Global Sources of Fine Particulate Matter: Interpretation of PM2.5

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


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


Page 2 of 35

Page 3 of 35

50 51


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



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


Page 4 of 35

Page 5 of 35

93


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



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


Page 6 of 35

Page 7 of 35


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



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


Page 8 of 35

Page 9 of 35


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



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

Global Sources of Fine Particulate Matter: Interpretation of PM2.5

Recommend Documents