Introduction
Tropospheric aerosols are important because of their strong influence on the
climate system through both direct and indirect effects. These include the
direct effect of scattering and absorbing radiant energy, and the indirect
effect of modifying the microphysical properties of clouds, and hence their
radiative properties and lifetime (Haywood and Boucher, 2000). They
also attract attention because of their adverse effects on visibility
(Watson, 2002) and human health (Delfino et al.,
2005; Pope III and Dockery, 2006). Therefore, the spatial and temporal
variation of aerosols is essential to understand, but remains a complex
subject because of their ephemeral nature and the complexity of their
physical and chemical properties (Ramanathan et al., 2001).
Particle size is considered a key parameter to define the impact of
particulate matter (PM) on human health; specifically, fine PM (PM2.5
and PM1) poses a greater health risk than coarse PM (PM10)
(Oberdörster et al., 2005). There have been numerous
network-based observation studies of the PM2.5 concentration and
chemical composition in North America and Europe. For example, based on a
data set across 19 Canadian sites, most of the PM2.5 concentrations were
found to be below 26 µgm-3, and PM2.5 accounted for 49 %
of the measured PM10 (Brook et al., 1997). Meanwhile,
Eldred et al. (1997) reported that PM2.5 and PM10 particulate
concentrations measured at 42 sites of the Interagency Monitoring of
Protected Visual Environments (IMPROVE) network over the 1993 seasonal year
(March 1993 to February 1994) showed the PM2.5 concentration had a
large gradient from west to east in the US, averaging 3 µgm-3 in
most of the west compared with 13 µgm-3 in the Appalachian
region. Another study, based on 143 IMPROVE sites in the year 2001, showed
that sulfates, carbon and crustal material were responsible for most of the
measured PM2.5 at the majority of sites in the US (Malm
et al., 2004). The temporal variation and spatial distribution of PM2.5
concentrations have also been reported in Switzerland (Gehrig and
Buchmann, 2003), Austria (Gomiscek et al., 2004), and six
central and eastern European countries (Houthuijs et al., 2001).
As a country with a rapidly developing economy, China has suffered from a
serious air pollution problem in recent years due to substantial increases
in energy consumption and other related production of large amounts of
aerosols and precursor gas emissions (Zhang
et al., 2009). At the coarse end of the spectrum (PM10), the spatial
distribution and interannual variation of concentrations has been
comprehensively studied using a data set accumulated from 86 Chinese cites
(Qu et al., 2010). Furthermore, the chemical
compositions of PM10 samples were investigated at 16 sites over China,
and the result indicated a dominant scattering feature of aerosols
(Zhang et al., 2012). Network-based studies of
PM2.5 observations have, however, been limited to certain seasons in a
single year (Cao et al., 2012), and most other research has
focused on one or more of the largest cities (He et al., 2001; G. Wang et
al., 2002; X. Wang et al., 2006; Wei et al., 1999; Yao et al., 2002; Zhao et
al., 2009; Zheng et al., 2005). The focus on PM2.5 needs to improve,
not least because the growing problem of heavy haze has compelled the
Chinese government to pay greater attention to PM2.5 monitoring and air
quality standards. Indeed, the Ministry of Environmental Protection of China
issued new ambient air quality standards in 2012, among which the PM2.5
concentration was the first to be included. Subsequently, the construction
of a network of national environmental PM2.5 monitoring stations began
in 2013.
In this paper, we present a long-term PM10, PM2.5 and PM1
monitoring data set from 2006 to 2014, based on 24 stations of CAWNET (China
Atmosphere Watch Network), operated by the China Meteorological
Administration (CMA). The spatial pattern of average PM concentration levels
and the relationships among them are reported. In addition, their seasonal
and interannual variations are presented.
Description of the PM stations.
Stations
Latitude (∘ N)
Longitude (∘ E)
Altitude (m)
Start time
Description
Zhengzhou
34.78
113.68
99.0
Jan 2006
Urban, in the center of Zhengzhou, 56 m building.
Chengdu
30.65
104.04
496.0
Mar 2006
Urban, in the center of Chengdu, 91 m building.
Xian
34.43
108.97
363.0
Jan 2006
Urban, in northern margin of Xian, 20 km north of center of Xian, 4 m sampling container.
Nanning
22.82
108.35
84.0
Jan 2006
Urban, in Nanning, 140 m hill.
Anshan
41.05
123.00
78.3
Oct 2007
Urban, in Anshan, 10 m building.
Shenyang
41.76
123.41
110.0
Oct 2007
Urban, in Shenyang, 15 m building.
Benxi
41.19
123.47
185.4
Oct 2007
Urban, in Benxi, 12 m building.
Fushun
41.88
123.95
163.0
Oct 2007
Urban, in Fushun, 10 m building.
Qingdao
36.07
120.33
77.2
Mar 2007
Urban, in Qingdao, top of Fulong Shan hill.
Lhasa
29.67
91.13
3663.0
Jan 2006
Urban, in Lhasa, 7 m building.
Siping
43.18
124.33
165.4
Mar 2007
Urban, in Siping, 4 m sampling container.
Panyu
23.00
113.35
5.0
Jan 2006
Suburban, in the Panyu district of Guangzhou, 140 m hill.
Gucheng
39.13
115.80
15.2
Jan 2006
Suburban, 38 km southwest of Baoding, within area of rapid urbanization, 8 m building.
42.27
118.97
568.0
Mar 2007
Rural, suburbs of Chifeng, 4 m sampling container.
Dandong
40.05
124.33
13.9
Mar 2007
Rural, suburbs of Dandong, 4 m sampling container.
Erlianhaote
43.65
111.97
965.9
Mar 2007
Rural, suburbs of Erlianhaote, 4 m sampling container.
Yulin
38.43
109.20
1135.0
Jan 2006
Rural, 10 km north of Yulin, at the southeastern edge of Mu Us desert.
Jinsha
29.63
114.20
416.0
Apr 2006
Rural, 105 km north of Wuhan, 8 m building.
Guilin
25.32
110.30
164.4
Jan 2006
Rural, north margin of Guilin, meteorological observation field.
Lushan
29.57
115.99
1165.0
Jan 2006
Rural, Kuniubei peak of Mount Lu.
Changde
29.17
111.71
563.0
Jan 2006
Rural, 18 km northwest of Changde, 8 m building.
Dongtan
31.50
121.80
10.0
May 2009
Rural, east of Chongming Island near Shanghai.
Akdala
47.12
87.97
562.0
Sep 2006
Remote, 55 km west of Fuhai, 10 m building.
Shangri-La
28.02
99.73
3580.0
Oct 2006
Remote, 12 km northeast of Shangri-La.
Results and discussion
Average PM10, PM2.5 and PM1 levels in China
The averaged PM concentration values are presented in Table 2, and their
distributions in Fig. 1. The highest PM10, PM2.5 and PM1
concentrations were observed at the stations of Xian (135.4, 93.6 and 77.0 µgm-3,
respectively), Zhengzhou (131.7, 84.8 and 71.0 µgm-3, respectively) and Gucheng (127.8, 89.7 and 79.4 µgm-3,
respectively), which are located in the most polluted areas of
the Huabei Plain (HBP) and the Guanzhong Plain. Although Gucheng is a suburban site, it is located in the rapid urbanization area around Beijing, and is
therefore subjected to associated large quantities of air pollutants. These
areas were also identified by Zhang et
al. (2012) as having experienced similar visibility changes and large
visibility loss in the past 40 years. The stations all recorded very high
coarse- and fine-PM concentrations, implying high emissions of both primary
emitted mineral particles and secondary anthropogenic particles in these
areas. Qingdao is a coastal city with relatively low PM concentrations
compared with inland cities on the HBP.
The PM concentrations were also high in northeast China, which is an
established industrial area. The ensemble average values of the five urban
stations of Ansan, Shenyang, Benxi, Fushun and Siping were 88.8, 58.4 and
49.8 µgm-3, for PM10, PM2.5 and PM1, respectively.
Dandong is a rural station with relatively low PM concentrations.
The similarity among the PM values for Chifeng, Erlianhaote and Yulin is due
to their location, far from regions of intensive economic development but
strongly affected by sandstorms and dust storms given their proximity to dust
source areas. Thus, the average PM10 concentrations were much higher
than the PM2.5 and PM1 concentrations at these sites. For example,
the average PM10 concentration at Chifeng, which is surrounded by sandy
land, was 88.0 µgm-3, compared with 42.4 and 32.6 µgm-3 for PM2.5 and PM1, respectively.
Chengdu, the capital of the province of Sichuan, is located in the Sichuan Basin,
another highly polluted area. High aerosol optical depth and low visibility,
due to the poor dispersion conditions and heavy local industrial emissions,
have been reported for this site (Li et al., 2003; Luo et al., 2001;
Zhang et al., 2012). In the present study, the average PM10, PM2.5
and PM1 concentrations were 78.0, 59.5 and 52.7 µgm-3,
respectively.
There are three stations in the south China area: Panyu, located in
Guangzhou, the capital of the province of Guangdong, which is the center of
the Pearl River Delta region; Nanning, the capital of the province of Guangxi; and
Guilin, a famous tourist city, also located in Guangxi. The
ensemble average PM concentrations of these three sites were 55.8, 43.1 and
38.8 µgm-3 for PM10, PM2.5 and PM1, respectively.
Significant visibility loss and relatively high PM10 concentrations
have been observed over the middle and lower reaches of the Yangtze River
after the 1980s due to the rapid economic development that has taken place
in this region (Qu et al., 2010;
Zhang et al., 2012). Although there was no urban site available for this
study to help quantify the high PM concentrations in this region, the
background conditions and temporal variance could be determined from the
rural site data. Dongtan, near Shanghai, is located on Chongming
Island, where there were low PM concentrations (31.9, 27.4 and 24.8 µgm-3 for PM10, PM2.5 and PM1, respectively) due to the
substantial influence of clean sea air mass. The ensemble average PM
concentrations for Lushan, Changde and Jinsha were 44.3, 37.2 and 33.6 µgm-3 for PM10, PM2.5 and PM1, respectively.
Lhasa, the capital of Tibet Autonomous Region, is located in the center of
the Tibetan Plateau at a very high altitude of 3663 m. The PM2.5 and
PM1 concentrations in Lhasa were low, with average values of 14.0 and
9.6 µgm-3, respectively, because of its relatively small
population and few industrial emissions. However, the average PM10
concentration was 37.7 µgm-3, mainly due to the high amounts of
fugitive dust from dry and bare land and the impacts of regional dust storm
events (Chen et al., 2013). As a result, minerals are the main
constituent of aerosol samples in this area (Zhang
et al., 2012).
The lowest PM concentration values were observed in the two remote sites of
Akdala and Shangri-La. The lower altitude and stronger contribution of soil
dust at Akdala (Qu et al., 2009), located in a dry region, lead to
higher PM concentrations than at the Shangri-La site.
Averaged PM10, PM2.5 and PM1 concentrations and
their interrelationships at each station. SB = substandard.
Stations
Averaged PM concentrations (µgm-3)a
SB ratio
SB ratio
PM2.5 /
PM1 /
R2 (PM2.5
R2 (PM1
PM10
PM2.5
PM1
(PM10)b
(PM2.5)b
PM10
PM2.5
to PM10)
to PM2.5)
Zhengzhou
131.7 (84.4)
84.8 (47.4)
71.0 (40.5)
0.31
0.51
0.68
0.84
0.68
0.91
Chengdu
78.0 (72.5)
59.5 (42.2)
52.7 (35.4)
0.11
0.27
0.83
0.91
0.76
0.94
Xian
135.4 (97.3)
93.6 (67.3)
77.0 (55.6)
0.34
0.52
0.73
0.83
0.77
0.93
Nanning
51.2 (56.3)
38.4 (24.7)
34.9 (22.2)
0.01
0.08
0.77
0.91
0.52
0.97
Anshan
97.8 (62.9)
60.9 (42.9)
52.3 (39.0)
0.17
0.25
0.65
0.85
0.72
0.98
Shenyang
85.0 (58.2)
59.1 (42.7)
50.8 (36.7)
0.11
0.25
0.69
0.85
0.88
0.97
Benxi
97.6 (57.4)
66.7 (45.0)
54.8 (36.4)
0.13
0.30
0.69
0.82
0.81
0.94
Fushun
80.3 (54.2)
50.1 (31.7)
42.8 (28.3)
0.07
0.17
0.66
0.85
0.64
0.97
Qingdao
64.8 (52.1)
47.3 (34.0)
41.1 (30.5)
0.05
0.17
0.76
0.86
0.76
0.95
Lhasa
37.7 (30.8)
14.0 (10.7)
9.6 (8.6)
0.01
0.00
0.40
0.66
0.72
0.94
Panyu
58.7 (33.1)
44.5 (24.4)
39.7 (22.1)
0.02
0.12
0.77
0.89
0.95
0.98
Gucheng
127.8 (75.1)
89.7 (53.0)
79.4 (48.8)
0.31
0.54
0.71
0.87
0.79
0.97
Siping
83.3 (54.3)
55.4 (35.2)
48.5 (32.5)
0.10
0.22
0.68
0.86
0.71
0.96
Chifeng
88.0 (68.9)
42.4 (33.1)
32.6 (27.8)
0.17
0.14
0.51
0.75
0.72
0.92
Dandong
66.8 (44.0)
45.6 (24.8)
39.3 (21.3)
0.03
0.11
0.71
0.86
0.64
0.90
Erlianhaote
49.1 (80.2)
22.0 (22.6)
15.9 (14.7)
0.03
0.03
0.51
0.72
0.71
0.61
Yulin
66.6 (67.1)
31.2 (21.0)
22.4 (15.9)
0.06
0.03
0.54
0.72
0.54
0.61
Jinsha
42.0 (38.6)
33.6 (24.1)
30.5 (21.9)
0.01
0.06
0.85
0.90
0.63
0.89
Guilin
57.6 (50.5)
46.5 (30.8)
41.7 (27.1)
0.04
0.15
0.85
0.90
0.70
0.96
Lushan
45.4 (32.7)
37.8 (27.9)
33.2 (26.7)
0.01
0.09
0.85
0.86
0.91
0.95
Changde
45.7 (33.8)
40.3 (29.1)
37.0 (27.5)
0.01
0.12
0.89
0.91
0.93
0.96
Dongtan
31.9 (34.0)
27.4 (25.9)
24.8 (23.8)
0.01
0.06
0.90
0.90
0.92
0.96
Akdala
17.1 (57.6)
9.8 (13.7)
7.7 (6.9)
0.00
0.00
0.67
0.79
0.80
0.53
Shangri-La
6.8 (6.3)
5.2 (5.3)
4.5 (5.0)
0.00
0.00
0.76
0.81
0.94
0.99
a Arithmetic mean value with standard deviation in
parentheses.b The ratio of substandard days (daily averaged PM10 or PM2.5
concentrations that exceed the standard values) to total observation days.
According to the latest air quality standards of China (annual averaged
PM10 and PM2.5 concentrations of 70 and 35 µgm-3), 14
stations reached the PM10 standard, while only 7 stations, mainly rural
and remote stations, reached the PM2.5 standard. The ratio of
substandard (daily averaged PM10 or PM2.5 concentrations that
exceed the standard values) days to total observation days at each station
was calculated using the standard daily averaged PM10 and PM2.5
concentrations of 150 and 75 µgm-3 (Table 2). Substandard days of
PM10 and PM2.5 represented more than 30 and 50 % of the
total period at the three most polluted sites (Xian, Zhengzhou and Gucheng).
The PM2.5 substandard day ratios at five other stations (Chengdu,
Anshan, Shenyang, Benxi and Siping) were also larger than 20 %.
Average PM10, PM2.5 and PM1 concentrations at urban/suburban
stations in this study were 83.6, 56.3 and 48.3 µgm-3,
respectively. Meanwhile, the values were 54.8, 36.3 and 30.8 µgm-3 at rural stations, and 11.9, 7.5 and 6.1 µgm-3 at
remote stations. All values were much higher than results from other
countries. For example, the observed PM concentration in Canada between 1984
to 1993 showed the average PM2.5 concentration was 14.1 and 10.7 µgm-3 at urban and rural stations, respectively
(Brook et al., 1997). The average PM2.5 values from
west to east across the IMPROVE network in 1993 (most stations located in
rural areas) were 3 to 13 µgm-3 (Eldred et al.,
1997). Observations in Switzerland from 1998 to 2001 showed average
PM10 and PM2.5 concentrations at urban/suburban stations of 27.7
and 20.1 µgm-3, respectively (Gehrig and Buchmann, 2003).
In Austria, in 1998, the annual mean mass concentrations of PM10,
PM2.5 and PM1 were around 28, 20 and 16 µgm-3,
respectively, at urban sites, and slightly lower at rural sites
(Gomiscek et al., 2004). The average PM10 and PM2.5
concentrations were 23.9 and 16.3 µgm-3, respectively, for the
period 2008–2009 in the Netherlands (Janssen et al.,
2013). Between October 2008 and April 2011, the 20 study areas of the
European ESCAPE project showed PM10 and PM2.5 with similar spatial
patterns; specifically, low concentrations in northern Europe and high
concentrations in southern and eastern Europe (Eeftens et al., 2012).
With the rapid urbanization and corresponding increase in traffic and energy
consumption in India, the ambient concentrations of fine PM are also high.
For example, measurements in New Delhi during August to December 2007 showed
that concentrations of PM10, PM2.5 and PM1 ranged from 20 to 180 µgm-3
during the monsoon season, and from 100 to 500 µgm-3 during winter (Tiwari et al., 2012).
Scatter plots of PM1 versus PM2.5 (a) with and
(b) without data from the strong sandstorm and dust storm at Akdala.
Spatial distribution of the average ratios of (a) PM2.5 / PM10 and (b) PM1 / PM2.5.
Relationships between PM10, PM2.5 and PM1 concentrations
The squared correlation coefficient (R2) values of the linear fit
between PM10 and PM2.5 and between PM1 and PM2.5 are
given in Table 2. Higher values indicate that the two PM size bins were
closer matched in terms of their sources. At most stations, the R2
values between PM1 and PM2.5 were higher than the values between
PM2.5 and PM10. This is because PM1 and PM2.5 both
belong to fine particle size bins, which are normally emitted from the same
sources. For example, the R2 values were 0.7857 between PM2.5 and
PM10, and 0.9689 between PM1 and PM2.5, at Gucheng.
Correlation analysis is sensitive to outliers, and thus sandstorm events
may have impacted upon the results considerably, due to abnormally high
concentration values. There were four strong dust storm event days at Akdala
in 2012, on 21 and 22 April, and 9 and 20 May, which resulted in the four
outliers shown in Fig. 2a, and the low R2 value of 0.5346 between
PM1 and PM2.5. The value increased to 0.9406 when the four
outliers were removed (Fig. 2b). Similar results were also observed at
Yulin and Erlianhaote around dust storm source regions (Table 2).
The average values of the daily PM2.5 / PM10 and PM1 / PM2.5
ratios are listed in Table 2. The spatial distribution of the average
PM2.5 / PM10 ratios (Fig. 3a) shows lower values in northern
China, influenced by Asian sandstorms and dust storms (Wang et al., 2008; Zhang et al.,
2003). The values were also influenced by fugitive dust due to the low
precipitation amounts in northern China, especially at Lhasa, Erlianhaote,
Yulin and Chifeng, with ratios of less than 0.6. The ratios at the stations
in northeast China were between 0.6 and 0.7, except at Dandong where the
value was 0.71. The values were also low at Zhengzhou and Akdala, at 0.68
and 0.67, respectively. The highest ratio was 0.9 at Dongtan, and the other
stations with ratios higher than 0.8 were Chengdu, Changde, Guilin, Jinsha
and Lushan. The values were between 0.7 and 0.8 at other stations. The
PM1 / PM2.5 ratios (Fig. 3b) showed a similar spatial
distribution, but the values were higher than PM2.5 / PM10. The
lowest ratio of 0.6 was also observed at Lhasa, and the values at most
stations in southern China were greater than or equal to 0.9.
Spatial distribution of the seasonal average concentrations
(µgm-3) of (a) PM10, (b) PM2.5, (c) PM1 and
(d) ratios of PM2.5 / PM10.
Interannual variations of PM2.5 concentrations at the
stations (a) on the HBP and Guanzhong Plain, (b) in northeast China,
(c) along the middle and lower reaches of the Yangtze River and (d) in southern
China.
Interannual variation of (a) PM10 concentration and
(b) PM1 concentration at Zhengzhou, Xian and Gucheng.
Seasonal variation
The seasonal variations of PM10 concentrations (Fig. 4a) show that
winter and spring were the most polluted seasons at all sites except Lushan,
where the highest value was observed in autumn. This result is consistent
with a previous study of PM10 variation across China from 2000 to 2006
(Qu et al., 2010). The higher winter concentrations
were caused by higher emissions during the cold season from heating, and
more stagnant weather conditions with a lower planetary boundary layer. The
opposite conditions and more precipitation due to the summer monsoon
resulted in the lowest PM10 concentration values in summer. Spring is
the dust storm season in east Asia (Qian et al., 2004; Wang et al., 2008;
Zhou and Zhang, 2003), which leads to high PM10 concentrations in dust
source regions and downwind areas in northern China. For example, the
PM10 concentrations in spring were much higher than other seasons at
the dust source sites of Yulin and Erlianhaote.
For PM2.5, winter was still the most polluted season at most sites,
while the contribution of spring decreased substantially in northern China
(Fig. 4b). This trend can be further observed from the PM1
distribution (Fig. 6c); hence, the average PM1 concentration in
spring was lowest at Yulin, Xian, Zhengzhou, Gucheng and Benxi. The seasonal
variation patterns were very similar for PM10, PM2.5 and PM1
at the sites in southern China.
A spatial distribution map of the seasonal average PM2.5 / PM10
ratios is given in Fig. 6d. For the reasons given above, lower
PM2.5 / PM10 ratios were observed in spring at the northern China
sites, while the seasonal variation was not significant at the southern
China sites.
Interannual variation
The interannual variation of PM2.5 at various stations is presented in
Fig. 5. Significant decreasing trends were observed at the HBP stations of
Zhengzhou and Gucheng (Fig. 5a). The annual averaged PM2.5
concentration decreased from 123.4 to 65.2 µgm-3 at Zhengzhou,
and from 101.0 to 69.1 µgm-3 at Gucheng, during 2006–2014. At
Zhengzhou, the lowest value of 63.7 µgm-3 occurred in 2012, and
this level was maintained in subsequent years; however, at Gucheng, the
value increased suddenly in 2012 to 95.1 µgm-3 and then declined
rapidly during 2013 and 2014. At Xian, the annual averaged PM2.5
concentration decreased from 2006 to 2009, increased until 2011, and then
decreased again until 2014 (Fig. 5a).
For the stations in northeast China, a significant increasing trend of the
PM2.5 concentration was observed at Shenyang and Benxi from 2006 to
2013, followed by a decrease in 2014 (Fig. 5b). The peak value at Shenyang
was especially high in 2013 at 123.1 µgm-3, while the values were
less than 60 µgm-3 in the other years. The highest values were
observed in 2009 at Anshan and Dandong, but the lowest values were in 2014
at Anshan and 2010 at Dandong. A general decreasing trend was observed at
Siping, with a few fluctuations. At Fushun, the value decreased from 2006 to
2011 and then increased to 2013, followed by a slight decrease in 2014.
For the stations along the middle and lower reaches of the Yangtze River, a
common trend was a clearly lower PM2.5 value in 2014 than in 2013, but
the general variation trend was not significant (Fig. 5c). A peak value of
33.7 µgm-3 was observed in 2012 at Dongtan, followed by a
decrease to 24.12 µgm-3 over the subsequent 2 years. At Jinsha
and Changde, the highest value was in 2013, while it was in 2009 at Lushan.
For the stations in southern China, a general decreasing trend was observed,
with obvious fluctuations (Fig. 5d). Panyu is a typical station in the
center of the Pearl River Delta economic area of China. The PM2.5 value
decreased from 64.6 µgm-3 in 2006 to 41.6 µgm-3 in
2014, and the lowest value was 36.4 µgm-3 in 2010. A similar
trend was observed in Guilin, with a stronger fluctuation from 2010 to 2012.
At Nanning, a peak value occurred in 2010 and the lowest value of 28.5 µgm-3 was observed in 2012.
Generally, the PM10 and PM1 interannual variation trends were
similar to that of PM2.5 at most stations. For example, a similar trend
and fluctuations were observed at the stations presented in Figs. 8 and
7a. A difference in the trend was observed at Zhengzhou from 2013 to
2014, with a significant increasing trend of PM10 and decreasing trend
of PM1.
Diurnal variation of PM2.5 concentrations at the stations
(a) on the HBP and Guanzhong Plain, (b) in northeast China, (c) along the middle
and lower reaches of the Yangtze River and (d) in southern China.
Anthropogenic emission distributions at a resolution of
0.1∘×0.1∘, based on HTAP_v2
data set: (a) BC; (b) PM2.5; (c) SO2; (d) NOx (units: kgm-2s-2).
Diurnal variation
The average diurnal variation of PM2.5 at various stations is presented
in Fig. 7. Pronounced diurnal variation of PM2.5 was observed at most
urban sites, with an obvious morning peak at around 07:00 to 08:00 (Beijing Time) and an
afternoon valley between 14:00 and 16:00. At some stations, an evening
peak could be recognized at around 19:00 to 21:00 (Siping, Benxi, Fushun,
Anshan, Guilin and Panyu) or midnight (Gucheng, Xian). This bimodal pattern
was also observed in Beijing (Zhao et al., 2009). A unimodal
pattern, without an evening peak, could be identified at some other stations
(Zhengzhou, Shenyang and Nanning). In urban areas, the morning and evening
peaks are contributed to by enhanced anthropogenic activity during rush
hour, and the afternoon valley is mainly due to a higher atmospheric mixing
layer, which is beneficial for air pollution diffusion. Panyu station is on
top of a 140 m hill at the edge of Guangzhou, so aged and mixing
aerosols were observed with a weak urban diurnal variation pattern. Similar
to Panyu station, the rural stations along the middle and lower reaches of
the Yangtze River showed no typical urban diurnal variation pattern (Fig. 7c). The diurnal variation in PM1 and PM10 concentrations was
similar to that of PM2.5 at most stations.
Emissions differences between 2010 and 2008 at a resolution of
0.1∘×0.1∘, based on HTAP_v2
data set: (a) BC; (b) PM2.5; (c) SO2; (d) NOx (units: kgm-2s-2).
Daily averaged PM2.5 concentrations vs wind speed and
relative humidity at (a, b) Zhengzhou, (c, d) Shenyang and (e, f) Nanning in
January 2013.
Interannual variation of PM10, PM2.5 and PM1 vs. wind speed and relative humidity at (a, b) Zhengzhou and (c, d) Nanning.
Emission and meteorological influences
PM loadings are controlled by both emissions and meteorological conditions.
Even mineral dust emissions from deserts and volatile organic compound (VOC)
emissions from vegetation are controlled by meteorological factors, e.g.,
wind speed and temperature. The major source of air pollution in China is
anthropogenic emissions, especially with the rapid economic development that
has taken place in recent years. As such, the average PM concentration
pattern is determined largely by emissions, but meteorological factors also
play an important role by affecting pollutant diffusion and deposition.
The distributions of the anthropogenic emissions of black carbon (BC),
PM2.5, SO2 and NO2 in 2010, based on the HTAP_v2 harmonized emissions database
(Janssens-Maenhout et al., 2015), are
presented in Fig. 8. The emissions data for the east Asia domain were
supplied by the MICS-Asia project. The spatial distributions of species show
a consistent pattern with the high emissions regions of the HBP, Guanzhong
Plain, Sichuan Basin, middle and lower reaches of the Yangtze River, Pearl
River Delta region and the industrial region of northeast China, which is
generally similar to the PM loadings pattern for China (Fig. 1). For
example, most stations subjected to PM pollution are located in the highest
emissions region of the HBP. This indicates that average PM loadings are
controlled by the quantity of anthropogenic emissions in central-eastern
China.
The trends in emissions for China during 2005–2010
(S. X. Wang et al., 2014) show that emissions of
SO2 and PM2.5 in east Asia decreased by 15 and 12 %,
respectively, while emissions of NOx and non-methane VOCs increased by
25 and 15 %, respectively. Driven by changes in emissions, PM2.5
concentrations decreased by 2–17 µgm-3 over most of the North
China Plain, the Yangtze River Delta and the Pearl River Delta
(Zhao et al., 2013). This could explain the general decreasing
trend with respect to PM during 2006–2010 at most stations (Fig. 5). The
spatial distributions of emissions differences between 2010 and 2008 for BC,
PM2.5, SO2 and NO2 are plotted in Fig. 9, based on the
HTAP_v2 emission data set. BC emissions decreased from 2008 to
2010 in most regions of east China, except the provinces of Hebei, Shanxi,
Hubei, Jiangxi and Inner Mongolia (Fig. 9a). More areas of China showed a
reduction in PM2.5 emissions, except Shanxi and Hubei (Fig. 9b). The difference in SO2 emissions (Fig. 9c) showed a similar pattern
to that of BC but with an increasing trend apparent in northeast China.
NOx
emissions increased in most regions of central-eastern China, except in the
provinces of Guangdong, Zhejiang and Taiwan (Fig. 9d). This trend was
driven by the rapid growth of industry and transportation, combined with
inadequate control strategies (S. X. Wang et al.,
2014).
Although there are no published emissions data after 2010, it is believed
that emissions have to a certain extent been controlled well since the end
of 2013, with the arrival of China's “Action Plan for the Control of Air
Pollution” document. This could explain the general decreasing trend for
the year 2014 at most stations (Fig. 5).
Central-eastern China experienced severe haze events in January 2013, with a
regionally stable planetary boundary layer and low mixing height (H. Wang
et al., 2014). The daily averaged PM2.5 concentrations and
meteorological factors of wind speed and relative humidity for this period
at Zhengzhou, Shenyang and Nanning are plotted in Fig. 10. Zhengzhou is
located in this haze region, and experienced very high PM2.5
concentrations, especially from 6 to 15 January. The wind speed variation was
negatively related with PM2.5 concentrations. The rapid increase in
PM2.5 concentrations from 1 to 6 January corresponded with the rapid
decrease in wind speed during the same period. Also, the strong wind speed
on 24 January resulted in low PM2.5 concentration. Shenyang and
Nanning are not located in this severe haze region, but still suffered some
fine-PM days that month. A negative correlation between PM2.5 and wind
speed was also observed at Shenyang and Nanning. In general, relative
humidity (RH) was positively related with the PM2.5 concentration if no
precipitation occurred. Otherwise, high RH with precipitation corresponded
to low PM concentrations due to wet deposition.
In terms of interannual variation, the negative correlation between
PM2.5 concentrations and wind speed, and the positive correlation
between PM2.5 concentrations and relative humidity, could not be well
identified (Fig. 11). Although a generally similar variation trend for the
PM10 concentration and relative humidity was observed at Zhengzhou,
this was not found at other stations. The PM2.5 concentration in 2014
was lower than in 2013, but the relative humidity was much higher and the
wind speed much lower. The interannual variation of PM concentrations could
not be explained solely by meteorological factors, although a recent model
simulation for the period 2004–2012 with anthropogenic emissions fixed at
the values for the year 2006 indicated that variations in meteorological
fields dominated the interannual variation in aerosols in China
(Mu and Liao, 2014). Long term, both emissions and meteorological
factors play important roles; while in the short term, meteorological
factors play a leading role – at least in the absence of significant changes in emissions.
Conclusion
Spatial and temporal trends in PM pollution were examined using PM10,
PM2.5 and PM1 concentration data at 24 stations from 2006 to 2014.
Relatively high PM concentrations were observed at most stations. There were
14 stations that reached the PM10 annual air quality standard, but only
7 stations, mostly rural and remote stations, reached the PM2.5 annual
air quality standard of China. The highest PM concentrations were observed
at the stations on the HBP and Guanzhong Plain. In addition, the percentage
value of substandard days of PM2.5 was greater than 50 %, indicating
very serious air pollution in these regions. PM pollutants are also a
serious problem in the industrial regions of northeast China and the Sichuan
Basin. The PM concentrations were relatively lower in southern areas of
China, but the averaged PM2.5 concentration was still higher than the
national standard.
Given they are both fine particles, PM1 and PM2.5 were more
closely correlated than PM2.5 and PM10. The correlations were
sensitive to the effect of outlier data at those stations heavily impacted
by dust storm events. More dust aerosol was observed in northern China, and
thus the PM2.5 / PM10 ratios increased from less than 0.6 to around
0.9 when moving from north to south China.
Pronounced seasonal variations were observed at most stations, with the
highest concentrations in winter and lowest concentrations in summer.
PM10 concentrations were also high in spring, due to the contribution
of dust storm events, especially at those stations near to dust source
regions. For PM2.5 and PM1, spring was a relatively low
concentration season, especially at the stations in northern China. Also,
low PM2.5 / PM10 ratios were observed in spring in northern China.
An interannual decreasing trend was observed in the HBP and southern China
from 2006 to 2014, but an increasing trend occurred at some stations in
northeast China, and no significant trend could be found over the middle and
lower reaches of the Yangtze River. Annual-averaged PM concentrations were
lower in 2014 than 2013 at most stations, which may indicate an improvement
in air quality following the “Action Plan for the Control of Air
Pollution” document issued by the Chinese government in September 2013.
Bimodal and unimodal diurnal variation patterns were identified at urban
stations. A negative correlation between PM concentrations and wind speed
was found for the short term, but variations in emissions must be
considered for long-term trend analyses, especially in rapidly developing
countries.
This network-based observation data set provides the longest continuous
record of fine particle concentrations in China, but it features a limited
number of stations and an uneven spatial distribution. Importantly, there is
no representative city site in the Yangtze River Delta region, which is an
important haze area in China. The emissions sources and meteorological
factors influencing PM spatial and temporal patterns in China still require
further study.