Subscriber access provided by Washington University | Libraries
Characterization of Natural and Affected Environments
Assessing effect of the long-term variations in aerosol characteristics on satellite remote sensing of PM2.5 using an observation-based model Changqing Lin, Alexis K. H. Lau, Jimmy C.H. Fung, Xiang Qian Lao, Ying Li, and Chengcai Li Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b06358 • Publication Date (Web): 28 Feb 2019 Downloaded from http://pubs.acs.org on March 2, 2019
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 29
Environmental Science & Technology
1
Assessing effect of the long-term variations in aerosol
2
characteristics on satellite remote sensing of PM2.5 using an
3
observation-based model
4
Changqing Lin,†,‡,* Alexis K.H. Lau,†,‡,* Jimmy C.H. Fung,‡,§ Xiang Qian Lao,∥ Ying Li,⊥
5
Chengcai Li#
6
†Department of Civil and Environmental Engineering, the Hong Kong University of Science and
7
Technology, Hong Kong, China
8
‡Division of Environment and Sustainability, the Hong Kong University of Science and
9
Technology, Hong Kong, China
10
§Department of Mathematics, the Hong Kong University of Science and Technology, Hong Kong,
11
China
12
∥Jockey Club School of Public Health and Primary Care, the Chinese University of Hong Kong,
13
Hong Kong, China
14
⊥Department of Ocean Science and Engineering, Southern University of Science and Technology,
15
Shenzhen, China
16
#Department of Atmospheric and Oceanic Sciences, School of Physics, Peking University,
17
Beijing, China
18
AUTHOR INFORMATION
19
Corresponding Author:
20
*(C.Q. Lin) Tel.: (852) 23586676. E-mail:
[email protected].
21
*(A.K.H. Lau) Tel.: (852) 23586944. E-mail:
[email protected].
22 1 ACS Paragon Plus Environment
Environmental Science & Technology
23
Abstract: Variations in aerosol characteristics play an essential role in satellite remote sensing
24
of PM2.5 concentrations. The lack of measurement of aerosol characteristics, however, limits the
25
assessment of their effects. This study presented an observation-based model that directly
26
considered the effects of aerosol characteristics. In this model, we used an integrated humidity
27
coefficient (γ’) and an integrated reference value (K) to delineate the effects of aerosol
28
characteristics. We then investigated the effects of the long-term variations in aerosol
29
characteristics on satellite remote sensing of PM2.5 concentration in Hong Kong from 2004 to
30
2012. The results show that the γ’ value peaked in 2009 because the percentages of highly
31
hygroscopic components (e.g., sulfate and nitrate) in aerosols reached their peaks. The K value
32
increased from 2004 to 2011 because of the increasing percentages of strong light-extinction
33
components (e.g., organic matter) and the decreasing fine mode fraction in aerosols. The
34
accuracy of PM2.5 retrieval improved greatly after accounting for the long-term variations in
35
aerosol characteristics (e.g., correlation coefficient increased from 0.56 to 0.80). The results
36
underscore the need to incorporate the variations in aerosol characteristics in the PM2.5
37
estimation models.
38 39 40 41 42 43 44 45
2 ACS Paragon Plus Environment
Page 2 of 29
Page 3 of 29
46
Environmental Science & Technology
TOC Art
47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 3 ACS Paragon Plus Environment
Environmental Science & Technology
65
Introduction
66
Traditional studies have relied on ground monitoring to identify the levels of PM2.5
67
concentration (particulate matter with an aerodynamic diameter below 2.5 μm).1 Spatial coverage
68
of such monitoring, however, is inherently limited by the resource. Satellite remote sensing
69
techniques provide an alternative method to delineate PM2.5 distribution with a large spatial
70
coverage.2–7 These techniques retrieve the ground-level PM2.5 concentration from aerosol optical
71
depth (AOD), which denotes a vertical integration of light extinction from atmospheric aerosols.
72
Aerosol characteristics play an essential role in the AOD-PM2.5 relationship through their
73
impacts on aerosol extinction efficiency and hygroscopic growth ability.8,9 Typical components
74
of particulate matter such as sulfate, nitrate, elemental carbon (EC), organic matter (OM), and
75
dust show distinctive light-extinction abilities.10,11 Light-extinction efficiency for OM is typically
76
higher than for sulfate and nitrate.12,13 Coarse particles, such as dust, scatter light less effectively
77
than fine particles.10 In addition, sulfate and nitrate are considered to be more hygroscopic than
78
OM and dust aerosols.14 Therefore, the effects of variations in aerosol characteristics should be
79
taken into account when developing a robust AOD-PM2.5 model.
80
During the past decade, various simulation-based and observation-based AOD-PM2.5 models
81
have been adopted to identify ground-level PM2.5 concentrations around the world.15–21 In the
82
simulation-based AOD-PM2.5 models, information on the aerosol characteristics are drawn from
83
Chemical Transport Models (CTMs).15–17 Observation-based AOD-PM2.5 models train various
84
empirical relationships between the AOD and PM2.5 concentration using ground PM2.5
85
measurements.18–21 Most of these empirical models adopt various meteorological and land-use
86
parameters as covariates to indicate the variation in aerosol characteristics.9 It is important to
87
develop observation-based model that directly addresses the effect of aerosol characteristics.
4 ACS Paragon Plus Environment
Page 4 of 29
Page 5 of 29
Environmental Science & Technology
88
Governments around the world have implemented various control measures to reduce the
89
levels of PM2.5 concentrations.22,23 These control efforts have reduced anthropogenic emissions
90
of primary particles and precursor gases such as sulfur dioxide and nitrogen oxides. As a result,
91
aerosol characteristics have changed considerably over a long period.24 In addition, different
92
seasons feature distinctive emission sources, meteorological conditions, and evolution processes
93
of pollutions.25 Therefore, aerosols tend to show distinctive characteristics among the four
94
seasons.26 These long-term variations in aerosol characteristics can pose a significant impact on
95
the AOD-PM2.5 relationship. The lack of measurement of aerosol characteristics, however, limits
96
the assessment of their effects.
97
Hong Kong locates in southeast of the Pearl River Delta (PRD) region of China. It is one of
98
the most developed and densely populated cities in the world. Compared with most cities in
99
mainland China, Hong Kong operates a much more comprehensive system for monitoring air
100
quality over a long period.1,27 In this study, we took advantage of long-term measurements of
101
mass concentrations and chemical compositions of particulate matter over Hong Kong from 2004
102
to 2012. Inter-annual and seasonal variations in aerosol characteristics were studied. We
103
presented an observation-based AOD-PM2.5 model that directly considered the effects of aerosol
104
characteristics. The effects of the long-term variations in aerosol characteristics on satellite
105
remote sensing of PM2.5 concentration were then investigated over Hong Kong during the study
106
period.
107
Data
108
From Moderate Resolution Imaging Spectroradiometer (MODIS) observations, we retrieved
109
AOD at a resolution of 1 km over Hong Kong from 2004 to 2012.28 The AOD dataset was built
110
using the dark-target land algorithm at 0.55 μm and our own look-up table. Verifications of the
5 ACS Paragon Plus Environment
Environmental Science & Technology
111
satellite-retrieved AOD showed a good agreement with ground observations.28,29 We obtained
112
hourly surface meteorological data at Tung Chung station (113.94°E 22.29°N, as shown in
113
Figure S1) from 2004 to 2012. The meteorological data included visibility-derived aerosol
114
extinction coefficient (σa,0) and relative humidity (RH).
115
We also obtained hourly PM10 (particulate matter with an aerodynamic diameter of below 10
116
μm) and PM2.5 data from air quality monitoring network over Hong Kong during the study
117
period. The air quality monitoring network included eleven general stations. Temporal coverage
118
of the PM2.5 concentration data was different at each station. The PM2.5 data with a nine-year
119
coverage were available at four stations, including Tung Chung [the other three stations were
120
Yuen Long (114.02°E 22.45°N), Tsuen Wan (114.11°E 22.37°N), and Tap Mun (114.36°E
121
22.47°N)]. PM2.5 concentrations were available at the other stations since some time in 2010.
122
The monthly averages of PM2.5 concentrations were estimated when >80% of the hourly PM2.5
123
data within that month were available. The annual averages of PM2.5 concentrations were
124
estimated when >80% of the hourly PM2.5 data within that year were available.
125
In addition, we obtained daily averages of PM10 chemical speciation data at an interval of six
126
days from the PM10 chemical speciation network over Hong Kong from 2004 to 2012. The PM10
127
chemical speciation data include collections of mass concentrations of nitrate, sulfate,
128
ammonium, EC, and OM in PM10 every six days. We normalized these mass concentrations by
129
the PM10 concentration and obtained the percentages of different chemical compositions in PM10.
130
Simultaneous observations of meteorological values, mass concentrations and chemical
131
compositions of particulate matter were available at Tung Chung station. These observations
132
were used to develop the AOD-PM2.5 model. The results of PM2.5 retrieval were evaluated using
6 ACS Paragon Plus Environment
Page 6 of 29
Page 7 of 29
Environmental Science & Technology
133
the observations in other stations. More information on these datasets can be found in the
134
supplementary material.
135
Methodology
136
The relationship between AOD and ground-level PM2.5 concentration is affected by aerosol
137
vertical structure, hygroscopic growth factor, aerosol extinction efficiency, and fine mode
138
fraction.8,17,20 Based on this physical understanding, Lin et al. (2015)8 developed an
139
observational-data driven algorithm to retrieve PM2.5 concentration from the AOD. We
140
summarize and extend this algorithm here.
141
To address the effect of the aerosol vertical structure, we introduced aerosol scale height (H).
142
The scale height H at Tung Chung station can be estimated from the ratio of the satellite-based
143
AOD over the station and the visibility-derived σa,0. By assuming a uniform H within Hong
144
Kong (an area of approximately 30 × 40 km2), spatial distribution of surface σa,0 over Hong
145
Kong can be derived by:
146 147 148
𝜎𝑎,0 =
𝐴𝑂𝐷
(1)
𝐻
Relationship between σa,0 and PM10 concentration can be expressed as: 𝑃𝑀10 =
𝜎𝑎,0 𝛼𝑒𝑥𝑡,10 ∙
(
(2)
1 ― 𝑅𝐻 ―𝛾 1 ― 𝑅𝐻0
)
149
where γ represents hygroscopic growth coefficient, which is dependent on aerosol characteristics;
150
RH0 = 40% is reference RH in dry conditions; 𝑓(𝑅𝐻) =
151
factor; and αext,10 represents aerosol mass extinction efficiency (MEE).
(
1 ― 𝑅𝐻
)
1 ― 𝑅𝐻0
―𝛾
is hygroscopic growth
152
Because aerosol characteristics change under different meteorological conditions, the MEE is
153
likely to be dependent on RH. We therefore merge the humidity dependence from the MEE with
7 ACS Paragon Plus Environment
Environmental Science & Technology
154
hygroscopic growth. The relationship between σa,0 and the PM10 concentration can then be
155
expressed as:
156
𝑃𝑀10 =
Page 8 of 29
𝜎𝑎,0
(
𝛼′𝑒𝑥𝑡,10 ∙
(3)
1 ― 𝑅𝐻 ― 𝛾0 1 ― 𝑅𝐻0
)
157
where γ0 is defined as the integrated humidity coefficient from the MEE and hygroscopic growth
158
in this study; α’ext,10 is the MEE under dry conditions. Both the γ0 and α’ext,10 values are associated
159
with the aerosol characteristics. The γ0 and α’ext,10 values at Tung Chung can be fitted using
160
simultaneous measurements of the visibility-derived σa,0, RH, and PM10 concentration.
161
To estimate PM2.5 concentration, we need to address the effect of fine mode fraction (FMF,
162
percentage of PM2.5 mass in PM10 mass). Relationship between σa,0 and PM2.5 concentration can
163
be expressed as:
164
165 166
𝜎𝑎,0
𝑃𝑀2.5 = 𝛼
𝑒𝑥𝑡,10 𝐹
∙
(
)
1 ― 𝑅𝐻 1 ― 𝑅𝐻0
(4)
―𝛾
where F represents the FMF, which is an indicator of aerosol size distribution. Because the FMF is likely to be dependent on RH, we merge the humidity dependence from
167
the MEE and FMF with hygroscopic growth. The relationship between σa,0 and PM2.5
168
concentration can then be expressed as:
169
𝑃𝑀2.5 = 𝛼′
𝜎𝑎,0
𝑒𝑥𝑡,10 𝐹′
∙
(
)
1 ― 𝑅𝐻 1 ― 𝑅𝐻0
= ― 𝛾′
𝜎𝑎,0 𝐾∙
(
1 ― 𝑅𝐻 ― 𝛾′ 1 ― 𝑅𝐻0
)
(5)
170
where γ’ is integrated humidity coefficient from the MEE, FMF, and hygroscopic growth; F’ is
171
the FMF under dry conditions; and K = α’ext,10/F’ is integrated reference value under dry
172
conditions. The γ’ and K values at Tung Chung can be fitted using simultaneous measurements of
173
the visibility-derived σa,0, RH, and PM2.5 concentration.
8 ACS Paragon Plus Environment
Page 9 of 29
174
Environmental Science & Technology
By introducing the γ0 value, we can separate the humidity dependence from the MEE and
175
hygroscopic growth with that from the FMF. Therefore, the γ’ value is associated with γ0 in
176
conjunction with the humidity dependence from the FMF. In this study, we estimate the γ0, γ’,
177
α’ext,10, F’, and K values on a monthly basis at Tung Chung. We then investigate the associations
178
between the long-term variations in the γ’/K values and aerosol characteristics. By assuming that
179
the parameters (e.g., γ’, K, and RH) derived at Tung Chung represent the conditions over all of
180
Hong Kong, spatial distribution of PM2.5 concentration can be derived by:
181
𝐴𝑂𝐷 𝐻
𝑃𝑀2.5 = 𝐾∙
(
1 ― 𝑅𝐻 ― 𝛾′ 1 ― 𝑅𝐻0
)
(6)
182
To assess the effect of the long-term variations in aerosol characteristics, we retrieve PM2.5
183
concentrations using two AOD-PM2.5 models. First, we retrieve PM2.5 concentration using a static
184
AOD-PM2.5 model, which assumes that the γ’ and K values remained unchanged during the study
185
period. Second, we retrieve PM2.5 concentration using a dynamic AOD-PM2.5 model, which takes
186
into account the long-term variations of γ’ and K. We then compare the performance of these two
187
models.
188
Results
189
Variation in PM10 chemical compositions. Figure 1 shows long-term variations in the
190
percentages of different chemical compositions in PM10 at Tung Chung from 2004 to 2012.
191
Nine-year averages of the percentages of nitrate, sulfate, ammonium, OM, and EC in PM10 were
192
1.4%, 21.8%, 4.4%, 18.9%, and 5.2%, respectively. Sulfate and OM showed the largest
193
contributions to PM10 concentration. Remaining components in the PM10 mass were associated
194
with other species, such as crustal material, sea salt, non-crustal trace elements, and unidentified
195
species.30 It is noted that annual averages of the percentages of nitrate and sulfate reached their
9 ACS Paragon Plus Environment
Environmental Science & Technology
196
peaks in 2009 with maximums of 1.9% and 23.0%, respectively. In addition, annual averages of
197
the percentages of ammonium and OM showed an ascending trend from 2004 to 2011 with
198
maximal levels of 5.4% and 21.7%, respectively. Annual average of the percentage of EC
199
showed a general descending trend.
200
(a)
201
(b)
202
(c)
203
(d)
204
(e) 10 ACS Paragon Plus Environment
Page 10 of 29
Page 11 of 29
Environmental Science & Technology
205
Figure 1. Monthly (solid lines) and yearly (dashed lines) variations in the percentages of
206
different chemical components [including (a) nitrate, (b) sulfate, (c) ammonium, (d) OM, and (e)
207
EC] in PM10 at Tung Chung from 2004 to 2012.
208
The monthly variations in the PM10 chemical components showed significant cyclic
209
variations. Figure S2 shows seasonal averages of the percentages of different chemical
210
components in PM10 at Tung Chung throughout the study period. In general, nitrate, ammonium,
211
and OM showed the highest percentages in PM10 (1.9%, 5.6%, and 21.8%, respectively) during
212
winter. Continental outflows pass through vast polluted regions in mainland China and then
213
boost the percentages of these secondary pollutants in Hong Kong during winter.31–33
214
Temperature is considered another factor that affects the concentration of ammonium nitrate.
215
Lower temperatures during the winter favor the formation of particulate ammonium nitrate from
216
its precursor gases.34 The highest percentages of sulfate were seen during spring and autumn
217
(both were about 23.4%), and the lowest percentage was seen during summer (19.5%). The high
218
percentages of sulfate during spring and autumn resulted from a regional accumulation of
219
secondary sulfate around the coastal regions in southeastern China.31,35 The transport of marine
220
vessels emissions from south contributed to the high percentage of EC during summer.33
221
Variation in γ0 and α’ext,10. Monthly γ0 and α’ext,10 values were fitted using simultaneous
222
measurements of the visibility-derived σa,0, RH, and PM10 concentration at Tung Chung. Figure 2
223
shows long-term variations in (a) γ0 and (b) α’ext,10 at Tung Chung from 2004 to 2012. The nine-
224
year averages of γ0 and α’ext,10 were estimated to be 0.89 and 3.34 m2/g, respectively. The γ0 value
225
reached its peak in 2009 with a maximal level of 1.06. This long-term pattern is consistent with
226
the percentages of highly hygroscopic components (e.g., sulfate and nitrate) in PM10. The α’ext,10
227
value experienced an ascending trend from 2004 to 2011 with a maximal level of 4.06 m2/g. This
11 ACS Paragon Plus Environment
Environmental Science & Technology
228
long-term pattern is consistent with the percentages of strong light-extinction components (e.g.,
229
OM and ammonium) in PM10.
230
(a)
231
(b)
232
Figure 2. Monthly (solid lines) and yearly (dashed lines) variations in (a) γ0 and (b) α’ext,10 at
233
Tung Chung from 2004 to 2012.
234
Figure S3 shows seasonal averages of the γ0 and α’ext,10 values at Tung Chung throughout the
235
study period. The γ0 value showed the highest levels in spring and winter (0.95 and 0.96,
236
respectively), when the highly hygroscopic components (e.g., sulfate, nitrate, and ammonium)
237
were abundant in PM10. In contrast, the γ0 value reached its lowest level of 0.78 in summer. The
238
α’ext,10 value reached its highest level of 3.81 m2/g in winter, which is associated with an
239
abundance of the strong light-extinction components (e.g., OM).
240
Variation in FMF. Hourly FMF at Tung Chung can be derived using simultaneous
241
observations of PM2.5 and PM10 concentrations. We then investigate the characteristics of the
242
variations in the FMF. Figure 3(a) shows long-term variation in correlation coefficient between
243
the FMF and RH at Tung Chung from 2004 to 2012. The correlation coefficients were positive in
244
most months and were as high as 0.6. These correlations were statistically significant at a 99% 12 ACS Paragon Plus Environment
Page 12 of 29
Page 13 of 29
Environmental Science & Technology
245
confidence level for most months (79 out of 108). These phenomena (i.e., positive humidity
246
dependence of the FMF, especially in winter) were similar to those observed in northern China,
247
showing that dry weathers were associated with more coarse particles.8
248
(a)
249
(b)
250
Figure 3. (a) Monthly (solid lines) and yearly (dashed lines) variations in correlation coefficient
251
between the FMF and RH at Tung Chung from 2004 to 2012. (b) Monthly (solid lines) and
252
yearly (dashed lines) variations in the average dry FMF (RH < 60%) at Tung Chung from 2004
253
to 2012.
254
Because of the moist climate, typical RH is high in Hong Kong. We characterized the dry
255
FMF by averaging the FMF under a condition of RH < 60%. Figure 3(b) shows long-term
256
variation in the average dry FMF (RH < 60%) at Tung Chung from 2004 to 2012. Annual
257
average of the dry FMF decreased from 0.74 to 0.60 during the study period. This phenomenon
258
(i.e., descending FMF) was observed at all stations over Hong Kong (as shown in Figure S4).
259
The descending FMF value corresponds to restrictions on vehicle exhaust and coal/biomass
260
burning in Hong Kong.36
13 ACS Paragon Plus Environment
Environmental Science & Technology
261
Figure S5 shows seasonal averages of the correlation coefficient between FMF and RH (blue
262
bars) and the dry FMF (red bars) at Tung Chung throughout the study period. The humidity
263
dependence of the FMF was more significant in spring and winter, with the average correlation
264
coefficients of 0.27 and 0.31, respectively. In contrast, the lowest humidity dependence of the
265
FMF was observed in summer with a weak correlation coefficient of -0.04. The dry FMF values
266
in autumn and winter (0.69 and 0.68, respectively) were higher than in spring and summer (both
267
were about 0.64). More fine particles in cold seasons suggest the effect of the long-range
268
transports of secondary aerosols from inland.
269
Variation in γ’ and K. Monthly γ’ and K values were fitted using simultaneous measurements
270
of the visibility-derived σa,0, RH, and PM2.5 concentration at Tung Chung. Figure 4 shows long-
271
term variations in the γ’ and K values at Tung Chung from 2004 to 2012. Because of the positive
272
humidity dependence of FMF in most months, the γ’ values were generally lower than γ0. Nine-
273
year average of γ’ was estimated to be 0.82 with a peak of 0.93 in 2009. The K value steadily
274
increased from 4.18 m2/g in 2004 to 6.17 m2/g in 2011. The increasing K was determined by the
275
combined effect of the increase in the dry MEE (α’ext,10) and the decrease in the dry FMF (F’).
276
(a)
277
(b) 14 ACS Paragon Plus Environment
Page 14 of 29
Page 15 of 29
Environmental Science & Technology
278
Figure 4. Monthly (solid lines) and yearly (dashed lines) variations in the (a) γ’ and (b) K values
279
at Tung Chung from 2004 to 2012.
280
Figure 5 shows seasonal averages of the γ’ and K values at Tung Chung throughout the study
281
period. Similar to the seasonal pattern of γ0, the γ’ values were highest in spring and winter (0.83
282
and 0.85 respectively). In contrast, the γ’ value’s lowest level of 0.79 was seen in summer. It is
283
noted that the γ’ value was slightly higher than the γ0 value in summer because of its weak-
284
negative humidity dependence on the FMF. Similar to the seasonal pattern of the dry MEE, the K
285
value’s highest level of 5.77 m2/g occurred in winter.
286 287
Figure 5. Seasonal averages of the γ’ (blue bars) and K (red bars) values at Tung Chung
288
throughout the study period.
289
Effect of variation in aerosol characteristics. To assess the effect of the long-term
290
variations in aerosol characteristics, we first retrieve PM2.5 concentration using the static AOD-
291
PM2.5 model, which assumes that the γ’ and K values remained unchanged (represented by the
292
nine-year averages) during the study period. Second, we retrieve PM2.5 concentration using the
293
dynamic AOD-PM2.5 model, which takes into account the monthly variations of γ’ and K. Figure
294
S6 shows an evaluation of hourly PM2.5 concentrations derived from the (a) static and (b) 15 ACS Paragon Plus Environment
Environmental Science & Technology
295
dynamic AOD-PM2.5 models against the ground observations at all stations during the study
296
period. The statistical metrics improved after consideration of the long-term variations of γ’ and
297
K. Correlation coefficient increased from 0.64 to 0.68 (N = 8444). Root mean square error
298
(RMSE) decreased from 23.6 µg/m3 to 22.0 µg/m3. The evaluations were also performed after
299
the training datasets at Tung Chung station were removed. The results show that correlation
300
coefficient increased from 0.60 to 0.63 (N = 7000), and RMSE decreased from 24.7 µg/m3 to
301
23.1 µg/m3.
302
Figure S7 shows an evaluation of monthly averages of the PM2.5 concentrations derived from
303
the (a) static and (b) dynamic AOD-PM2.5 models against the ground observations at all stations
304
during the study period. All statistical metrics improved after consideration of the long-term
305
variations of γ’ and K. Correlation coefficient increased from 0.78 to 0.82 (N = 636). RMSE
306
reduced from 11.2 µg/m3 to 9.7 µg/m3, and mean absolute percentage error reduced from 26.4%
307
to 22.9%. The evaluations were also performed after the training datasets at Tung Chung station
308
were removed. The results show that correlation coefficient increased from 0.77 to 0.80 (N =
309
528). RMSE reduced from 11.5 µg/m3 to 10.1 µg/m3, and mean absolute percentage error
310
reduced from 26.7% to 23.7%.
311
Because the nine-year ground observations of PM2.5 concentrations were available at four
312
specific stations (e.g., Tung Chung, Yuen Long, Tsuen Wan, and Tap Mun), we further evaluate
313
time series of the satellite-derived PM2.5 concentrations at these stations. The solid red lines in
314
Figure S8 represent monthly variations of the average satellite-derived PM2.5 concentrations at
315
the four stations from 2004 to 2012 using the (a) static and (b) dynamic AOD-PM2.5 models. The
316
dashed blue lines represent the corresponding ground observations. Although the static model
317
captured the monthly variation of PM2.5 concentration to a certain extent, correlation coefficient
16 ACS Paragon Plus Environment
Page 16 of 29
Page 17 of 29
Environmental Science & Technology
318
further increased from 0.85 to 0.90 (N = 108) after accounting for the long-term variations in
319
aerosol characteristics. In addition, the RMSE reduced from 9.6 µg/m3 to 7.2 µg/m3, and the mean
320
absolute percentage error reduced from 22.9% to 17.6%. Annual averages of the satellite-retrieved PM2.5 concentrations were derived by averaging the
321 322
monthly-mean PM2.5 concentrations. Figure 6 shows an evaluation of annual averages of the
323
PM2.5 concentrations derived from the (a) static and (b) dynamic AOD-PM2.5 models against
324
ground observations at all stations during the study period. After consideration of the long-term
325
variations in aerosol characteristics, the correlation coefficient greatly increased from 0.56 to
326
0.80 (N = 51). This city-scale correlation is comparable to those from other studies on national
327
scales.16,17,37,38 The RMSE decreased from 5.4 µg/m3 to 4.0 µg/m3, and the mean absolute
328
percentage error decreased from 12.3% to 9.5%. The evaluations were also performed after the
329
training datasets at Tung Chung station were removed. The results show that correlation
330
coefficient increased from 0.58 to 0.78 (N = 42). The RMSE decreased from 5.5 µg/m3 to 4.3
331
µg/m3, and the mean absolute percentage error decreased from 12.6% to 10.6%.
332
(a)
333
Figure 6. Evaluation of annual averages of the PM2.5 concentrations derived from the (a) static
334
and (b) dynamic AOD-PM2.5 models against ground observations at all stations during the study
(b)
17 ACS Paragon Plus Environment
Environmental Science & Technology
335
period. Statistical metrics include correlation coefficient (R), root mean square error (RMSE),
336
mean error (E), mean absolute error (|E|), mean percentage error (PE), and mean absolute
337
percentage error (|PE|). Figure 7 shows spatial distributions of the annual satellite-derived PM2.5 concentrations over
338 339
Hong Kong from 2004 to 2012 using the (a) static and (b) dynamic AOD-PM2.5 models. The
340
points represent the available ground observations. Central and northwestern areas of Hong
341
Kong experienced the highest level of PM2.5 concentration. Using the static AOD-PM2.5 model,
342
the PM2.5 concentrations were underestimated during the early years, whereas they were
343
overestimated during the recent years. After accounting for the long-term variations in aerosol
344
characteristics, both the satellite-derived and ground-observed PM2.5 concentrations show a
345
significant decreasing trend during the study period.
346
(a)
347
(b)
18 ACS Paragon Plus Environment
Page 18 of 29
Page 19 of 29
Environmental Science & Technology
348
Figure 7. Spatial distributions of the annual satellite-derived PM2.5 concentrations over Hong
349
Kong from 2004 to 2012 using the (a) static and (b) dynamic AOD-PM2.5 models. The points
350
represent the available ground observations. The solid red lines in Figure 8 represent inter-annual variations of the average satellite-
351 352
derived PM2.5 concentrations at the four specific stations (i.e., Tung Chung, Yuen Long, Tsuen
353
Wan, and Tap Mun) from 2004 to 2012 using the (a) static and (b) dynamic AOD-PM2.5 models.
354
The dashed blue lines represent the corresponding ground observations. After consideration of
355
the long-term variations in aerosol characteristics, the correlation coefficient greatly increased
356
from 0.64 to 0.94 (N = 9). In addition, the RMSE decreased from 4.3 µg/m3 to 2.0 µg/m3, and the
357
mean absolute percentage error decreased from 10.2% to 4.7%.
358
(a)
359
Figure 8. The solid red lines represent inter-annual variations of the average satellite-derived
360
PM2.5 concentrations at the four specific stations from 2004 to 2012 using the (a) static and (b)
361
dynamic AOD-PM2.5 models. The dashed blue lines represent the corresponding ground
362
observations. Statistical metrics include correlation coefficient (R), root mean square error
363
(RMSE), mean error (E), mean absolute error (|E|), mean percentage error (PE), and mean
364
absolute percentage error (|PE|).
(b)
365
Both the model parameters (e.g., γ’ and K) are associated with aerosol characteristics. We
366
further develop the relationships between the long-term variations in the γ’/K values and aerosol
367
characteristics. The γ’ value reached its peak in 2009, which was consistent with the variations of 19 ACS Paragon Plus Environment
Environmental Science & Technology
Page 20 of 29
368
the highly hygroscopic components (e.g., sulfate and nitrate). The K value increased from 2004
369
to 2011, which was determined by the increasing abundance of the strong light-extinction
370
components (e.g., OM and ammonium) and decreasing dry FMF. A multilinear regression using
371
the inter-annual variations of the γ’ and K values, and the aerosol physical and chemical
372
characteristics is performed to empirically establish their associations:
373
(7.1)
𝛾′ = 𝛽1·[𝑆𝑢𝑙𝑓𝑎𝑡𝑒] + 𝛽2·[𝑁𝑖𝑡𝑟𝑎𝑡𝑒] + 𝛽0 1
(7.2)
374
𝐾 = 𝜆1·[𝑂𝑀] + 𝜆2·[𝐴𝑚𝑚𝑜𝑛𝑖𝑢𝑚] + 𝜆3·𝐹′ + 𝜆0
375
where [Sulfate], [Nitrate], [OM], and [Ammonium] represent the percentages of different
376
chemical compositions in PM10; β1, β2, λ1, λ2, and λ3 represent the slopes; β0 and λ0 denote the
377
intercepts. The β1, β2, and β0 values were estimated to be 1.75, 23.16, and 0.11, respectively. The
378
λ1, λ2, λ3, and λ0 values were estimated to be 25.42 m2/g, 7.73 m2/g, 3.36 m2/g, and -5.01 m2/g,
379
respectively. Using these empirical relationships, the γ’ and K values are predicted from the
380
chemical composition data. Figure S9 shows the inter-annual variations of the predicted γ’ and K
381
values using the chemical composition data at Tung Chung station. Because the chemical
382
composition data at Tung Chung were not available for 2007, the predicted γ’ and K values in
383
2007 were filled by the averages of the values in 2006 and 2008. The prediction results are
384
reasonable, with the γ’ value reached its peak in 2009 and the K value increased from 2004 to
385
2011.
386
We then retrieve PM2.5 concentration using the AOD-PM2.5 model that takes the predicted γ’
387
and K values as inputs. Figure S10 shows an evaluation of annual averages of the retrieved PM2.5
388
concentrations against ground observations at all stations during the study period. The
389
correlation coefficient, RMSE, and mean absolute percentage error were estimated to be 0.81 (N
390
= 51), 3.9 µg/m3, and 8.7%, respectively. The evaluations were also performed after the training 20 ACS Paragon Plus Environment
Page 21 of 29
Environmental Science & Technology
391
datasets at Tung Chung station were removed. The results show that correlation coefficient,
392
RMSE, and mean absolute percentage error were 0.80 (N = 42), 4.2 µg/m3, and 9.8%,
393
respectively. These statistical metrics are similar to those obtained from the dynamic AOD-PM2.5
394
model that takes the raw γ’ and K values as inputs. These results underscore the importance of
395
incorporation of the long-term variations in aerosol characteristics in the PM2.5 estimation model.
396
Discussion
397
Traditional MODIS AOD products have a spatial resolution of 10 km (recently upgraded to 3
398
km). Hong Kong is a compact city with an area of about 30 × 40 km2. To better delineate the
399
spatial distribution of PM2.5 concentration over Hong Kong, we constructed the AOD data at a
400
high resolution of 1 km. The verifications of the satellite-retrieved AOD against the ground-
401
based observations showed a good agreement.28,29,39 The MODIS dataset is affected by missing
402
data (e.g., resulting from the bright surface and cloud effects) and cannot provide information on
403
diurnal variations in PM2.5 concentration. Both effects introduced an error of within 10%.40 To
404
compensate for the data loss, future studies can obtain other AOD datasets derived from the
405
Deep-Blue algorithm or geostationary satellites.41,42
406
Various models have been developed to retrieve ground-level PM2.5 concentration using
407
satellite remote sensing techniques. The statistical models (e.g., land-use regression models)
408
input as many data as possible. These statistical models therefore tend to have lower prediction
409
errors. Looking from a scientific point of view is important because it can provide supporting
410
information for researches in other regions. Aerosol characteristics play an essential role in the
411
AOD-PM2.5 relationship. A common way to retrieve PM2.5 concentration based on science is
412
relying on the model outputs from the CTMs. It is important to improve our scientific
413
understanding of the relationship between AOD and PM2.5 concentration. We studied the role of
21 ACS Paragon Plus Environment
Environmental Science & Technology
414
the aerosol characteristics in the AOD-PM2.5 relationship using an observation-based model. The
415
relationship between the model parameters and aerosol chemical compositions developed in this
416
study has good applicability. In addition, our model do not need to input extensive ground PM2.5
417
measurements, promoting the applicability of the model.
418
The K value is associated with the dry mass extinction efficiency of PM10 particles (α’ext,10)
419
and dry fine mode fraction (F’). The γ’ value is associated with hygroscopic growth coefficient of
420
PM10 particles (γ0), humidity dependence of mass extinction efficiency of PM10 particles, and
421
humidity dependence of fine mode fraction. The α’ext,10 and γ0 values are associated with
422
chemical compositions of PM10 particles. The effect of size distribution of particles is taken into
423
account by using the fine mode fraction.
424
In this study, we assumed that the model parameters (e.g., γ’ and K) derived at Tung Chung
425
station represented the conditions over all of Hong Kong. Uncertainty caused by this assumption
426
was estimated by assessing the spatial variability of different chemical compositions in PM10
427
over Hong Kong. The long-term measurements of PM10 chemical compositions were available at
428
six general stations (e.g., Tung Chung, Yuen Long, Tsuen Wan, Central Western, Kwun Tong,
429
and Sham Shui Po) over Hong Kong.27 Figure S11 shows the spatial variability of the nine-year
430
average of the percentages of different chemical compositions in PM10 at these six stations. The
431
mean and standard deviation of the nine-year average of the percentages of nitrate, ammonium,
432
EC, OM, and sulfate in PM10 were estimated to be 1.66 ± 0.15%, 4.68 ± 0.13%, 6.39 ± 0.35%,
433
20.73 ± 1.16%, and 21.12 ± 0.91%, respectively. The relative standard deviations (a ratio of
434
standard deviation and mean) for all chemical compositions were within 10%. Future studies can
435
introduce the effect of spatial variation of aerosol characteristics in the model when simultaneous
22 ACS Paragon Plus Environment
Page 22 of 29
Page 23 of 29
Environmental Science & Technology
436
observations of meteorological values, mass concentrations and chemical compositions of
437
particulate matter are available at more stations over Hong Kong.
438
This study focused on an investigation of the effect of the long-term variations in aerosol
439
characteristics on satellite remote sensing of PM2.5 concentrations. The annual-average K value
440
changed from 4.18 m2/g in 2004 to 5.34 m2/g in 2012. During this nine-year period, the annual
441
average of K experienced a change rate of around 20%. The K value also showed distinctive
442
levels among the four seasons. It changed from 4.92 m2/g in spring to 5.77 m2/g in winter, with a
443
change rate of around 15%. According to Eq. (6), the variations in the K value can affect the
444
retrieved PM2.5 concentration by the same rate, suggesting a need to incorporate variations in
445
aerosol characteristics in the PM2.5 estimation models.
446
By using the γ0 value, we separated the combined humidity dependence from the MEE and
447
hygroscopic growth with that from the FMF. Because of the humidity dependence from the
448
FMF, the γ’ value was lower than γ0 by about 0.07 (about 8%). It suggests that the MEE and
449
hygroscopic growth dominated the humidity dependence. One limitation of this method is that
450
the separation of the humidity dependence from the MEE and hygroscopic growth is difficult.
451
For a city (e.g., Hong Kong) with abundant highly hygroscopic components in aerosols, the
452
hygroscopic growth effect should occupy a substantial part in the humidity dependence.
453
The lack of measurement of aerosol characteristics limits the assessment of their effects.
454
Compared with most cities in mainland China, Hong Kong operates a much more comprehensive
455
system for monitoring air quality over a long period. Therefore, we used the long-term
456
measurements of mass concentrations and chemical compositions of particulate matter over
457
Hong Kong. Given the vast territory of China, aerosol characteristics are likely to show a
458
significant regional disparity. Studies in several cities in China demonstrated that the percentages
23 ACS Paragon Plus Environment
Environmental Science & Technology
459
of sulfate, nitrate, and organic carbon in particulate matter varied from 25% to 54%, from 4% to
460
27%, and from 20% to 48%, respectively.24 In addition, regional disparities were found in their
461
long-term trends. These results suggest that the variations in aerosol characteristics in different
462
regions of China should pose different impacts on PM2.5 retrievals. In addition to this regional
463
disparity, their effects on the AOD-PM2.5 relationship should show some similar features.
464
Variations in components such as sulfate and nitrate should pose impacts on the PM2.5 retrievals
465
through the aerosol hygroscopic-growth ability. Variations in components such as OM should
466
pose impacts on the PM2.5 retrievals through the aerosol light-extinction ability. Therefore, it is
467
of great value for future studies to examine the similarity and disparity of their impacts for
468
different regions.
469
High levels of PM2.5 concentration can cause severe visibility impairments.43 As a result of
470
control efforts, both the PM2.5 concentration and visibility-derived σa,0 showed a descending
471
trend in Hong Kong during the study period. Figure S12 shows the long-term variation in the
472
ratio between the visibility-derived σa,0 and PM2.5 concentration at Tung Chung from 2004 to
473
2012. An ascending trend in the ratio was observed, suggesting that the reduction of σa,0 was
474
slower than the reduction of PM2.5 concentration. In terms of improving visibility, the benefit of
475
the reduction in PM2.5 concentration was offset by other factors, such as the changes in aerosol
476
chemical compositions. To better improve visibility in Hong Kong, it is suggested that more
477
control efforts focus on the reduction of strong light-extinction components, such as OM.
478
Associated content
479
Supporting Information
480
Additional data information and figures (Figure S1-S12).
481
Acknowledgments
24 ACS Paragon Plus Environment
Page 24 of 29
Page 25 of 29
Environmental Science & Technology
482
We thank the Hong Kong Environmental Protection Department for providing air-quality
483
monitoring data. This work was supported by the National Natural Science Foundation of China
484
(Grant No. 41575106), the Science and Technology Plan Project of Guangdong Province of China
485
(Grant No. 2015A020215020 and 2017A050506003), NSFC/RGC (Grant N_HKUST631/05), and
486
the Fok Ying Tung Graduate School (NRC06/07.SC01). The authors declare they have no actual
487
or potential competing financial interests.
488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504
25 ACS Paragon Plus Environment
Environmental Science & Technology
505
Reference
506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549
(1)
Zhong, L.; Louie, P. K. K.; Zheng, J.; Yuan, Z.; Yue, D.; Ho, J. W. K.; Lau, A. K. H. Science–Policy Interplay: Air Quality Management in the Pearl River Delta Region and Hong Kong. Atmos. Environ. 2013, 76, 3–10. https://doi.org/10.1016/j.atmosenv.2013.03.012. (2) Brauer, M.; Amann, M.; Burnett, R. T.; Cohen, A.; Dentener, F.; Ezzati, M.; Henderson, S. B.; Krzyzanowski, M.; Martin, R. V.; Van Dingenen, R.; van Donkelaar, A.; Thurston, G.D. Exposure Assessment for Estimation of the Global Burden of Disease Attributable to Outdoor Air Pollution. Environ. Sci. Technol. 2012, 46 (2), 652–660. https://doi.org/10.1021/es2025752. (3) Chan, T.-C.; Zhang, Z.; Lin, B.-C.; Lin, C.; Deng, H.-B.; Chuang, Y. C.; Chan, J. W. M.; Jiang, W. K.; Tam, T.; Chang, L.; Hoek, G.; Lau, A.K.H.; Lao, X.Q. Long-Term Exposure to Ambient Fine Particulate Matter and Chronic Kidney Disease: A Cohort Study. Environ. Health Perspect. 2018, 126 (10), 107002. https://doi.org/10.1289/EHP3304. (4) Guo, C.; Zhang, Z.; Lau, A. K. H.; Lin, C. Q.; Chuang, Y. C.; Chan, J.; Jiang, W. K.; Tam, T.; Yeoh, E.-K.; Chan, T.-C.; Chang, T.; Lao, X.Q. Effect of Long-Term Exposure to Fine Particulate Matter on Lung Function Decline and Risk of Chronic Obstructive Pulmonary Disease in Taiwan: A Longitudinal, Cohort Study. Lancet Planet. Health 2018, 2 (3), e114–e125. https://doi.org/10.1016/S2542-5196(18)30028-7. (5) Zhang, Z.; Hoek, G.; Chang, L.; Chan, T.-C.; Guo, C.; Chuang, Y. C.; Chan, J.; Lin, C.; Jiang, W. K.; Guo, Y.; Vermeulen, R.; Yeoh, E.K.; Tam, T.; Lau, A.K.H.; Griffiths, S.; Lao, X.Q. Particulate Matter Air Pollution, Physical Activity and Systemic Inflammation in Taiwanese Adults. Int. J. Hyg. Environ. Health 2018, 221 (1), 41–47. https://doi.org/10.1016/j.ijheh.2017.10.001. (6) Lao, X. Q.; Zhang, Z.; Lau, A. K.; Chan, T.-C.; Chuang, Y. C.; Chan, J.; Lin, C.; Guo, C.; Jiang, W. K.; Tam, T.; Hoek, G.; Kan, H.D.; Yeoh, E.K.; Chang, L. Exposure to Ambient Fine Particulate Matter and Semen Quality in Taiwan. Occup Env. Med 2017, oemed2017-104529. https://doi.org/10.1136/oemed-2017-104529. (7) Lin, C. Q.; Lau, A. K. H.; Li, Y.; Fung, J. C. H.; Li, C. C.; Lu, X. C.; Li, Z. Y. Difference in PM2.5 Variations between Urban and Rural Areas over Eastern China from 2001 to 2015. Atmosphere 2018, 9 (8), 312. https://doi.org/10.3390/atmos9070000. (8) Lin, C.; Li, Y.; Yuan, Z.; Lau, A. K. H.; Li, C.; Fung, J. C. H. Using Satellite Remote Sensing Data to Estimate the High-Resolution Distribution of Ground-Level PM2.5. Remote Sens. Environ. 2015, 156, 117–128. https://doi.org/10.1016/j.rse.2014.09.015. (9) Liu, Y. Monitoring PM2.5 from Space for Health: Past, Present, and Future Directions. Air & Waste Management Association 2014, em, 6–10. (10) Hand, J. L.; Malm, W. C. Review of Aerosol Mass Scattering Efficiencies from GroundBased Measurements since 1990. J. Geophys. Res. Atmospheres 2007, 112, D16203 1-24. https://doi.org/10.1029/2007JD008484. (11) Lee, S.; Ghim, Y. S.; Kim, S.-W.; Yoon, S.-C. Seasonal Characteristics of Chemically Apportioned Aerosol Optical Properties at Seoul and Gosan, Korea. Atmos. Environ. 2009, 43 (6), 1320–1328. https://doi.org/10.1016/j.atmosenv.2008.11.044. (12) Kim, K. W.; Kim, Y. J.; Oh, S. J. Visibility Impairment during Yellow Sand Periods in the Urban Atmosphere of Kwangju, Korea. Atmos. Environ. 2001, 35 (30), 5157–5167. https://doi.org/10.1016/S1352-2310(01)00330-2. 26 ACS Paragon Plus Environment
Page 26 of 29
Page 27 of 29
550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595
Environmental Science & Technology
(13) Watson, J. G. Visibility: Science and Regulation. J. Air Waste Manag. Assoc. 2002, 52 (6), 628–713. https://doi.org/10.1080/10473289.2002.10470813. (14) Li, Y.; Huang, H. X. H.; Griffith, S. M.; Wu, C.; Lau, A. K. H.; Yu, J. Z. Quantifying the Relationship between Visibility Degradation and PM2.5 Constituents at a Suburban Site in Hong Kong: Differentiating Contributions from Hydrophilic and Hydrophobic Organic Compounds. Sci. Total Environ. 2017, 575, 1571–1581. https://doi.org/10.1016/j.scitotenv.2016.10.082. (15) Boys, B. L.; Martin, R. V.; van Donkelaar, A.; MacDonell, R. J.; Hsu, N. C.; Cooper, M. J.; Yantosca, R. M.; Lu, Z.; Streets, D. G.; Zhang, Q.; Wang, S.W. Fifteen-Year Global Time Series of Satellite-Derived Fine Particulate Matter. Environ. Sci. Technol. 2014, 48 (19), 11109–11118. https://doi.org/10.1021/es502113p. (16) Geng, G.; Zhang, Q.; Martin, R. V.; van Donkelaar, A.; Huo, H.; Che, H.; Lin, J.; He, K. Estimating Long-Term PM2.5 Concentrations in China Using Satellite-Based Aerosol Optical Depth and a Chemical Transport Model. Remote Sens. Environ. 2015, 166, 262– 270. https://doi.org/10.1016/j.rse.2015.05.016. (17) van Donkelaar, A.; Martin, R. V.; Brauer, M.; Kahn, R.; Levy, R.; Verduzco, C.; Villeneuve, P. J. Global Estimates of Ambient Fine Particulate Matter Concentrations from Satellite-Based Aerosol Optical Depth: Development and Application. Environ. Health Perspect. 2010, 118 (6), 847–855. (18) Ma, Z.; Hu, X.; Huang, L.; Bi, J.; Liu, Y. Estimating Ground-Level PM2.5 in China Using Satellite Remote Sensing. Environ. Sci. Technol. 2014, 48 (13), 7436–7444. https://doi.org/10.1021/es5009399. (19) Hu, X.; Waller, L. A.; Lyapustin, A.; Wang, Y.; Liu, Y. 10-Year Spatial and Temporal Trends of PM2.5 Concentrations in the Southeastern US Estimated Using High-Resolution Satellite Data. Atmos Chem Phys 2014, 14 (12), 6301–6314. https://doi.org/10.5194/acp14-6301-2014. (20) Liu, Y.; Sarnat, J. A.; Kilaru, V.; Jacob, D. J.; Koutrakis, P. Estimating Ground-Level PM2.5 in the Eastern United States Using Satellite Remote Sensing. Environ. Sci. Technol. 2005, 39 (9), 3269–3278. https://doi.org/10.1021/es049352m. (21) Lee, H. J.; Coull, B. A.; Bell, M. L.; Koutrakis, P. Use of Satellite-Based Aerosol Optical Depth and Spatial Clustering to Predict Ambient PM2.5 Concentrations. Environ. Res. 2012, 118, 8–15. https://doi.org/10.1016/j.envres.2012.06.011. (22) Wang, J.; Zhao, B.; Wang, S.; Yang, F.; Xing, J.; Morawska, L.; Ding, A.; Kulmala, M.; Kerminen, V.-M.; Kujansuu, J.; Wang, Z.; Ding, D.; Zhang, X.; Wang, H.; Tian, M.; Petaja, T.; Jiang, J.; Hao, J.M. Particulate Matter Pollution over China and the Effects of Control Policies. Sci. Total Environ. 2017, 584–585, 426–447. https://doi.org/10.1016/j.scitotenv.2017.01.027. (23) Klimont, Z.; Kupiainen, K.; Heyes, C.; Purohit, P.; Cofala, J.; Rafaj, P.; Borken-Kleefeld, J.; Schöpp, W. Global Anthropogenic Emissions of Particulate Matter Including Black Carbon. 2017, 17 (14), 8681–8723. (24) Geng, G.; Zhang, Q.; Tong, D.; Li, M.; Zheng, Y.; Wang, S.; He, K. Chemical Composition of Ambient PM2. 5 over China and Relationship to Precursor Emissions during 2005–2012. Atmospheric Chem. Phys. 2017, 17 (14), 9187–9203. https://doi.org/https://doi.org/10.5194/acp-17-9187-2017. (25) Sun, Y. L.; Wang, Z. F.; Du, W.; Zhang, Q.; Wang, Q. Q.; Fu, P. Q.; Pan, X. L.; Li, J.; Jayne, J.; Worsnop, D. R. Long-Term Real-Time Measurements of Aerosol Particle 27 ACS Paragon Plus Environment
Environmental Science & Technology
596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640
(26) (27) (28)
(29)
(30) (31) (32) (33)
(34)
(35)
(36)
(37)
Composition in Beijing, China: Seasonal Variations, Meteorological Effects, and Source Analysis. Atmos Chem Phys 2015, 15 (17), 10149–10165. https://doi.org/10.5194/acp-1510149-2015. Cao, J.; Shen, Z.; Chow, J. C.; Qi, G.; Watson, J. G. Seasonal Variations and Sources of Mass and Chemical Composition for PM10 Aerosol in Hangzhou, China. Particuology 2009, 7 (3), 161–168. https://doi.org/10.1016/j.partic.2009.01.009. Yuan, Z.; Yadav, V.; Turner, J. R.; Louie, P. K. K.; Lau, A. K. H. Long-Term Trends of Ambient Particulate Matter Emission Source Contributions and the Accountability of Control Strategies in Hong Kong over 1998–2008. Atmos. Environ. 2013, 76, 21–31. Li, C.; Lau, A. K. H.; Mao, J.; Chu, D. A. Retrieval, Validation, and Application of the 1Km Aerosol Optical Depth from MODIS Measurements over Hong Kong. IEEE Trans. Geosci. Remote Sens. 2005, 43 (11), 2650–2658. https://doi.org/10.1109/TGRS.2005.856627. Lin, C. Q.; Liu, G.; Lau, A. K. H.; Li, Y.; Li, C. C.; Fung, J. C. H.; Lao, X. Q. HighResolution Satellite Remote Sensing of Provincial PM2.5 Trends in China from 2001 to 2015. Atmos. Environ. 2018, 180, 110–116. https://doi.org/10.1016/j.atmosenv.2018.02.045. Ho, K. F.; Lee, S. C.; Cao, J. J.; Chow, J. C.; Watson, J. G.; Chan, C. K. Seasonal Variations and Mass Closure Analysis of Particulate Matter in Hong Kong. Sci. Total Environ. 2006, 355 (1–3), 276–287. https://doi.org/10.1016/j.scitotenv.2005.03.013. Louie, P. K. K.; Watson, J. G.; Chow, J. C.; Chen, A.; Sin, D. W. M.; Lau, A. K. H. Seasonal Characteristics and Regional Transport of PM2.5 in Hong Kong. Atmos. Environ. 2005, 39 (9), 1695–1710. https://doi.org/10.1016/j.atmosenv.2004.11.017. Pathak, R. K.; Yao, X.; Lau, A. K. H.; Chan, C. K. Acidity and Concentrations of Ionic Species of PM2.5 in Hong Kong. Atmos. Environ. 2003, 37 (8), 1113–1124. https://doi.org/10.1016/S1352-2310(02)00958-5. Yu, J. Z.; Tung, J. W. T.; Wu, A. W. M.; Lau, A. K. H.; Louie, P. K.-K.; Fung, J. C. H. Abundance and Seasonal Characteristics of Elemental and Organic Carbon in Hong Kong PM10. Atmos. Environ. 2004, 38 (10), 1511–1521. https://doi.org/10.1016/j.atmosenv.2003.11.035. Huang, X. H. H.; Bian, Q. J.; Ng, W. M.; Louie, P. K. K.; Yu, J. Z. Characterization of PM2.5 Major Components and Source Investigation in Suburban Hong Kong: A One Year Monitoring Study. Aerosol Air Qual. Res. 2014, 14, 237–250. https://doi.org/10.4209/aaqr.2013.01.0020. So, K. L.; Guo, H.; Li, Y. S. Long-Term Variation of PM2.5 Levels and Composition at Rural, Urban, and Roadside Sites in Hong Kong: Increasing Impact of Regional Air Pollution. Atmos. Environ. 2007, 41 (40), 9427–9434. https://doi.org/10.1016/j.atmosenv.2007.08.053. Zhang, X.; Yuan, Z.; Li, W.; Lau, A. K. H.; Yu, J. Z.; Fung, J. C. H.; Zheng, J.; Yu, A. L. C. Eighteen-Year Trends of Local and Non-Local Impacts to Ambient PM10 in Hong Kong Based on Chemical Speciation and Source Apportionment. Atmospheric Res. 2018, 214, 1–9. https://doi.org/10.1016/j.atmosres.2018.07.004. van Donkelaar, A.; Martin, R. V.; Brauer, M.; Boys, B. L. Use of Satellite Observations for Long-Term Exposure Assessment of Global Concentrations of Fine Particulate Matter. Environ. Health Perspect. 2015, 123 (2), 135–143. https://doi.org/10.1289/ehp.1408646.
28 ACS Paragon Plus Environment
Page 28 of 29
Page 29 of 29
641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661
Environmental Science & Technology
(38) Peng, J.; Chen, S.; Lü, H.; Liu, Y.; Wu, J. Spatiotemporal Patterns of Remotely Sensed PM2.5 Concentration in China from 1999 to 2011. Remote Sens. Environ. 2016, 174, 109– 121. https://doi.org/10.1016/j.rse.2015.12.008. (39) Li, Y.; Lin, C.; Lau, A. K. H.; Liao, C.; Zhang, Y.; Zeng, W.; Li, C.; Fung, J. C. H.; Tse, T. K. T. Assessing Long-Term Trend of Particulate Matter Pollution in the Pearl River Delta Region Using Satellite Remote Sensing. Environ. Sci. Technol. 2015, 49 (19), 11670–11678. https://doi.org/10.1021/acs.est.5b02776. (40) Lin, C. Q.; Li, Y.; Lau, A. K. H.; Deng, X. J.; Tse, K. T.; Fung, J. C. H.; Li, C. C.; Li, Z. Y.; Lu, X. C.; Zhang, X. G.; Yu, Q.W. Estimation of Long-Term Population Exposure to PM2.5 for Dense Urban Areas Using 1-Km MODIS Data. Remote Sens. Environ. 2016, 179, 13–22. https://doi.org/10.1016/j.rse.2016.03.023. (41) Hsu, N. C.; Tsay, S.-C.; King, M. D.; Herman, J. R. Aerosol Properties Over BrightReflecting Source Regions. IEEE Trans. Geosci. Remote Sens. 2004, 42 (3), 557–569. https://doi.org/10.1109/TGRS.2004.824067. (42) Wang, W.; Mao, F.; Du, L.; Pan, Z.; Gong, W.; Fang, S. Deriving Hourly PM2.5 Concentrations from Himawari-8 AODs over Beijing–Tianjin–Hebei in China. Remote Sens. 2017, 9 (8), 858. https://doi.org/10.3390/rs9080858. (43) Pui, D. Y. H.; Chen, S.-C.; Zuo, Z. PM2.5 in China: Measurements, Sources, Visibility and Health Effects, and Mitigation. Particuology 2014, 13, 1–26. https://doi.org/10.1016/j.partic.2013.11.001.
29 ACS Paragon Plus Environment