Site 3
151
48
14
1.62
0.25
0.18
1.09
6.43
0.20
2.43
13.2
7.30
99.8
24.2
1830
2013/2/9 Site 1
148
53
12
1.28
0.25
0.15
0.58
5.85
0.12
2.25
11.9
3.39
90.0
24.5
877
2013/2/9 Site 2
404
160
28
1.71
0.61
0.19
0.49
15.5
0.34
5.45
29.0
7.21
187
57.9
1060
2013/2/9 Site 3
no sample
Sample 3 2013/1/3 Site 1
Sample 4 2013/1/12 Site 1
Sample 5 2013/1/22 Site 1
Sample 6
Sample 7
Sample 8 2013/2/17 Site 1
77.5
26
8
0.93
0.15
0.13
0.52
3.72
0.06
1.25
8.39
2.04
49.3
15.0
798
2013/2/17 Site 2
156
61
14
1.77
0.24
0.17
0.30
5.37
0.10
1.55
13.4
2.68
89.3
21.9
1330
2013/2/17 Site 3
81.6
26
9
1.00
0.16
0.13
1.00
3.70
0.05
1.98
8.52
2.13
50.7
11.1
917
Sampling Date
Sample
OC EC Ca2＋ K＋ Mg2＋ Na＋ NH4＋ FClSO42- NO3Zn Pb Fe PM10 (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (μg/m3) (ng/m3) (ng/m3) (ng/m3)
Sample 9 2013/5/18 Site 1
33.2
2013/5/18 Site 2
64.7
5
2
1.42
0.07
0.12
0.10
0.65
0.03
0.63
1.79
0.90
16.6
8.4
n.a.
2013/5/18 Site 3
234
31
14
8.26
0.31
0.31
0.56
1.28
0.08
1.61
5.57
1.93
93.3
27.4
n.a.
86.3
8
3
2.63
0.08
0.15
0.16
0.34
0.03
0.34
1.54
0.64
32.2
10.5
n.a.
10
3
1.13
0.05
0.10
0.13
0.60
0.02
0.21
2.00
0.53
17.3
5.8
n.a.
Sample 10 2013/5/21 Site 1 2013/5/21 Site 2
238
29
13
5.75
0.17
0.29
0.29
0.58
0.05
0.71
5.12
0.92
38.8
15.0
n.a.
2013/5/21 Site 3
151
13
7
4.82
0.15
0.20
0.28
0.36
0.05
0.65
2.21
0.86
52.8
12.4
n.a.
Sample 11 2013/6/29 Site 1
no sample
2013/6/29 Site 2
24.5
7
2
0.69
