Subscriber access provided by University of South Dakota
Environmental Modeling
Global Sources of Fine Particulate Matter: Interpretation of PM Chemical Composition Observed by SPARTAN using a Global Chemical Transport Model 2.5
Crystal L Weagle, Graydon Snider, Chi Li, Aaron van Donkelaar, Sajeev Philip, Paul Bissonnette, Jaqueline Burke, John Jackson, Robyn Latimer, Emily Stone, Ihab Abboud, Clement Akoshile, Nguyen Xuan Anh, Jeffrey Robert Brook, Aaron Cohen, Jinlu Dong, Mark D. Gibson, Derek Griffith, Kebin He, Brent N Holben, Ralph Kahn, Christoph A Keller, Jong Sung Kim, Nofel Lagrosas, Puji Lestari, Yeo Lik Khian, Yang Liu, Eloise A. Marais, J. Vanderlei Martins, Amit Misra, Ulfi Muliane, Rizki Pratiwi, Eduardo J. Quel, Abdus Salam, Lior Segev, Sachchida Nand Tripathi, Chien Wang, Qiang Zhang, Michael Brauer, Yinon Rudich, and Randall V Martin Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.8b01658 • Publication Date (Web): 14 Sep 2018 Downloaded from http://pubs.acs.org on September 15, 2018
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 35
Environmental Science & Technology
Lagrosas, Nofel; Ateneo de Manila University, Manila Observatory Lestari, Puji; Institute of Technology Bandung, Faculty of Civil and Environmental Engineering Lik Khian, Yeo; MIT, Center for Global Change Science Liu, Yang; Emory University, Rollins School of Public Health Marais, Eloise; University of Birmingham, School of Geography, Earth and Environmental Sciences Martins, J.; University of Maryland Baltimore County, Department of Physics and Joint Center for Earth Systems Technology Misra, Amit ; Indian Institute of Technology Kanpur, Center for Environmental Science and Engineering Muliane, Ulfi; Institute of Technology Bandung, Faculty of Civil and Environmental Engineering Pratiwi, Rizki ; Institute of Technology Bandung, Faculty of Civil and Environmental Engineering Quel, Eduardo; UNIDEF (CITEDEF-CONICET) , División Lidar - CEILAP Salam, Abdus; Dhaka University, Chemistry Segev, Lior; Weizmann Institute of Science, Department of Earth and Planetary Sciences Tripathi, Sachchida; Indian Institute of Technology Kanpur, Center for Environmental Science and Engineering Wang, Chien; MIT, Center for Global Change Science Zhang, Qiang; Tsinghua University, Center for Earth System Science Brauer, Michael; University of British Columbia, School of Population and Public Health Rudich, Yinon; Weizmann Institute of Science, Department of Earth and Planetary Sciences Martin, Randall; Dalhousie University, Department of Chemistry; Dalhousie University, Department of Physics and Atmospheric Science; HarvardSmithsonian Center for Astrophysics
ACS Paragon Plus Environment
Environmental Science & Technology
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49
Global Sources of Fine Particulate Matter: Interpretation of PM2.5 Chemical Composition Observed by SPARTAN using a Global Chemical Transport Model Crystal L. Weagle1,2*, Graydon Snider2, Chi Li2, Aaron van Donkelaar2, Sajeev Philip2,3, Paul Bissonnette2, Jaqueline Burke2, John Jackson2, Robyn Latimer2, Emily Stone2, Ihab Abboud4, Clement Akoshile5, Nguyen Xuan Anh6, Jeffrey Robert Brook7, Aaron Cohen8, Jinlu Dong9, Mark D. Gibson10, Derek Griffith11, Kebin B. He9, Brent N. Holben12, Ralph Kahn12, Christoph A. Keller13,14, Jong Sung Kim15, Nofel Lagrosas16, Puji Lestari17, Yeo Lik Khian18, Yang Liu19, Eloise A. Marais20, J. Vanderlei Martins21, Amit Misra22, Ulfi Muliane17, Rizki Pratiwi17, Eduardo J. Quel23, Abdus Salam24, Lior Segev25, Sachchida N. Tripathi22, Chien Wang18, Qiang Zhang9, Michael Brauer26, Yinon Rudich25, Randall V. Martin1,2,27
* Correspondence to
[email protected]; Tel: 902-494-1820; Fax: 902-494-5191 Affiliations 1 Department of Chemistry, Dalhousie University, Halifax, Canada 2 Department of Physics and Atmospheric Science, Dalhousie University, Halifax, Canada 3 NASA Ames Research Center, Moffett Field, California, United States of America 4 Centre for Atmospheric Research Experiments, Environment and Climate Change Canada, Egbert, Ontario, Canada 5 Department of Physics, University of Ilorin, Ilorin, Nigeria 6 Institute of Geophysics, Vietnam Academy of Science and Technology, Hanoi, Vietnam 7 Department of Public Health Sciences, University of Toronto, Toronto, Ontario, Canada M5S 1A8 8 Health Effects Institute, 75 Federal Street Suite 1400, Boston, MA 02110-1817, USA 9 Department of Earth System Science, Tsinghua University, Beijing, China 10 Department of Civil and Resource Engineering, Dalhousie University, Halifax, Canada 11 Council for Scientific and Industrial Research (CSIR), Pretoria, South Africa 12 Earth Science Division, NASA Goddard Space Flight Center, Greenbelt, Maryland, USA 13 Universities Space Research Association/Goddard Earth Science Technology and Research, Columbia, MD, USA 14 Global Modeling and Assimilation Office, NASA Goddard Space Flight Center, Greenbelt, Maryland, USA 15 Department of Community Health and Epidemiology, Dalhousie University, Halifax, Canada 16 Manila Observatory, Ateneo de Manila University campus, Quezon City, Philippines 17 Faculty of Civil and Environmental Engineering, ITB, JL. Ganesha No.10, Bandung 40132, Indonesia 18 Center for Global Change Science, Massachusetts Institute of Technology, Cambridge, MA, 02139, USA 19 Rollins School of Public Health, Emory University, 1518 Clifton Road NE, Atlanta, GA 30322, United States 20 School of Geography, Earth and Environmental Sciences, University of Birmingham, Birmingham B15 2TT, United Kingdom 21 Department of Physics and Joint Center for Earth Systems Technology, University of Maryland, Baltimore County, Baltimore, Maryland, USA 22 Center for Environmental Science and Engineering, Indian Institute of Technology Kanpur, India 23 UNIDEF (CITEDEF-CONICET) Juan B. de la Salle 4397 – B1603ALO Villa Martelli, Buenos Aires, Argentina 24 Department of Chemistry, University of Dhaka, Dhaka - 1000, Bangladesh 25 Department of Earth and Planetary Sciences, Weizmann Institute, Rehovot 76100, Israel 26 School of Population and Public Health, University of British Columbia, Vancouver, British Columbia, Canada 27 Harvard-Smithsonian Center for Astrophysics, Cambridge, MA 02138, USA
1
ACS Paragon Plus Environment
Page 2 of 35
Page 3 of 35
50 51
Environmental Science & Technology
Abstract
52 53
Exposure to ambient fine particulate matter (PM2.5) is a leading risk factor for the global burden
54
of disease. However, uncertainty remains about PM2.5 sources. We use a global chemical
55
transport model (GEOS-Chem) simulation for 2014, constrained by satellite-based estimates of
56
PM2.5 to interpret globally dispersed PM2.5 mass and composition measurements from the
57
ground-based Surface PARTiculate mAtter Network (SPARTAN). Measured site mean PM2.5
58
composition varies substantially for secondary inorganic aerosols (2.4-19.7 µg/m3), mineral dust
59
(1.9-14.7 µg/m3), residual/organic matter (2.1-40.2 µg/m3), and black carbon (1.0-7.3 µg/m3).
60
Interpretation of these measurements with the GEOS-Chem model yields insight into sources
61
affecting each site. Globally, combustion sectors such as residential energy use (7.9 µg/m3),
62
industry (6.5 µg/m3), and power generation (5.6 µg/m3) are leading sources of outdoor global
63
population-weighted PM2.5 concentrations. Global population-weighted organic mass is driven
64
by the residential energy sector (64%) whereas population-weighted secondary inorganic
65
concentrations arise primarily from industry (33%) and power generation (32%). Simulation-
66
measurement biases for ammonium nitrate and dust identify uncertainty in agricultural and
67
crustal sources. Interpretation of initial PM2.5 mass and composition measurements from
68
SPARTAN with the GEOS-Chem model constrained by satellite-based PM2.5 provides insight
69
into sources and processes that influence the global spatial variation in PM2.5 composition.
2
ACS Paragon Plus Environment
Environmental Science & Technology
70 71 72 73
Introduction
74
micrometers or less) is a leading risk factor for increased mortality and morbidity.1,2 The Global
75
Burden of Disease study attributed 4.1 million premature deaths to PM2.5 exposure in 2016.3 A
76
strong need exists to understand the sources contributing to this PM2.5 burden to inform
77
mitigation efforts.4 PM2.5 formation is influenced by a range of emission sources, atmospheric
78
transport, and atmospheric chemistry.5 A chemical transport model constrained by observations
79
offers a powerful tool to understand these sources. Until recently, long-term measurements of
80
PM2.5 mass and chemical composition were primarily limited to North America and Europe, with
81
different networks using a variety of sampling techniques and standards to determine chemical
82
composition. A global dataset of ground-based PM2.5 compositional measurements could offer
83
valuable information to understand the sources and processes that control the spatial diversity of
84
PM2.5 mass and chemical composition.
Exposure to ambient fine particulate matter (PM2.5; aerodynamic diameter of 2.5
85
The relationship between emissions and PM2.5 loading is complex. PM2.5 can be emitted
86
directly as particles from combustion or mechanical processes, but it can also form and grow in
87
the atmosphere through the condensation of low volatility products of atmospheric chemical
88
reactions of inorganic and organic precursors.5,6 Chemical transport models have been applied to
89
represent this complexity through source apportionment studies aimed at characterizing the
90
global sources contributing to PM2.5 mass and composition,7 with increasingly fine resolution.8
91
Although insightful, simulations could benefit from stronger observational constraints to
92
evaluate and improve accuracy and spatial representativeness.
3
ACS Paragon Plus Environment
Page 4 of 35
Page 5 of 35
93
Environmental Science & Technology
Two observational datasets have emerged recently with information about the global
94
distribution of PM2.5 mass and composition, a dedicated ground-based network of PM2.5
95
composition, and increasingly accurate satellite-based estimates of PM2.5 mass. The Surface
96
PARTiculate mAtter Network (SPARTAN) measures ground-based PM2.5 mass and chemical
97
composition using consistent instrumentation and standardized chemical analysis techniques in
98
diverse global locations with high population densities relevant for health. Measurements from
99
SPARTAN include PM2.5 filter samples that are analyzed for PM2.5 mass and chemical
100
composition including sulfate, nitrate, ammonium, black carbon (BC), crustal material, and sea
101
salt.9,10 Satellite-based estimates of PM2.5 mass complement the ground-based measurement
102
network by offering additional global constraints on PM2.5 mass at resolution finer than global
103
simulations. Case studies investigating the effects of model resolution on calculated PM2.5
104
mortality rates find that those calculated from coarse (2°x2.5°) resolution are systematically
105
lower than those calculated using finer (e.g. 0.5°x0.66°) resolution.8,11,12 The latest global
106
satellite-based PM2.5 estimates combine observations from multiple retrieval algorithms13–19 and
107
instruments (MODIS, MISR, SeaWiFs) weighted inversely by error with respect to ground-based
108
AOD measurements from AERONET20 with additional statistical constraints from ground-based
109
PM2.5 measurements.21–23
110
We apply the GEOS-Chem global chemical transport model, constrained by satellite-
111
derived PM2.5, to interpret ground-based measurements of PM2.5 mass and composition from
112
SPARTAN to gain insight into the main sources determining the spatial distribution of PM2.5
113
mass and composition. We explore the annual average influence of major emission source
114
categories on global population-weighted PM2.5.
4
ACS Paragon Plus Environment
Environmental Science & Technology
Methods
115 116 117 118 119
SPARTAN filter measurements and analysis
120
ions, trace elements, and aerosol optical properties at globally dispersed, densely-populated areas
121
of relevance to human health. Snider et al.10 provide an overview of SPARTAN. Instrumentation
122
includes an AirPhoton 3-wavelength integrating nephelometer and an AirPhoton SS4i filter
123
sampler. We focus on the latter here. The PM2.5 is collected on a pre-weighed PTFE filter (2 µm
124
pore size, SKC). Filters sample one diurnal cycle over a 9-day period beginning and ending at
125
09:00 local time before shipment under ambient conditions to Dalhousie University for analysis
126
in a central laboratory. Samples have been collected for periods of 2 months to over 3 years
127
during 2013-2017 at 11 carefully chosen, regionally diverse sites. Although initially sparse,
128
SPARTAN is unique in covering poorly sampled regions of the world with a consistent
129
methodology.
130
SPARTAN is an ongoing, long-term project that measures aerosol mass, water-soluble
An extensive overview of the SPARTAN PM2.5 sampling methodology, filter chemical
131
analysis protocols, and the filter-based hygroscopicity parameter, κ, are provided by Snider et
132
al.9 Details on relevant SPARTAN chemical analysis procedures and the uncertainty associated
133
with each major chemical component based on collocated measurements is presented in the
134
Supporting Information (SI). Seasonal mean uncertainties range from 4.0% for sulfate, to 4.6%
135
for PM2.5, to 9.2% for nitrate. Briefly, gravimetric analysis follows USEPA protocols in a
136
cleanroom with controlled temperature at 20-23°C and relative humidity at 35 ± 5%. Black
137
carbon content is estimated by surface reflectance measurements,24 water-soluble ions are
138
determined by ion chromatography,25 and trace elements are determined by inductively coupled
139
plasma-mass spectrometry.25,26 The residual matter (RM) component, estimated by subtracting
5
ACS Paragon Plus Environment
Page 6 of 35
Page 7 of 35
Environmental Science & Technology
140
the dry inorganic mass and particle-bound water from total PM2.5 mass, is expected to be
141
predominately organic.9 Activities are ongoing to directly measure organics through Aerosol
142
Mass Spectrometry27 and FT-IR spectroscopy.28
143 144 145
GEOS-Chem Simulation
146
chem.org) to determine the daily distribution of PM2.5 major chemical component mass
147
concentrations. GEOS-Chem solves for the evolution of atmospheric aerosols and gases using
148
assimilated meteorology from the NASA Goddard Earth Observing System (GEOS), global and
149
regional emission inventories, and algorithms that represent the physics and chemistry of
150
atmospheric processes. The simulation uses assimilated meteorological observations (GEOS
151
MERRA-2) at 2° x 2.5° horizontal resolution with 47 vertical levels for the year 2014 for overlap
152
with SPARTAN and due to emissions availability. The simulation uses anthropogenic emissions
153
from the Emissions Database for Global Atmospheric Research (EDGAR) version 4.3
154
inventory29 and the MIX regional anthropogenic emission inventory for 29 Asian regions and
155
countries.30
156
We use the GEOS-Chem 3-dimensional chemical transport model (v11.01, http://geos-
SPARTAN measurements are used to inform developments to the simulation. We include
157
the anthropogenic fugitive, combustion, and industrial dust (AFCID) emission inventory from
158
Philip et al.,31 which increased correlation (r) with SPARTAN fine dust concentrations in this
159
study from 0.35 to 0.67. Simulated total PM2.5 mass concentration is calculated at 35% RH for
160
consistency with PM2.5 measurement protocols. The calculation employs the kappa
161
hygroscopicity formulation32 used by SPARTAN, as described by Snider et al.9; this formulation
162
exhibits hygroscopic growth consistent with the Aerosol Inorganic Model (AIM)33 and
163
laboratory measurements.9 Compared to the default GEOS-Chem model, the kappa
6
ACS Paragon Plus Environment
Environmental Science & Technology
164
hygroscopicity formulation decreases the bias (reduced major axis slope) of simulated PM2.5
165
versus SPARTAN PM2.5 concentrations from 1.39 to 1.29. We also include an aqueous-phase
166
mechanism for secondary organic aerosol (SOA) formation from isoprene from Marais et al.34
167
that better represents SPARTAN PM2.5 mass as shown by reduced root-mean-square-error from
168
14.2 to 13.3 µg/m3. We perform sensitivity simulations for which emissions from individual
169
source categories were removed to calculate the contribution of sources to PM2.5 mass and
170
composition by mass balance. Further details about the simulation are provided in the SI.
171 172 173
Constraining the simulation with satellite-based PM2.5
174
distribution at spatial scales commensurate with population density distributions.35 To reduce
175
uncertainties in PM2.5 exposure estimates caused by model resolution,8,36 satellite-derived PM2.5
176
concentrations21 for the year 2014 are used to downscale GEOS-Chem PM2.5 mass and
177
composition from 2°x2.5° to 0.1°x0.1° resolution following a widely used approach.36–42 The
178
spatial map of the annual ratio of satellite-derived to simulated PM2.5 concentration is applied to
179
all simulated PM2.5 components, thus retaining the simulated fraction and temporal variation of
180
PM2.5 composition.
181
Satellite observations of AOD offer an additional constraint on the global PM2.5
Supplemental Table S2 shows the Pearson’s correlation coefficient between SPARTAN
182
measurements and PM2.5 components from the simulation (r) and simulated values scaled by
183
local satellite-derived PM2.5 (rsat). For most PM2.5 components, downscaling to satellite-derived
184
PM2.5 increases the consistency in capturing PM2.5 spatial diversity. Correlations tend to increase,
185
most notably for organic mass (OM, rsat=0.92 vs. r=0.64) as well as total PM2.5 mass (rsat=0.93
186
vs. r=0.88), ammonium (rsat=0.86 vs. r=0.81), and BC (rsat=0.67 vs. r=0.61). The root-mean-
187
square-error of simulated total PM2.5 decreased from 13.3 µg /m3 to 12.8 µg/m3. Prior work has
7
ACS Paragon Plus Environment
Page 8 of 35
Page 9 of 35
Environmental Science & Technology
188
similarly found that the downscaled simulation better represents observations for both mass and
189
composition.36,41 Therefore, all values of total PM2.5 mass and chemical composition reported
190
herein are from the downscaled simulation (scaled by the local annual ratio of PM2.5,sat to
191
PM2.5,model) at 0.1°x0.1° resolution.
192 193 194 195 196
Sources Affecting PM2.5 Mass and Composition Global Distribution of PM2.5 Chemical Composition
197
concentric circles depicting concentrations at SPARTAN sites. The center of each concentric
198
circle indicates the measured value, with corresponding downscaled simulated concentration
199
indicated by the outer ring. Differences between the outer ring and background map represent the
200
effects of sampling the simulation for the same months as the measurements versus complete
201
annual sampling. These seasonal sampling effects are generally much smaller than the global
202
spatial variation, providing evidence of temporal representativeness. Inset values represent the
203
global population-weighted mean concentration inferred from the downscaled simulation. Table
204
1 contains numerical values of measured and simulated concentrations for the specified sampling
205
periods of SPARTAN sampling sites. The spatial variation of SPARTAN site-mean
206
concentrations exceeds a factor of five (e.g. Kanpur, India vs. Buenos Aires, Argentina) for most
207
PM2.5 major chemical components, as described in more detail in the following sections.
208
Figure 1 shows the annual mean simulated PM2.5 chemical composition with overlaid
SPARTAN measurements of secondary inorganic aerosol (SIA) vary by a factor of 8
209
across sampling sites from 2.4 µg/m3 in Ilorin, Nigeria to 19.7 µg/m3 in Beijing, China and
210
account for over 20% of total PM2.5 mass at sampling sites, except Manila, Philippines (16%)
211
and Ilorin (15%). The downscaled simulation captures the spatial heterogeneity of SIA
212
concentrations (r=0.87). SIA tends to be overestimated at SPARTAN sites, including an
8
ACS Paragon Plus Environment
Environmental Science & Technology
213
overestimate of nitrate concentrations in Beijing, as discussed below. SIA dominates population-
214
weighted PM2.5, accounting for 37% globally.
215
Measured concentrations of crustal material (dust) vary by an order of magnitude from
216