Embed Size (px)
Citation preview
1
Global Chemical Composition of Ambient Fine Particulate Matter Estimated from 1
Satellite Observations and a Chemical Transport Model 2
Sajeev Philip1, Randall V. Martin
1, 2, Aaron van Donkelaar
1, Jason Wai-Ho Lo
1, Yuxuan Wang
3, 3
Dan Chen4, Lin Zhang
5, Prasad S. Kasibhatla
6, Siwen W. Wang
7, Qiang Zhang
7, Zifeng Lu
8, 4
David G. Streets8, Shabtai Bittman
9and Douglas J. Macdonald
10 5
1Department of Physics and Atmospheric Science, Dalhousie University, Halifax, Nova Scotia, 6
Canada 7
2Also at Harvard-Smithsonian Center for Astrophysics, Cambridge, Massachusetts, USA 8
3Ministry of Education Key Laboratory for Earth System Modeling, Center for Earth System 9
Science, Institute for Global Change Studies, Tsinghua University, Beijing, China 10
4Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, 11
California, USA. 12
5Department of Atmospheric and Oceanic Sciences, School of Physics, Peking University, China 13
6Nicholas School of the Environment and Earth Sciences, Duke University, Durham, North 14
Carolina, USA. 15
7State Key Joint Laboratory of Environment Simulation and Pollution Control, School of 16
Environment, Tsinghua University, Beijing, China 17
8Decision and Information Sciences Division, Argonne National Laboratory, Argonne, IL, USA 18
9Agriculture and Agri-Food Canada, Agassiz, British Columbia, Canada 19
10Environment Canada, Canada 20
Corresponding author: S. Philip, Department of Physics and Atmospheric Science, Dalhousie 21
University, Halifax, NS, B3H 4R2, Canada ([email protected]) 22
2
Key words: PM2.5, AOD, aerosol composition, exposure 23
Short title: Chemical composition of global fine particulate matter 24
Acknowledgements: This work was supported by Health Canada and the Natural Sciences and 25
Engineering Research Council of Canada. We thank the MODIS, MISR, CALIOP, GEOS-Chem, 26
IMPROVE, AQS, CASTNET, NAPS, CAPMoN, EMEP, EANET, and ACENET teams. 27
28
Abstract 29
Background: Epidemiologic and health impact studies are inhibited by the paucity of global, 30
long-term measurements of the chemical composition of fine particulate matter. 31
Objective: We fuse information from satellite observations and chemical transport model 32
simulations to estimate the chemical composition of global, long-term average, ground-level 33
PM2.5. 34
Methods: We inferred PM2.5 chemical composition at a spatial resolution of approximately 10 35
km x 10 km for 2004-2008 by combining aerosol optical depth retrieved from satellite 36
instruments, MODIS and MISR with coincident profile and composition information from a 37
global chemical transport model (GEOS-Chem). 38
Results: Evaluation of the satellite-model PM2.5 composition dataset with North American in situ 39
measurements indicated a significant spatial agreement for secondary inorganic aerosol, 40
particulate organic mass, black carbon, mineral dust and sea salt. We found that global 41
population-weighted PM2.5 concentrations are dominated by particulate organic mass (11 g/m3), 42
secondary inorganic components (8.6 g/m3), and mineral dust (7.6 g/m
3) components. 43
3
Secondary inorganic aerosol concentrations exceeded 30 g/m3
over East China and northern 44
India. The global population-weighted mean uncertainty (1 SD) is 4.4 g/m3 for particulate 45
organic mass, 3.0 g/m3
for both secondary inorganic aerosol and mineral dust, as estimated 46
from uncertainty in aerosol optical depth, accuracy of the simulated aerosol vertical profile and 47
composition, and sampling. 48
Conclusions: Global chemical transport model simulations combined with satellite-derived 49
total-column AOD observations provide constraints into the global chemical composition of 50
PM2.5. 51
52
Introduction 53
A large body of evidence has established that short-term human exposure to various chemical 54
constituents of fine particulate matter (PM) with aerodynamic diameter less than 2.5 g/m3 55
(PM2.5) causes adverse health effects including increased hospital admissions (e.g., Bell et al. 56
2009; Ito et al. 2011; Kim et al. 2012), cardiovascular, respiratory, and all-cause mortality (e.g., 57
Burnett et al. 2000; Ostro et al. 2010; Zhou et al. 2011; Cao et al. 2012; Son et al. 2012). 58
However, the health impacts of long-term exposure to PM2.5 chemical components are less well 59
understood, in contrast to the well-established relationship of the total PM2.5 mass with adverse 60
health effects (e.g., Pope et al. 2009; Brook et al. 2010; Lepeule et al. 2012). Epidemiologic and 61
health impact assessments of PM2.5 composition have been impeded by the paucity of long-term 62
measurements of PM2.5 composition. Spatial mapping of aerosol composition could help in 63
elucidating the health impacts of fine particulate matter components. 64
4
Satellite remote sensing for surface air quality has developed rapidly over the last decade (Martin 65
2008; Hoff and Christopher 2009). Aerosol optical depth (AOD), an optical measure of the 66
column integrated aerosol abundance in the atmosphere, can now be reliably retrieved from 67
satellite remote sensing over land. Several studies have demonstrated close relationships between 68
AOD and PM2.5 (e.g., Wang and Christopher 2003; Engel-Cox et al. 2004; Kloog et al. 2011) to 69
the point that AOD is being used for operational air quality forecasting (Al-Saadi et al. 2005, van 70
Donkelaar et al. 2012). However, the relation of AOD to PM2.5 is complex (Paciorek and Liu 71
2009) and current satellite remote sensing provides little information on the chemical 72
composition of PM2.5 (Liu et al. 2007). 73
Chemical transport models (CTMs) also have developed markedly over the last decade. Current 74
CTMs are capable of simulating the atmospheric distribution of aerosols and calculating the 75
local, coincident relationship of satellite AOD with ground-level PM2.5 concentration at a 76
regional (Liu et al. 2004) and global (van Donkelaar et al. 2010) scale. CTMs also offer the 77
capability to simulate the major chemical components of PM2.5 including secondary inorganic 78
aerosol (sulfate, nitrate, and ammonium), primary and secondary organic aerosol, black carbon, 79
mineral dust, sea salt, and aerosol water. These model developments provide information about 80
the relation of AOD with ground-level PM2.5 and its chemical composition. CTMs are also being 81
used to quantify source impacts on PM2.5 (e.g., Wang et al. 2009; Anenberg et al. 2011). 82
Scientific understanding of PM2.5 chemical composition has been closely coupled with advances 83
in measurements. For example, several in situ aerosol composition monitoring networks across 84
the U.S. and Canada routinely measure the major components of PM2.5 (e.g., Malm et al. 1994; 85
Dabek-Zlotorzynska et al. 2011). Numerous studies combined these in situ data to study the 86
5
spatial and temporal variation of PM2.5 chemical composition (e.g., Bell et al. 2007; Hand et al. 87
2011, 2012). Other established networks are the European Monitoring and Evaluation 88
Programme (EMEP; Europe; http://www.emep.int/) and the Acid Deposition Monitoring 89
Network in East Asia (EANET; East Asia; http://www.eanet.cc/) which measure some 90
components of PM2.5. These in situ measurements are too sparse to represent population 91
exposure in most regions of the world. However, they provide an opportunity to evaluate 92
satellite-model combined (hereafter, satellite-model) PM2.5 composition. 93
Here, we combined satellite remote sensing of AOD with global modeling of coincident aerosol 94
vertical profile and composition to produce a global long-term (2004-2008) mean ambient 95
satellite-model PM2.5 composition dataset at a spatial resolution of 0.1ºx 0.1
º, or approximately 10 96
km x 10 km at mid-latitudes. We evaluated this dataset with in situ measurements across North 97
America, and where available in the rest of the world. We subsequently estimated the various 98
emission sources of total PM2.5 mass. 99
Materials and Methods 100
Satellite AOD observations 101
We began with AOD retrievals from the two MODIS instruments onboard the Terra and Aqua 102
satellites and the MISR instrument onboard Terra. Aerosol retrievals (collection 5) from each 103
MODIS instrument provide near-daily global coverage of cloud-free regions at a resolution of 10 104
km x 10 km (Levy et al. 2007). The MISR retrieval algorithm uses multi-angle, multi-spectral 105
observations to provide aerosol optical properties at a spatial resolution of 18 km x 18 km and 106
global coverage of cloud-free regions within nine days (Kahn et al. 2007). The operational 107
6
MODIS and MISR retrievals together provide more reliable global AOD than from either 108
instrument alone (van Donkelaar et al. 2010). 109
Following van Donkelaar et al. (2010), we collected daily AOD retrievals of these three satellite 110
sensors from 2004 to 2008, and regridded them separately onto a resolution of 0.1º x 0.1
0. We 111
then divided the world into nine regions with distinct surface type based on the MODIS 112
BRDF/Albedo product (MOD43, Collection 5; Schaaf et al. 2002). We used all the available 113
ground-based sun photometer AOD measurements (Aerosol Robotic Network, AERONET, 114
Holben et al. 1998) over these regions to identify the average monthly bias of satellite AOD for 115
each region. We retained the daily satellite AOD observations with a local monthly bias less than 116
± (20% or 0.1). In addition, we retained the MODIS and MISR AOD with retrieved fine mode 117
fraction greater than 20% to reduce the influence of large particles. We included two textural 118
filters for MODIS AOD to reduce cloud contamination by excluding data with no adjacent 119
retrievals, and grids with AOD and coefficient of variation greater than 0.5 (Zhang and Reid 120
2006; Hyer et al. 2011). These three different (MODIS/Terra, MODIS/Aqua, MISR/Terra) 121
albedo-filtered, fine-mode-filtered AODs at 0.10
x 0.10
resolution are the major observational 122
inputs for this study. 123
Figure 1 shows the long-term (2004-2008) mean AOD from the MODIS and MISR satellite 124
instruments. Features from near ground-level PM2.5 components from anthropogenic and natural 125
sources are apparent. Enhancements exist over anthropogenic pollution sources of South and 126
East Asia, over mineral dust source regions of the Sahara, and over biomass burning regions of 127
South America, Central Africa, and Equatorial and Southeast Asia. 128
129
7
Inferring PM2.5 composition from AOD 130
Our approach to infer the 24-hr average ground-level dry PM2.5 concentrations of each chemical 131
component i from the observed aerosol optical depth, AODsat involved a chemical transport 132
model (GEOS-Chem) to calculate that relationship, 133
(1) 134
The major PM2.5 components include sulfate (SO42-
), nitrate (NO3-), ammonium (NH4
+), total 135
secondary inorganic aerosol (SIA, sum of SO42-
, NO3- and NH4
+ ions), particulate organic mass 136
(OM), black carbon (BC), mineral dust and sea salt. The subscript CTM indicates values from a 137
chemical transport model. The simulated conversion factor, defined as the ratio of components to 138
AOD, relates the observed AOD to the ground-level PM2.5 components. We used the GEOS-139
Chem global CTM (http://geos-chem.org) to calculate the local conversion factors coincident 140
with each satellite observation. The GEOS-Chem model simulates the temporal and three-141
dimensional spatial distributions of various aerosol components and gases using assimilated 142
meteorology and emission inventories as major inputs [see Supplemental Material]. 143
We conducted a global and three regional, higher resolution, simulations from 2004 to 2008, 144
rather than relying on previous coarse-resolution model simulations by van Donkelaar et al. 145
(2010). The top-left panel of Figure 2 shows the boundaries of the three regional simulations. 146
The aerosol dry mass composition of the lowest layer of the model was taken to represent the 147
ground-level PM2.5 composition. We averaged the simulated AOD between 1000 and 1200 hrs 148
local solar time to correspond with Terra overpass, and 1300 and 1500 hours local solar time to 149
correspond with Aqua overpass. We calculated the daily local Terra and Aqua conversion factors 150
ii CTMSat Sat
CTM
componentcomponent AOD
AOD
8
as the ratio of 24 hr average PM2.5 components to the corresponding AOD at satellite overpass 151
period. We subsequently interpolated these two daily conversion factors to a resolution of 0.10
x 152
0.10. 153
Figure 2 shows the long-term mean (2004-2008) simulated ratio of PM2.5 components to AOD. 154
The relative importance of various aerosol components to the total AOD over different regions of 155
the globe is represented by these maps. High values indicate regions with relatively large 156
abundance of that aerosol component near the ground. Non-hygroscopic components (e.g., 157
mineral dust) also exhibit high conversion factors due to the low contribution of aerosol water to 158
AOD. The high mass/AOD ratios for secondary inorganic, OM and mineral dust indicate that 159
these components are the dominant contributors to global AOD over land. Secondary inorganic 160
aerosols dominate over industrial regions. Particulate organic mass from biomass burning is the 161
primary contributor to AOD over the Amazon, Central Africa, Northern India and Oceania. 162
Mineral dust is the primary contributor to AOD over deserts. Black carbon is a small component 163
of AOD, but is more apparent in local hotspots. Sea salt generally has the lowest conversion 164
factor over land. 165
We applied equation (1) to produce PM2.5 from individual AOD observations from the two 166
MODIS and the MISR instruments from 2004 to 2008. We averaged the five-year composition 167
data from these three satellite instruments on a monthly basis, and capped the variation of the 168
long-term monthly mean composition from the simulated composition within a factor of three. 169
We then averaged the monthly mean data to obtain the long-term mean satellite-model combined 170
composition. We retained grids with at least 50 successful satellite observations which cover 171
99% of the global population (66% of the land grids). We accounted for sampling bias by 172
9
scaling this dataset with the ratio of annual mean model composition to the model sampled 173
coincidently with satellite observations following van Donkelaar et al. (2010). We calculated the 174
regional population exposure for the GBD regions (Global Diseases, Injuries, and Risk Factors 175
2010 study; Figure 1 shows the 21 GBD regions) using population data for 2005 (described in 176
Brauer et al. 2011). 177
We used continuous PM2.5 composition measurements from several networks over North 178
America, and annually representative composition measurements from Europe and East Asia and 179
from several published papers to evaluate our satellite-model dataset [see Supplemental 180
Material]. 181
Estimating the Error 182
The error associated with the satellite-model long-term mean PM2.5 composition arises from 183
three dominant sources, 1) AOD retrieval bias, 2) bias in simulating the component/AOD ratio, 184
and 3) incomplete sampling of the long-term mean. We used 20% to represent the AOD 185
retrieval bias, as we excluded regions with an expected AOD bias greater than that value. The 186
bias associated with the component/AOD ratio arises from the combination of errors in emission 187
estimates, meteorology, and atmospheric chemistry processes in the model. Quantification is 188
difficult due to the lack of coincident AOD and ground-level composition observations. A major 189
source of this error, however, is represented by the accuracy of simulated relative vertical profile. 190
Therefore, we calculated the relative model vertical profile error by comparison with Cloud-191
Aerosol LIDAR with Orthogonal Polarisation (CALIOP) data (Winker et al. 2007; van 192
Donkelaar et al. 2013), and apply the mean model ground-level relative composition (ratio of 193
PM2.5 components to total PM2.5 mass) error (1 SD) for each component in comparison with 194
10
North American in situ measurements. Sampling error is estimated as the difference between the 195
annual mean model values and the coincidently sampled simulated values. We combined in 196
quadrature the errors from the AOD retrieval, GEOS-Chem AOD to ground-level relative 197
composition, and incomplete sampling. We then scaled this error estimate such that 1 SD of the 198
error is consistent with the error in our product versus North American in situ measurements. 199
Estimating the PM2.5 emission sources 200
PM2.5 constituents arise from different emission source categories such as fossil fuel, biofuel and 201
biomass burning. Quantitative determination of these emission source sector‟s contribution to 202
PM2.5 can facilitate mitigation strategies. We therefore estimated the emission sources of total 203
PM2.5 using sensitivity simulation to exclude specific emission sectors. We calculated the 204
satellite-model PM2.5 mass, PM2.5jSat
from an emission sector “j” such as fossil fuel, biofuel, and 205
biomass burning: 206
(2) 207
The superscript, “Base” represents a simulation including all the emission sources, and “Sens” 208
refers to the sensitivity simulation excluding a specific emission sector. The term within the 209
parenthesis is the simulated ratio of PM2.5 from an emission sector to the total AOD. 210
Results 211
PM2.5 composition over North America 212
Figure 3 shows the satellite and ground-based observations of North American PM2.5 213
composition. The in situ observation of a large sulfate burden in the East is well reproduced in 214
Sat
Base
Sens
j
Base
Sat
j AODAOD
PMPMPM
5.25.2
5.2
11
the satellite-model product. Nitrate and ammonium are enhanced south of the Great Lakes where 215
intense agriculture sources of ammonia and weak sulfur sources contribute to excess ammonia 216
gas that is available for forming ammonium nitrate. The Californian nitrate enhancements are 217
under-predicted reflecting difficulties in representing this heterogeneous region (Walker et al. 218
2012; Heald et al. 2012). Together these secondary inorganic ions comprise a major fraction of 219
the total PM2.5 in the Eastern U.S. and California, reaching concentrations of approximately 10 220
g/m3. The spatial pattern of particulate organic mass over the southeastern U.S. particularly 221
from the biogenic sources is captured well in the satellite-model product. Black carbon 222
concentrations exhibit hotspots in industrial regions; performance elsewhere is more variable 223
given the stochastic nature of fires. Fine-mode dust and fine sea salt emissions are weak 224
contributors to PM2.5 mass throughout the continent with typical mass concentrations below 1 225
g/m3. The primary exception is for mineral dust over deserts in the southwest. 226
Figure 4 shows scatter plots of satellite-model PM2.5 components with North American ground-227
level measurements. The correlation between satellite-model sulfate and ground monitors is high 228
(r = 0.91, slope = 0.94). Concentrations are also well predicted for nitrate (r = 0.68, slope=1.04) 229
and ammonium (r = 0.85, slope = 1.01). The correlation for OM is weaker (r = 0.43, slope = 230
0.99) likely due to sampling differences for fires, and difficulty simulating fires and secondary 231
organic aerosol. Black carbon, mineral dust and sea salt have modest agreement with in situ 232
measurements with correlations of 0.56, 0.58 and 0.63 respectively. The bias for all components 233
is within 30%. The in situ observations are similarly correlated with the satellite-model product 234
and the GEOS-Chem simulation, with improvements over in the slope versus the GEOS-Chem 235
simulation for sulfate (0.82), nitrate (0.78), and ammonium (0.83). 236
12
Global PM2.5 composition 237
Figure 5 shows the global satellite-model estimate of long-term mean PM2.5 composition. 238
Secondary inorganic aerosol concentrations over East China exceed 30 g/m3 with half from 239
sulfate. The Indo-Gangetic Plain, China, and biomass burning regions of South America and 240
Central Africa are highlighted in the OM map. Hotspots of black carbon are most apparent in 241
China and the Indo-Gangetic Plain. Mineral dust is the largest contributor to PM2.5 over the 242
desert regions of Northern Africa and the Middle East with concentrations greater than 50 g/m3
243
over broad regions. 244
Table 1 compares the satellite-model composition with the global (non-North American) in situ 245
measurements, whose locations and values are also shown in Figure 5 [see Supplemental 246
Material]. The satellite-model product exhibits high correlations for secondary inorganic aerosol 247
(r = 0.94) and its components, with slopes near unity. Carbonaceous aerosols are again less well 248
represented; sparse in situ monitors may play a role for fires. The satellite-model product 249
outperforms the pure model simulation for all components (e.g., for secondary inorganic aerosols 250
the slope improved from 0.65 to 0.93; for organic matter the correlation improved from 0.61 to 251
0.70). 252
Errors in the PM2.5 composition 253
Figure 6 shows the estimated absolute uncertainty in the satellite-model composition. Secondary 254
inorganic aerosols have the lowest relative uncertainty, ± (35% + 0.5 g/m3). Particulate organic 255
mass and black carbon have uncertainty of ± (40% + 0.5 g/m3) and ± (50% + 0.2 g/m
3) 256
respectively. Mineral dust ± (40% + 0.2 g/m3) and sea salt ± (85% + 0.1 g/m
3) exhibit large 257
13
uncertainty, but this occurs primarily in unpopulated regions. The global population-weighted 258
uncertainty for the dominant components are 4.4 g/m3 for particulate organic mass, 3.0 g/m
3 259
for mineral dust, and 3.0 g/m3
for secondary inorganic aerosol. 260
Estimate of PM2.5 from emission sources 261
Figure 7 shows the contributions to PM2.5 from fossil fuel combustion, biofuel combustion, and 262
biomass burning. The other category is dominated by mineral dust. Enhanced anthropogenic 263
sources are apparent over the industrial and populated regions. Biofuel sources over Asia reflect 264
open cooking. Biomass burning dominates in Central Africa and the Amazon. Other sources are 265
dominated by mineral dust with large enhancements over and downwind of major deserts. 266
Population exposure to outdoor ambient PM2.5 267
Table 2 summarizes the global and regional statistics of population exposure to outdoor ambient 268
PM2.5 composition. The dominant population-weighted component is particulate organic mass 269
with a global mean concentration of 11 g/m3. This is followed by secondary inorganic 270
components (8.6 g/m3), mineral dust (7.6 g/m
3), and black carbon (2.3 g/m
3). The secondary 271
inorganic aerosol is dominated by sulfate (4.7 g/m3), followed by ammonium (2.1 g/m
3) and 272
nitrate (1.7 g/m3) concentrations. 273
On a regional scale, secondary inorganic aerosol concentrations are noteworthy in East Asia (22 274
g/m3) and South Asia (8.0 g/m
3). Mineral dust concentrations exceed 15 g/m
3 in the Middle 275
East, North Africa, and West Africa. Mean outdoor ambient particulate organic mass 276
concentrations are 21 g/m3
in South Asia and 20 g/m3
in East Asia. 277
14
Population-weighted PM2.5 is dominated by fossil fuel combustion (15 g/m3) followed by 278
biofuel combustion (10 g/m3) and biomass burning (1.3 g/m
3). PM2.5 is high from emissions 279
from fossil fuel over East Asia (40 g/m3), from biofuel over East Asia (21 g/m
3) and South 280
Asia (19 g/m3), and from biomass emissions over Central Africa (14 g/m
3). 281
Discussion 282
The chemical composition of ambient ground-level fine particulate mass is of relevance for 283
epidemiological and health impact studies (Health Effects Institute 2004, National Research 284
Council 2004; U.S. Environmental Protection Agency 2009; Bell et al. 2012), but few 285
measurements exist in many regions of the world. Satellite remote sensing offers valuable 286
information on the total integrated column abundance of aerosol in the atmosphere but limited 287
information on its ground-level chemical composition. Recent advances in chemical transport 288
models provide accurate estimates of the major chemical components present in fine particulate 289
mass. Together, satellite observations combined with chemical transport model simulations 290
provide insight into the global abundance of PM2.5 chemical composition. 291
In our study, we developed a method to map fine particulate components globally using MODIS 292
and MISR satellite observations and the GEOS-Chem chemical transport model. We produced a 293
long-term (2004-2008) mean global map of PM2.5 components at a spatial resolution 0.10
x 0.10 294
and examined the population exposure to PM2.5 composition. These estimates should facilitate 295
studies of long-term exposure to PM2.5 chemical components in regions of the world with 296
insufficient ground-based monitors. Our estimates suggest that particulate organic mass is the 297
dominant form of outdoor ambient PM2.5 with global population-weighted concentrations of 11 298
g/m3. Other major contributors to global population-weighted PM2.5 mass are secondary 299
15
inorganic components (8.6 g/m3) and mineral dust (7.6 g/m
3). We also identified regions of 300
concern. For example, secondary inorganic aerosol and particulate organic mass concentrations 301
exceed 30 g/m3
over parts of East China and northern India. We additionally estimated the 302
different emission sources contributing to PM2.5 mass. These estimates should enhance further 303
studies to determine the association between long-term exposure to PM2.5 components and their 304
adverse health effects. 305
To our knowledge, this is the first study to assess the global exposure to all the major PM2.5 306
chemical components. Evaluation of our dataset with North American in situ measurements 307
shows significant agreement for all major chemical components. The best performance is for 308
secondary inorganic aerosol (r = 0.86, slope = 1.05, n = 419). The weakest performance versus 309
particulate organic mass (r = 0.43, slope = 0.99, n = 336) reflects uncertainties in model 310
representation of secondary organic aerosol formation and biomass burning sources, and 311
differences in sampling fires. Subcomponents of secondary inorganic aerosol also exhibits high 312
correlation with in situ measurements (r = 0.91 for sulfate, r = 0.68 for nitrate, r = 0.85 for 313
ammonium). 314
We estimated the uncertainty in the global PM2.5 composition through error propagation and 315
comparison with independent observations. Major sources of uncertainty are the retrieval of 316
aerosol optical depth, accuracy of the aerosol vertical profile and composition simulation, and 317
sampling. The global population-weighted mean uncertainty (1 SD) is 4.4 g/m3 for particulate 318
organic mass, 3.0 g/m3
for mineral dust, and 3.0 g/m3
for secondary inorganic aerosol. 319
Multiple opportunities exist to improve the estimates. Advances in satellite remote sensing 320
(Mishchenko et al. 2007) could yield more observational information on aerosol components. 321
16
Future developments in the modeling of aerosol composition such as organic mass could 322
improve the estimates. Other emerging sources of information on the sources of aerosol 323
precursors include satellite observations of tracer gases (Streets et al. 2013) such as NO2 324
(Boersma et al. 2011), SO2 (Lee et al. 2011), and NH3 (Clarisse et al. 2009). Assimilation of 325
these components into a chemical transport model would provide additional constraints on 326
aerosol composition. Simulations of aerosol microphysics could improve the AOD/composition 327
estimate. Trace metals are an important aerosol component that should be added as their 328
simulation capability improves. 329
330
331
332
333
334
335
336
337
338
339
340
17
References 341
Al-Saadi J, Szykman J, Pierce RB, Kittaka C, Neil D, Chu DA et al. 2005. Improving national air 342
quality forecasts with satellite aerosol observations. Bull Am Meteorol Soc 86(9); doi: 343
10.1175/BAMS-86-9-1249. 344
Anenberg SC, Talgo K, Arunachalam S, Dolwick P, Jang C, West JJ. 2011. Impacts of global, 345
regional, and sectoral black carbon emission reductions on surface air quality and human 346
mortality. Atmospheric Chemistry and Physics 11(14):7253-7267; doi: 10.5194/acp-11-7253-347
2011. 348
Bell ML, Dominici F, Ebisu K, Zeger SL, Samet JM. 2007. Spatial and temporal variation in 349
PM2.5 chemical composition in the united states for health effects studies. Environ Health 350
Perspect 115(7); doi: 10.1289/ehp.9621. 351
Bell ML, Ebisu K, Peng RD, Samet JM, Dominici F. 2009. Hospital admissions and chemical 352
composition of fine particle air pollution. American Journal of Respiratory and Critical Care 353
Medicine 179(12); doi: 10.1164/rccm.200808-1240OC. 354
Bell ML, HEI Health Review Committee. 2012. Assessment of the health impacts of particulate 355
matter characteristics. Res Rep Health Eff Inst(161):5-38. 356
Boersma KF, Eskes HJ, Dirksen RJ, van der A RJ, Veefkind JP, Stammes P et al. 2011. An 357
improved tropospheric NO2 column retrieval algorithm for the ozone monitoring instrument. 358
Atmospheric Measurement Techniques 4(9):1905-1928; doi: 10.5194/amt-4-1905-2011. 359
Brauer M, Amann M, Burnett RT, Cohen A, Dentener F, Ezzati M et al. 2012. Exposure 360
assessment for estimation of the global burden of disease attributable to outdoor air pollution. 361
Environ Sci Technol 46(2); doi: 10.1021/es2025752. 362
Brook RD, Rajagopalan S, Pope CA,III, Brook JR, Bhatnagar A, Diez-Roux AV et al. 2010. 363
Particulate matter air pollution and cardiovascular disease an update to the scientific statement 364
from the american heart association. Circulation 121(21); doi: 365
10.1161/CIR.0b013e3181dbece1. 366
Burnett RT, Brook J, Dann T, Delocla C, Philips O, Cakmak S et al. 2000. Association between 367
particulate- and gas-phase components of urban air pollution and daily mortality in eight 368
canadian cities. Inhal Toxicol 12; doi: 10.1080/08958370050164851. 369
18
Cao J, Xu H, Xu Q, Chen B, Kan H. 2012. Fine particulate matter constituents and 370
cardiopulmonary mortality in a heavily polluted chinese city. Environ Health Perspect 120(3); 371
doi: 10.1289/ehp.1103671. 372
Clarisse L, Clerbaux C, Dentener F, Hurtmans D, Coheur P. 2009. Global ammonia distribution 373
derived from infrared satellite observations. Nature Geoscience 2(7):479-483; doi: 374
10.1038/ngeo551. 375
Dabek-Zlotorzynska E, Dann TF, Martinelango PK, Celo V, Brook JR, Mathieu D et al. 2011. 376
Canadian national air pollution surveillance (NAPS) PM2.5 speciation program: Methodology 377
and PM2.5 chemical composition for the years 2003-2008. Atmos Environ 45(3); doi: 378
10.1016/j.atmosenv.2010.10.024. 379
Engel-Cox JA, Holloman CH, Coutant BW, Hoff RM. 2004. Qualitative and quantitative 380
evaluation of MODIS satellite sensor data for regional and urban scale air quality. Atmos 381
Environ 38(16); doi: 10.1016/j.atmosenv.2004.01.039. 382
Hand JL, et al. 2011. IMPROVE (Interagency Monitoring of Protected Visual Environments): 383
Spatial and seasonal patterns and temporal variability of haze and its constituents in the 384
United States, Rep. V, Coop. Inst. For Res. In the Atmos., Fort Collins, Colo. [Available at 385
http:// vista.cira.colostate.edu/improve/Publications/Reports/2011/2011.htm.] 386
Hand JL, Schichtel BA, Pitchford M, Malm WC, Frank NH. 2012. Seasonal composition of 387
remote and urban fine particulate matter in the united states. Journal of Geophysical 388
Research-Atmospheres 117:D05209; doi: 10.1029/2011JD017122. 389
Heald CL, Collett JL,Jr., Lee T, Benedict KB, Schwandner FM, Li Y et al. 2012. Atmospheric 390
ammonia and particulate inorganic nitrogen over the united states. Atmospheric Chemistry 391
and Physics 12(21):10295-10312; doi: 10.5194/acp-12-10295-2012. 392
Health Effects Institute. 2004. Understanding the Health Effects of Components of the Particulate 393
Matter Mix: Progress and Next Steps. 2004, Boston, MA: Health Effects Institute. 394
Hoff RM, Christopher SA. 2009. Remote sensing of particulate pollution from space: Have we 395
reached the promised land? J Air Waste Manage Assoc 59(6); doi: 10.3155/1047-396
3289.59.6.645. 397
19
Holben BN, Eck TF, Slutsker I, Tanre D, Buis JP, Setzer A et al. 1998. AERONET - A federated 398
instrument network and data archive for aerosol characterization. Remote Sens Environ 66(1); 399
doi: 10.1016/S0034-4257(98)00031-5. 400
Hyer EJ, Reid JS, Zhang J. 2011. An over-land aerosol optical depth data set for data assimilation 401
by filtering, correction, and aggregation of MODIS collection 5 optical depth retrievals. 402
Atmospheric Measurement Techniques 4(3):379-408; doi: 10.5194/amt-4-379-2011. 403
Ito K, Mathes R, Ross Z, Nadas A, Thurston G, Matte T. 2011. Fine particulate matter constituents 404
associated with cardiovascular hospitalizations and mortality in new york city. Environ Health 405
Perspect 119(4); doi: 10.1289/ehp.1002667. 406
Kahn RA, Li W-, Moroney C, Diner DJ, Martonchik JV, Fishbein E. 2007. Aerosol source plume 407
physical characteristics from space-based multiangle imaging. Journal of Geophysical 408
Research-Atmospheres 112(D11):D11205; doi: 10.1029/2006JD007647. 409
Kim S, Peel JL, Hannigan MP, Dutton SJ, Sheppard L, Clark ML et al. 2012. The temporal lag 410
structure of short-term associations of fine particulate matter chemical constituents and 411
cardiovascular and respiratory hospitalizations. Environ Health Perspect 120(8). 412
Kloog I, Koutrakis P, Coull BA, Lee HJ, Schwartz J. 2011. Assessing temporally and spatially 413
resolved PM2.5 exposures for epidemiological studies using satellite aerosol optical depth 414
measurements. Atmos Environ 45(35); doi: 10.1016/j.atmosenv.2011.08.066. 415
Lee C, Martin RV, van Donkelaar A, Lee H, Dickerson RR, Hains JC et al. 2011. SO2 emissions 416
and lifetimes: Estimates from inverse modeling using in situ and global, space-based 417
(SCIAMACHY and OMI) observations. Journal of Geophysical Research-Atmospheres 418
116:D06304; doi: 10.1029/2010JD014758. 419
Lepeule J, Laden F, Dockery D, Schwartz J. 2012. Chronic exposure to fine particles and 420
mortality: An extended follow-up of the harvard six cities study from 1974 to 2009. Environ 421
Health Perspect 120(7); doi: 10.1289/ehp.1104660. 422
Levy RC, Remer LA, Mattoo S, Vermote EF, Kaufman YJ. 2007. Second-generation operational 423
algorithm: Retrieval of aerosol properties over land from inversion of moderate resolution 424
imaging spectroradiometer spectral reflectance. Journal of Geophysical Research-425
Atmospheres 112(D13):D13211; doi: 10.1029/2006JD007811. 426
20
Liu Y, Park RJ, Jacob DJ, Li QB, Kilaru V, Sarnat JA. 2004. Mapping annual mean ground-level 427
PM2.5 concentrations using multiangle imaging spectroradiometer aerosol optical thickness 428
over the contiguous united states. Journal of Geophysical Research-Atmospheres 429
109(D22):D22206; doi: 10.1029/2004JD005025. 430
Liu Y, Koutrakis P, Kahn R. 2007. Estimating fine particulate matter component concentrations 431
and size distributions using satellite-retrieved fractional aerosol optical depth: Part 1 - method 432
development. J Air Waste Manage Assoc 57(11); doi: 10.3155/1047-3289.S7.11.1351. 433
Malm WC, Sisler JF, Huffman D, Eldred RA, Cahill TA. 1994. Spatial and seasonal trends in 434
particle concentration and optical extinction in the united-states. Journal of Geophysical 435
Research-Atmospheres 99(D1); doi: 10.1029/93JD02916. 436
Martin RV. 2008. Satellite remote sensing of surface air quality. Atmos Environ 42(34); doi: 437
10.1016/j.atmosenv.2008.07.018. 438
Mishchenko MI, Cairns B, Kopp G, Schueler CF, Fafaul BA, Hansen JE et al. 2007. Accurate 439
monitoring of terrestrial aerosols and total solar irradiance - introducing the glory mission. 440
Bull Am Meteorol Soc 88(5):677-+; doi: 10.1175/BAMS-88-5-677. 441
National Research Council 2004. Research Priorities for Airborne Particulate Matter: IV. 442
Continuing Research Progress. Committee on Research Priorities for Airborne Particulate 443
Matter. Washington, DC: National Academies Press. 444
Ostro B, Lipsett M, Reynolds P, Goldberg D, Hertz A, Garcia C et al. 2010. Long-term exposure 445
to constituents of fine particulate air pollution and mortality: Results from the california 446
teachers study. Environ Health Perspect 118(3); doi: 10.1289/ehp.0901181. 447
Paciorek CJ, Liu Y. 2009. Limitations of remotely sensed aerosol as a spatial proxy for fine 448
particulate matter. Environ Health Perspect 117(6); doi: 10.1289/ehp.0800360. 449
Pope CA,III, Ezzati M, Dockery DW. 2009. Fine-particulate air pollution and life expectancy in 450
the united states. N Engl J Med 360(4); doi: 10.1056/NEJMsa0805646. 451
Schaaf CB, Gao F, Strahler AH, Lucht W, Li XW, Tsang T et al. 2002. First operational BRDF, 452
albedo nadir reflectance products from MODIS. Remote Sens Environ 83(1-2):PII S0034-453
4257(02)00091-3; doi: 10.1016/S0034-4257(02)00091-3. 454
21
Son J, Lee J, Kim K, Jung K, Bell ML. 2012. Characterization of fine particulate matter and 455
associations between particulate chemical constituents and mortality in seoul, korea. Environ 456
Health Perspect 120(6):872-878; doi: 10.1289/ehp.1104316. 457
Streets DG, Canty T et al. 2013. Emissions estimation from satellite retrievals: A review of current 458
capability. Atmos Environ, doi:10.1016/j.atmosenv.2013.05.051. 459
U.S. EPA. Integrated Science Assessment for Particulate Matter (Final Report). U.S. 460
Environmental Protection Agency, Washington, DC, EPA/600/R-08/139F, 2009. 461
van Donkelaar A, Martin RV, et al. 2013. Optimal estimation for global ground-level fine 462
particulate matter concentrations. Journal of Geophysical Research-Atmospheres, Accepted, 463
doi: 10.1002/jgrd.50479. 464
van Donkelaar A, Martin RV, Pasch AN, Szykman JJ, Zhang L, Wang YX et al. 2012. Improving 465
the accuracy of daily satellite-derived ground-level fine aerosol concentration estimates for 466
north america. Environ Sci Technol 46(21):11971-11978; doi: 10.1021/es3025319. 467
van Donkelaar A, Martin RV, Brauer M, Kahn R, Levy R, Verduzco C et al. 2010. Global 468
estimates of ambient fine particulate matter concentrations from satellite-based aerosol optical 469
depth: Development and application. Environ Health Perspect 118(6); doi: 470
10.1289/ehp.0901623. 471
Walker JM, Philip S, Martin RV, Seinfeld JH. 2012. Simulation of nitrate, sulfate, and ammonium 472
aerosols over the united states. Atmospheric Chemistry and Physics 12(22):11213-11227; doi: 473
10.5194/acp-12-11213-2012. 474
Wang J, Christopher SA. 2003. Intercomparison between satellite-derived aerosol optical 475
thickness and PM2.5 mass: Implications for air quality studies. Geophys Res Lett 476
30(21):2095; doi: 10.1029/2003GL018174. 477
Wang ZS, Chien C, Tonnesen GS. 2009. Development of a tagged species source apportionment 478
algorithm to characterize three-dimensional transport and transformation of precursors and 479
secondary pollutants. Journal of Geophysical Research-Atmospheres 114:D21206; doi: 480
10.1029/2008JD010846. 481
Winker DM, Hunt WH, McGill MJ. 2007. Initial performance assessment of CALIOP. Geophys 482
Res Lett 34(19):L19803; doi: 10.1029/2007GL030135. 483
22
Zhang J, Reid JS. 2006. MODIS aerosol product analysis for data assimilation: Assessment of 484
over-ocean level 2 aerosol optical thickness retrievals. Journal of Geophysical Research-485
Atmospheres 111(D22):D22207; doi: 10.1029/2005JD006898. 486
Zhou J, Ito K, Lall R, Lippmann M, Thurston G. 2011. Time-series analysis of mortality effects of 487
fine particulate matter components in detroit and seattle. Environ Health Perspect 119(4); doi: 488
10.1289/ehp.1002613. 489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
23
Table 1: Comparison of satellite-model versus observed global (non-North American) PM2.5 507
composition 508
PM2.5 composition R Slope Offset(g/m3) N
Sulfate 0.94 0.89 -0.04 54
Nitrate 0.69 0.85 -0.36 54
Ammonium 0.91 1.10 -0.20 54
Total Secondary inorganic ions 0.91 0.93 -0.67 54
Particulate Organic Matter 0.70 0.44 -1.08 56
Black Carbon 0.62 1.06 -1.83 65
509
Correlation statistics are calculated with the reduced major axis regression. 510
511
512
513
514
515
516
517
518
519
520
24
Table 2: Population-weighted regional PM2.5 composition 521
522
Note: Figure 1 shows the borders of GBD (Global Diseases, Injuries, and Risk Factors 2010 523
study) regions. SD is standard deviation 524
Po
pu
lati
on
-wei
gh
ted
sta
tist
ics
(ug
/m3
)
PM
2.5
P
M2
.5
PM
2.5
PM
2.5
Po
pu
lati
on
Reg
ion
SO
42
-
N
O3
-
N
H4
+
S
IA
O
M
BC
D
ust
Sea
Sa
lt
F
oss
il f
uel
B
iofu
el
B
iom
ass
Oth
er(%
)
Mea
nS
DM
ean
SD
Mea
nS
DM
ean
SD
Mea
nS
DM
ean
SD
Mea
nS
DM
ean
SD
Mea
nS
DM
ean
SD
Mea
nS
DM
ean
SD
Wo
rld
4.7
4.6
1.7
2.6
2.1
2.3
8.6
9.2
11
.01
1.7
2.3
2.6
7.6
10
.10
.40
.41
4.6
17
.71
0.0
13
.31
.32
.91
0.2
12
.11
00
.0
Asi
a P
aci
fic
4.2
1.4
1.2
0.9
1.6
0.8
6.9
3.0
4.6
2.7
2.0
1.1
2.8
1.1
0.9
0.2
15
.77
.12
.21
.20
.31
.54
.81
.22
.7
Asi
a C
entr
al
2.3
0.7
0.5
0.4
0.9
0.3
3.7
1.2
3.7
1.7
0.5
0.2
12
.14
.70
.10
.14
.41
.33
.92
.60
.30
.21
2.5
4.6
1.3
Asi
a E
ast
11
.35
.15
.13
.55
.42
.82
1.7
11
.02
0.3
10
.65
.42
.88
.77
.10
.40
.54
0.1
22
.52
0.7
16
.10
.30
.31
0.6
7.3
21
.2
Asi
a S
ou
th5
.12
.10
.90
.72
.11
.08
.03
.62
0.6
13
.03
.52
.29
.26
.00
.40
.31
1.1
5.0
18
.61
3.9
0.5
0.3
12
.68
.82
2.9
Asi
a S
ou
th E
ast
2.5
2.1
0.2
0.2
0.8
0.8
3.4
3.0
6.1
4.2
0.9
0.7
0.9
0.9
0.8
0.4
6.9
7.1
5.5
4.9
3.6
5.6
3.6
1.5
8.7
Au
stra
lasi
a0
.40
.20
.10
.10
.10
.10
.60
.30
.50
.30
.10
.10
.60
.50
.70
.20
.80
.70
.20
.20
.20
.22
.41
.20
.4
Ca
rib
bea
n1
.00
.30
.20
.10
.20
.11
.30
.40
.70
.30
.10
.03
.71
.01
.40
.31
.20
.60
.10
.10
.20
.14
.31
.20
.5
Eu
rop
e C
entr
al
3.5
1.0
3.3
1.3
2.3
0.5
9.1
2.2
5.0
1.3
1.1
0.4
2.7
1.2
0.3
0.1
13
.03
.65
.21
.80
.50
.22
.61
.21
.9
Eu
rop
e E
ast
ern
2.7
0.7
2.3
1.2
1.6
0.6
6.5
2.3
3.5
1.4
0.6
0.3
2.3
1.5
0.2
0.1
9.0
3.4
3.6
1.9
0.9
0.7
1.9
2.2
3.3
Eu
rop
e W
este
rn2
.00
.72
.91
.71
.50
.66
.42
.52
.20
.90
.90
.42
.62
.80
.70
.39
.94
.91
.61
.10
.30
.53
.73
.56
.1
La
tin
Am
eric
a A
nd
ean
1.2
0.8
0.0
0.0
0.3
0.1
1.5
0.9
1.7
1.3
0.2
0.1
0.1
0.0
0.3
0.3
6.2
5.3
1.5
1.1
4.6
3.6
2.8
1.4
0.8
La
tin
Am
eric
a C
entr
al
2.1
1.9
0.3
0.3
0.7
0.6
3.1
2.8
1.7
0.7
0.3
0.3
0.9
0.9
0.5
0.4
5.8
4.2
1.0
1.4
1.7
1.1
3.4
1.9
3.4
La
tin
Am
eric
a S
ou
ther
n0
.90
.60
.10
.10
.20
.11
.20
.71
.91
.00
.30
.21
.11
.00
.30
.22
.72
.71
.11
.00
.81
.03
.92
.20
.9
La
tin
Am
eric
a t
rop
ica
l0
.70
.40
.10
.10
.20
.20
.90
.62
.51
.70
.30
.30
.10
.20
.40
.31
.81
.61
.41
.31
.62
.71
.30
.62
.9
No
rth
Afr
ica
/Mid
dle
Ea
st2
.61
.00
.30
.40
.90
.43
.81
.61
.90
.60
.40
.21
9.9
11
.10
.40
.34
.71
.90
.70
.40
.20
.12
6.0
15
.66
.4
No
rth
Am
eric
a3
.01
.31
.31
.01
.40
.75
.72
.73
.41
.40
.80
.60
.70
.40
.40
.41
1.1
5.4
0.9
0.4
0.6
0.3
1.5
0.9
5.0
Oce
an
ia0
.30
.20
.00
.00
.00
.00
.30
.20
.10
.10
.00
.00
.00
.00
.40
.30
.20
.20
.10
.20
.10
.12
.20
.70
.1
Su
b-S
ah
ara
n A
fric
a C
entr
al
0.7
0.4
0.0
0.0
0.2
0.1
1.0
0.6
8.0
2.2
0.5
0.2
2.7
3.1
0.1
0.1
1.3
1.3
4.2
3.6
13
.95
.66
.05
.21
.3
Su
b-S
ah
ara
n A
fric
a E
ast
0.5
0.3
0.0
0.0
0.1
0.1
0.7
0.4
2.9
1.7
0.3
0.2
5.7
10
.40
.30
.30
.80
.63
.43
.32
.52
.78
.71
3.6
4.8
Su
b-S
ah
ara
n A
fric
a S
ou
ther
n1
.10
.60
.10
.10
.40
.21
.60
.92
.51
.10
.30
.20
.40
.30
.40
.22
.92
.20
.90
.52
.51
.51
.50
.71
.0
Su
b-S
ah
ara
n A
fric
a W
est
0.9
0.6
0.1
0.1
0.3
0.3
1.3
0.9
4.2
1.9
0.3
0.1
30
.71
9.9
0.2
0.2
1.7
1.4
1.9
1.1
4.2
3.1
38
.41
8.8
4.4
25
Figure Legends 525
Figure 1: Combined aerosol optical depth (AOD) from the MODIS and MISR satellite 526
instruments for 2004-2008. Gray denotes water or < 50 successful satellite observations. The 527
thick border lines represent the GBD (Global Diseases, Injuries, and Risk Factors 2010 study) 528
regions. 529
Figure 2: Mean ratio of PM2.5 composition to AOD for 2004-2008. PM2.5 composition is 530
represented as dry mass. SIA is the sum of SO42-,
NO3- and NH4
+ ions. OM denotes particulate 531
organic mass. BC is black carbon. Gray space denotes water. The top-left panel contains the 532
boundaries of the three nested GEOS-Chem regions. 533
Figure 3: PM2.5 composition from satellite and ground-based measurements across North 534
America. PM2.5 composition is represented as dry mass. Gray denotes water or no in situ 535
measurements. 536
Figure 4: Comparison of satellite-model versus observed North American PM2.5 composition. 537
The solid black line is 1:1 line, dashed line is the best fit line, and dotted lines are the 1 SD error 538
lines. The inset contains correlation statistics calculated with reduced major axis regression. 539
Figure 5: Satellite-model global long-term mean (2004-2008) PM2.5 composition. In situ 540
observations are overlaid as colored circles. PM2.5 composition is represented as dry mass. Gray 541
denotes water. 542
Figure 6: Absolute uncertainty (1 SD) of satellite-model PM2.5 composition. Gray denotes water. 543
Figure 7: Estimate of emission sources contributing to PM2.5. 544
26
Figure 1. 545
546
547
548
549
550
551
552
553
554
555
556
557
27
Figure 2. 558
559
560
561
562
563
564
565
28
Figure 3. 566
567
29
Figure 4. 568
569
570
571
572
573
574
575
576
577
578
30
Figure 5. 579
580
581
582
583
584
585
586
31
Figure 6. 587
588
589
590
591
592
593
594
32
Figure 7. 595
596
597
598
599
600
601
602
603
604
605
606
607
608
33
Supplemental Material: Global Chemical Composition of Ambient Fine Particulate Matter 609
Estimated from Satellite Observations and a Chemical Transport Model 610
Sajeev Philip1, Randall V. Martin
1, 2, Aaron van Donkelaar
1, Jason Wai-Ho Lo
1, Yuxuan Wang
3, 611
Dan Chen4, Lin Zhang
5, Prasad S. Kasibhatla
6, Siwen W. Wang
7, Qiang Zhang
7, Zifeng Lu
8, 612
David G. Streets8, Shabtai Bittman
9and Douglas J. Macdonald
10 613
1Department of Physics and Atmospheric Science, Dalhousie University, Halifax, Nova Scotia, 614
Canada 615
2Also at Harvard-Smithsonian Center for Astrophysics, Cambridge, Massachusetts, USA 616
3Ministry of Education Key Laboratory for Earth System Modeling, Center for Earth System 617
Science, Institute for Global Change Studies, Tsinghua University, Beijing, China 618
4Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, 619
California, USA. 620
5Department of Atmospheric and Oceanic Sciences, School of Physics, Peking University, China 621
6Nicholas School of the Environment and Earth Sciences, Duke University, Durham, North 622
Carolina, USA. 623
7State Key Joint Laboratory of Environment Simulation and Pollution Control, School of 624
Environment, Tsinghua University, Beijing, China 625
8Decision and Information Sciences Division, Argonne National Laboratory, Argonne, IL, USA 626
9Agriculture and Agri-Food Canada, Agassiz, British Columbia, Canada 627
10Environment Canada, Canada 628
Corresponding author: S. Philip, Department of Physics and Atmospheric Science, Dalhousie 629
University, Halifax, NS, B3H 4R2, Canada ([email protected]) 630
34
Description of the GEOS-Chem aerosol simulation 631
We used the GEOS-Chem global three-dimensional chemical transport model (version 9-01-03; 632
http://geos-chem.org) to calculate the local conversion factors (ratio of component to AOD) 633
coincident with each satellite observation of AOD. The GEOS-Chem uses assimilated 634
meteorological data from the Goddard Earth Observing System (GEOS-5) at the NASA Global 635
Modeling Assimilation Office (GMAO). The meteorological data includes instantaneous fields, 636
surface variables (e.g., mixed layer depth) at a temporal resolution of 3 hours, and other variables 637
at 6 hours. We reduced the stratospheric grid boxes of the native GEOS-5 vertical grids (72 638
hybrid eta levels) for computational expediency. The vertical layers of the current model extend 639
from the Earth‟s surface to the top of the atmosphere (0.01 hPa) with 47 vertical levels. The 640
lowest layer of the model is centered at approximately 70 meters, and used here to represent the 641
ground-level aerosol concentrations. 642
The native horizontal resolution for the GEOS-5 meteorological data is at 0.50
x 0.6670. First, we 643
regridded the native resolution to a coarser resolution of 20
x 2.50 for computational expediency, 644
and performed the global simulation at this resolution. Second, we conducted three regional 645
(nested) simulations at the native horizontal resolution of 0.50
x 0.6670 for three regions of the 646
globe: North America, Europe, and East Asia. The global simulation outputs were used as 647
boundary conditions for the regional grids. The top left panel of Figure 2 provides the boundaries 648
of these regions. This high resolution simulation preserves the finer spatial patterns of the 649
chemical components (Chen et al. 2009; van Donkelaar et al. 2012). We spun up the model for 650
one month before each global and regional simulation to remove the effects of initial conditions 651
to the aerosol simulation. The dynamical processes (transport and convection) have a temporal 652
35
resolution of 10 minutes for the nested simulations and 15 minutes for the global simulation. We 653
used a timestep of 60 minutes for chemical processes and emissions for both nested and global 654
resolutions. We use full mixing of PM components below the boundary layer, with a correction 655
to the GEOS-5 predicted nocturnal mixed layer depth (Heald et al. 2012; Walker et al. 2012). 656
GEOS-Chem contains a detailed simulation of HOx-NOx-VOC-ozone-aerosol chemistry (Bey et 657
al. 2001; Park et al. 2004). The simulation of secondary inorganic ions is directly coupled with 658
gas phase chemistry (Park et al. 2004). Aerosol-gas interactions are simulated through 659
heterogeneous chemistry (Jacob, 2000) with updated aerosol uptake of N2O5 (Evans and Jacob 660
2005) and HO2 (Thornton et al. 2008), and aerosol extinction effects on photolysis rates (Martin 661
et al. 2003). The ISORROPIA II thermodynamic scheme is used to partition aerosols from gas 662
(Fontoukis and Nenes 2007) as implemented by Pye et al. (2009). GEOS-Chem uses in-cloud 663
sulfate formation using the cloud liquid water content and cloud volume fractions of the GEOS-5 664
data (Fischer et al. 2011). We artificially limited the nitric acid to two thirds of its value for each 665
timestep to correct for an overestimation in HNO3 found in comparison with measurements over 666
the eastern U.S. (Heald et al. 2012). GEOS-Chem calculates AOD based on the relative humidity 667
dependent aerosol optical properties (Martin et al. 2003) with an updated growth factor for 668
organic matter, and updates to the dust and ammonium sulfate optics by Ridley et al. (2012). 669
The GEOS-Chem simulation uses emission inventories of aerosol and its precursor gases as 670
input. We used regional anthropogenic emission inventories of NOx and SO2 over Canada (CAC; 671
http://www.ec.gc.ca/inrp‐npri/), the U.S. (Environmental protection Agency-National Emissions 672
Inventory 2005; http://www.epa.gov/ttnchie1/net/2005inventory.html), Mexico (BRAVO; Kuhns 673
et al. 2005), Europe (EMEP; http://www.emep.int/), and East Asia (Zhang et al. 2009 for NOx; 674
36
Lu et al. 2011 for SO2. Elsewhere, we used anthropogenic emissions from EDGAR v32-FT2000 675
global inventory for 2000 (Olivier et al. 2005), and scaled it based on the energy statistics (van 676
Donkelaar et al. 2008) to the subsequent years up to 2010. Anthropogenic NOx emissions were 677
scaled from 2006 to the subsequent years based on the NO2 column density data retrieved from 678
the OMI satellite sensor (Lamsal et al. 2011). Non-anthropogenic emissions include biomass 679
burning (Mu et al. 2011), soil NOX (Yienger and Levy 1995; Wang et al. 1998), lightning NOx 680
(Price and Rind 1992; Sauvage et al. 2007; Martin et al. 2007; Hudman et al. 2007; Murray et al. 681
2012), ship SO2 from the ICOADS inventory (Lee et al. 2011; Vinker et al. 2012), and volcanic 682
emissions (Fischer et al. 2011). GEOS-Chem includes seasonality for NOx and SO2 based on 683
statistics from the regional inventories, and a diurnal variation for NOx as described in van 684
Donkelaar et al. (2008). 685
Global ammonia emission in GEOS-Chem is from Bouwman et al. (1997) with a seasonality 686
imposed by Park et al. (2004). Spatial and seasonal NH3 variation over Canada is based on 687
monthly varying agricultural activity statistics provided by the Agriculture Canada (Sheppard et 688
al. 2011). Other regional inventories are over the U.S. (EPA-NEI), Europe (EMEP), and East 689
Asia (Streets et al. 2003). East Asian annual emissions are superimposed with a relative seasonal 690
variation (Fischer et al. 2011; Kharol et al. 2013), and a reduction of 30% by Kharol et al. (2013) 691
motivated by comparison with other inventories (Huang et al. 2012a; 2012b). We doubled NH3 692
emissions over California as suggested by Heald et al. (2012) and Walker et al. (2012). 693
GEOS-Chem carbonaceous aerosols include black carbon (BC), organic carbon (OC), and 694
secondary organic aerosols (SOA) simulations (Park et al. 2003; Heald et al. 2011; Wang et al. 695
2011). The global anthropogenic organic OC and BC inventory is from Bond et al. (2007), with 696
37
Cooke et al. (1999) inventory over North America (Leibensperger et al. 2012), Lu et al. (2011) 697
inventory over East Asia. We doubled the East Asian OC and BC emissions based on a 698
comparison with top-down inversion of regional inventories by Fu et al. (2012) and recognize 699
there is ongoing discussion on this topic (Wang et al. 2013). GEOS-Chem simulates the 700
formation of secondary organic aerosol (SOA) from the oxidation of volatile organic compounds 701
(Henze et al. 2006; 2008; Liao et al. 2007; Fu et al. 2008). Global biomass burning emission is 702
from the GFED3 inventory at 3-day temporal resolution (van der Werf et al. 2010; Mu et al. 703
2011), and global biofuel emission is from Yevich and Logan (2003) superimposed by the 704
regional inventories mentioned above. We calculated OM as the sum of model OC and SOA. We 705
used the OM/OC ratio estimated from the observations by the OMI satellite sensor and Aerosol 706
Mass Spectrometer to convert OC to OM to account for the presence of non-carbon elements 707
(Philip et al. in review). 708
GEOS-Chem includes the simulation of natural particles such as mineral dust and sea salt. The 709
mineral dust simulation is described by Fairlie et al. (2007). We use the first dust size bin and 710
37% of the second dust size bin (out of the 5 bins) to get a PM2.5 size range. Sea salt emission in 711
the model is described by Alexander et al. (2005) with updates by Jaegle et al. (2011). We use 712
sea salt accumulation mode size range from 0.1 to 1 μm, which in normal coastal condition 713
represent the approximate PM2.5 size range. 714
GEOS-Chem includes dry deposition (Wesley 1990; Wang et al. 1998; Fischer et al. 2011), and 715
wet deposition (Liu et al. 2001; Amos et al. 2012; Wang et al. 2011). 716
Numerous studies have evaluated the GEOS-Chem ground-level aerosol concentrations and its 717
seasonal variation (e.g., Park et al. 2003, 2004, 2006; Fairlie et al. 2007; Pye et al. 2009; 718
38
Leibensperger et al. 2012; Zhang et al. 2012; Fu et al. 2012; Heald et al. 2012; Walker et al. 719
2012). Vertical profiles of aerosol composition were also compared with various aircraft 720
observations (e.g., van Donkelaar et al. 2008; Drury et al. 2010; Heald et al. 2011), and with 721
CALIOP satellite observations (van Donkelaar et al. 2010; 2013; Ford and Heald 2012). Six-year 722
coincident comparisons of GEOS-Chem and CALIOP suggest simulated near-surface to column 723
extinction ratios are often within 25%, but can approach a factor of two in certain seasons and 724
locations (van Donkelaar et al., 2013). 725
Description of the ground-based PM2.5 composition measurements 726
We utilized filter-based in situ measurements by several networks, such as, the National Air 727
Pollution Surveillance Network (NAPS) and the Canadian Air and Precipitation Monitoring 728
Network (CAPMoN) over Canada (http://www.on.ec.gc.ca/natchem). The U.S. measurement 729
networks include the Clean Air Status and Trends Network (CASTNET, 730
http://java.epa.gov/castnet/epa_jsp/sites.jsp), the Interagency Monitoring of Protected Visual 731
Environments (IMPROVE, http://vista.cira.colostate.edu/improve/), and the U.S. 732
Environmental Protection Agency Air Quality System (EPA-AQS, 733
http://www.epa.gov/ttn/airs/airsaqs/). The NAPS network provides 24-hr composition every 734
third day across Canada as described in Dabek-Zlotorzynska et al. (2011). We used weekly 735
average sulfate and ammonium ion measurements from the CAPMoN and the CASTNET 736
networks even though they are devoid of PM2.5 filters (Zhang et al. 2008). The IMPROVE 737
network provides 24-hr PM2.5 composition data except for ammonium for every third day from 738
several national parks in the U.S. The EPA-AQS network mainly operates over rural areas 739
which report 24-hr averages of all the major composition measurements every consecutive third 740
39
or sixth day. Here, we used the long-term mean (2005-2008) data from EPA-AQS and 741
IMPROVE networks reported by Hand et al. (2011). We calculated ammonium from sulfate and 742
nitrate measurements of EPA-AQS and IMPROVE networks by assuming a fully neutralized 743
sulfuric acid by ammonia gas. We averaged the long-term (2004-2008) in situ measurements of 744
other networks, and subsequently regridded onto the 0.10x0.1
0 grid by ensuring that the in situ 745
data at each grid box represents the long-term mean concentration. We used this dataset to 746
evaluate the satellite-based composition over North America. 747
We treated these in situ data as „truth‟ to evaluate our product. However, it is worth noting some 748
uncertainties. Carbon measurements are prone to errors due to filter contamination (e.g., Rattigan 749
et al. 2011). The ratio of OM to OC varies from 1.2 to 2.6 depending on the spatial and seasonal 750
differences (e.g., Turpin and Lim 2001; Simon et al. 2011). Mineral dust concentrations were 751
from the elemental measurements of IMPROVE and EPA-AQS following Malm et al. (1994) 752
even though measurements of five elements alone are inadequate to determine the ambient 753
mineral dust (Malm and Hand 2007; Hand et al. 2011). Sea salt were from elemental chlorine or 754
chlorine ion measurements, by accounting for 55% chlorine by weight; the selection of sea salt 755
marker as sodium or chlorine is also uncertain (White 2008; Hand et al. 2011). Hand et al. (2012) 756
obtained relative errors for PM2.5 and its chemical components from a comparison of collocated 757
IMPROVE and AQS measurements (17% for PM2.5, 5% for ammonium sulfate, 11% for 758
ammonium nitrate, 10% for OC, 12% for EC, 33% for dust, and 77% for sea salt). Finally, the 759
gridded in situ data are prone to representation error as an individual measurement is used to 760
represent a 10 km x10 km area. 761
40
In addition, we collected annually representative inorganic and organic composition 762
measurements from the European Monitoring and Evaluation Programme (EMEP; 763
http://www.emep.int/), the Acid Deposition Monitoring Network in East Asia (EANET; 764
http://www.eanet.cc/), and several field measurements around the world from published papers 765
such as, Lu et al. 2011, Zhang XY et al. 2008, Cao et al., 2007 (for organic measurements) and 766
several others (Table S-1). We used this dataset to evaluate the global satellite-model 767
composition (see Figure 5 and Table 1). We included only sites with all SIA components to 768
achieve a consistent evaluation. 769
770
771
772
773
774
775
776
777
778
779
780
41
References 781
Alexander B, Park RJ, Jacob DJ, Li QB, Yantosca RM, Savarino J et al. 2005. Sulfate formation 782 in sea-salt aerosols: Constraints from oxygen isotopes. Journal of Geophysical Research-783
Atmospheres 110(D10):D10307; doi: 10.1029/2004JD005659. 784
Amos HM, Jacob DJ, Holmes CD, Fisher JA, Wang Q, Yantosca RM et al. 2012. Gas-particle 785 partitioning of atmospheric hg(II) and its effect on global mercury deposition. Atmospheric 786
Chemistry and Physics 12(1):591-603; doi: 10.5194/acp-12-591-2012. 787
Bey I, Jacob DJ, Yantosca RM, Logan JA, Field BD, Fiore AM et al. 2001. Global modeling of 788
tropospheric chemistry with assimilated meteorology: Model description and evaluation. Journal 789
of Geophysical Research-Atmospheres 106(D19); doi: 10.1029/2001JD000807. 790
Bond TC, Bhardwaj E, Dong R, Jogani R, Jung S, Roden C et al. 2007. Historical emissions of 791
black and organic carbon aerosol from energy-related combustion, 1850-2000. Global 792
Biogeochem Cycles 21(2):GB2018; doi: 10.1029/2006GB002840. 793
Bouwman AF, Lee DS, Asman WAH, Dentener FJ, VanderHoek KW, Olivier JGJ. 1997. A 794
global high-resolution emission inventory for ammonia. Global Biogeochem Cycles 11(4):561-795
587; doi: 10.1029/97GB02266. 796
Brown KW, Bouhamra W, Lamoureux DP, Evans JS, Koutrakis P. 2008. Characterization of 797 particulate matter for three sites in kuwait. J Air Waste Manage Assoc 58(8); doi: 10.3155/1047-798
3289.58.8.994. 799
Cao JJ, Lee SC, Chow JC, Watson JG, Ho KF, Zhang RJ et al. 2007. Spatial and seasonal 800
distributions of carbonaceous aerosols over china. Journal of Geophysical Research-801
Atmospheres 112(D22):D22S11; doi: 10.1029/2006JD008205. 802
Chen D, Wang Y, McElroy MB, He K, Yantosca RM, Le Sager P. 2009. Regional CO pollution 803 and export in china simulated by the high-resolution nested-grid GEOS-chem model. 804
Atmospheric Chemistry and Physics 9(11):3825-3839. 805
Chowdhury M. 2004. Characterization of fine particle air pollution in the Indian subcontinent. 806
PhD Thesis, Georgia Institute of Technology, USA. 807
Cooke WF, Liousse C, Cachier H, Feichter J. 1999. Construction of a 1 degrees x 1 degrees 808
fossil fuel emission data set for carbonaceous aerosol and implementation and radiative impact in 809 the ECHAM4 model. Journal of Geophysical Research-Atmospheres 104(D18):22137-22162; 810
doi: 10.1029/1999JD900187. 811
Dabek-Zlotorzynska E, Dann TF, Martinelango PK, Celo V, Brook JR, Mathieu D et al. 2011. 812 Canadian national air pollution surveillance (NAPS) PM2.5 speciation program: Methodology 813
and PM2.5 chemical composition for the years 2003-2008. Atmos Environ 45(3):673-686; doi: 814
10.1016/j.atmosenv.2010.10.024. 815
42
Drury E, Jacob DJ, Spurr RJD, Wang J, Shinozuka Y, Anderson BE et al. 2010. Synthesis of 816
satellite (MODIS), aircraft (ICARTT), and surface (IMPROVE, EPA-AQS, AERONET) aerosol 817 observations over eastern north america to improve MODIS aerosol retrievals and constrain 818 surface aerosol concentrations and sources. Journal of Geophysical Research-Atmospheres 819
115:D14204; doi: 10.1029/2009JD012629. 820
Evans MJ, Jacob DJ. 2005. Impact of new laboratory studies of N(2)O(5) hydrolysis on global 821 model budgets of tropospheric nitrogen oxides, ozone, and OH. Geophys Res Lett 32(9):L09813; 822
doi: 10.1029/2005GL022469. 823
Fairlie TD, Jacob DJ, Park RJ. 2007. The impact of transpacific transport of mineral dust in the 824
united states. Atmos Environ 41(6):1251-1266; doi: 10.1016/j.atmosenv.2006.09.048. 825
Fisher JA, Jacob DJ, Wang Q, Bahreini R, Carouge CC, Cubison MJ et al. 2011. Sources, 826
distribution, and acidity of sulfate-ammonium aerosol in the arctic in winter-spring. Atmos 827
Environ 45(39):7301-7318; doi: 10.1016/j.atmosenv.2011.08.030. 828
Ford B, Heald CL. 2012. An A-train and model perspective on the vertical distribution of 829
aerosols and CO in the northern hemisphere. Journal of Geophysical Research-Atmospheres 830
117:D06211; doi: 10.1029/2011JD016977. 831
Fu T-, Cao JJ, Zhang XY, Lee SC, Zhang Q, Han YM et al. 2012. Carbonaceous aerosols in 832 china: Top-down constraints on primary sources and estimation of secondary contribution. 833
Atmospheric Chemistry and Physics 12(5):2725-2746; doi: 10.5194/acp-12-2725-2012. 834
Fu T, Jacob DJ, Wittrock F, Burrows JP, Vrekoussis M, Henze DK. 2008. Global budgets of 835
atmospheric glyoxal and methylglyoxal, and implications for formation of secondary organic 836 aerosols. Journal of Geophysical Research-Atmospheres 113(D15):D15303; doi: 837
10.1029/2007JD009505. 838
Hand JL, et al. 2011. IMPROVE (Interagency Monitoring of Protected Visual Environments): 839 Spatial and seasonal patterns and temporal variability of haze and its constituents in the United 840 States, Rep. V, Coop. Inst. For Res. In the Atmos., Fort Collins, Colo. [Available at http:// 841
vista.cira.colostate.edu/improve/Publications/Reports/2011/2011.htm.] 842
Hand JL, Schichtel BA, Pitchford M, Malm WC, Frank NH. 2012. Seasonal composition of 843 remote and urban fine particulate matter in the united states. Journal of Geophysical Research-844
Atmospheres 117:D05209; doi: 10.1029/2011JD017122. 845
Heald CL, Coe H, Jimenez JL, Weber RJ, Bahreini R, Middlebrook AM et al. 2011. Exploring 846 the vertical profile of atmospheric organic aerosol: Comparing 17 aircraft field campaigns with a 847
global model. Atmospheric Chemistry and Physics 11(24):12673-12696; doi: 10.5194/acp-11-848
12673-2011. 849
Heald CL, Collett JL,Jr., Lee T, Benedict KB, Schwandner FM, Li Y et al. 2012. Atmospheric 850 ammonia and particulate inorganic nitrogen over the united states. Atmospheric Chemistry and 851
Physics 12(21):10295-10312; doi: 10.5194/acp-12-10295-2012. 852
43
Henze DK, Seinfeld JH, Ng NL, Kroll JH, Fu T-, Jacob DJ et al. 2008. Global modeling of 853
secondary organic aerosol formation from aromatic hydrocarbons: High- vs. low-yield pathways. 854
Atmospheric Chemistry and Physics 8(9):2405-2420. 855
Henze DK, Seinfeld JH. 2006. Global secondary organic aerosol from isoprene oxidation. 856
Geophys Res Lett 33(9):L09812; doi: 10.1029/2006GL025976. 857
Huang K, Zhuang G, Lin Y, Fu JS, Wang Q, Liu T et al. 2012a. Typical types and formation 858 mechanisms of haze in an eastern asia megacity, shanghai. Atmospheric Chemistry and Physics 859
12(1):105-124; doi: 10.5194/acp-12-105-2012. 860
Huang X, Song Y, Li M, Li J, Huo Q, Cai X et al. 2012b. A high-resolution ammonia emission 861
inventory in china. Global Biogeochem Cycles 26:GB1030; doi: 10.1029/2011GB004161. 862
Hudman RC, Jacob DJ, Turquety S, Leibensperger EM, Murray LT, Wu S et al. 2007. Surface 863 and lightning sources of nitrogen oxides over the united states: Magnitudes, chemical evolution, 864
and outflow. Journal of Geophysical Research-Atmospheres 112(D12):D12S05; doi: 865
10.1029/2006JD007912. 866
Jacob DJ. 2000. Heterogeneous chemistry and tropospheric ozone. Atmos Environ 34(12-867
14):2131-2159; doi: 10.1016/S1352-2310(99)00462-8. 868
Jaegle L, Quinn PK, Bates TS, Alexander B, Lin J-. 2011. Global distribution of sea salt 869
aerosols: New constraints from in situ and remote sensing observations. Atmospheric Chemistry 870
and Physics 11(7):3137-3157; doi: 10.5194/acp-11-3137-2011. 871
Kharol SK, Martin RV, Philip S et al. 2013. Persistent sensitivity of Asian aerosol to emissions 872
of nitrogen oxides, Geophys. Res. Lett., 40, doi:10.1002/grl.50234. 873
Kuhns H, Knipping EM, Vukovich JM. 2005. Development of a united states-mexico emissions 874
inventory for the big bend regional aerosol and visibility observational (BRAVO) study. J Air 875
Waste Manage Assoc 55(5):677-692. 876
Lamsal LN, Martin RV, Padmanabhan A, van Donkelaar A, Zhang Q, Sioris CE et al. 2011. 877 Application of satellite observations for timely updates to global anthropogenic NOx emission 878
inventories. Geophys Res Lett 38:L05810; doi: 10.1029/2010GL046476. 879
Lee C, Martin RV, van Donkelaar A, Lee H, Dickerson RR, Hains JC et al. 2011. SO2 emissions 880 and lifetimes: Estimates from inverse modeling using in situ and global, space-based 881
(SCIAMACHY and OMI) observations. Journal of Geophysical Research-Atmospheres 882
116:D06304; doi: 10.1029/2010JD014758. 883
Leibensperger EM, Mickley LJ, Jacob DJ, Chen W-, Seinfeld JH, Nenes A et al. 2012. Climatic 884 effects of 1950-2050 changes in US anthropogenic aerosols - part 1: Aerosol trends and radiative 885
forcing. Atmospheric Chemistry and Physics 12(7):3333-3348; doi: 10.5194/acp-12-3333-2012. 886
Liao H, Henze DK, Seinfeld JH, Wu S, Mickley LJ. 2007. Biogenic secondary organic aerosol 887
over the united states: Comparison of climatological simulations with observations. Journal of 888
Geophysical Research-Atmospheres 112(D6):D06201; doi: 10.1029/2006JD007813. 889
44
Liu HY, Jacob DJ, Bey I, Yantosca RM. 2001. Constraints from pb-210 and be-7 on wet 890
deposition and transport in a global three-dimensional chemical tracer model driven by 891 assimilated meteorological fields. Journal of Geophysical Research-Atmospheres 892
106(D11):12109-12128; doi: 10.1029/2000JD900839. 893
Lu Z, Zhang Q, Streets DG. 2011. Sulfur dioxide and primary carbonaceous aerosol emissions in 894
china and india, 1996-2010. Atmospheric Chemistry and Physics 11(18):9839-9864; doi: 895
10.5194/acp-11-9839-2011. 896
Malm WC, Sisler JF, Huffman D, Eldred RA, Cahill TA. 1994. Spatial and seasonal trends in 897 particle concentration and optical extinction in the united-states. Journal of Geophysical 898
Research-Atmospheres 99(D1):1347-1370; doi: 10.1029/93JD02916. 899
Malm WC, Hand JL. 2007. An examination of the physical and optical properties of aerosols 900
collected in the IMPROVE program. Atmos Environ 41(16):3407-3427; doi: 901
10.1016/j.atmosenv.2006.12.012. 902
Martin RV, Jacob DJ, Yantosca RM, Chin M, Ginoux P. 2003. Global and regional decreases in 903
tropospheric oxidants from photochemical effects of aerosols. Journal of Geophysical Research-904
Atmospheres 108(D3):4097; doi: 10.1029/2002JD002622. 905
Martin RV, Sauvage B, Folkins I, Sioris CE, Boone C, Bernath P et al. 2007. Space-based 906 constraints on the production of nitric oxide by lightning. Journal of Geophysical Research-907
Atmospheres 112(D9):D09309; doi: 10.1029/2006JD007831. 908
Mu M, Randerson JT, van der Werf GR, Giglio L, Kasibhatla P, Morton D et al. 2011. Daily and 909
3-hourly variability in global fire emissions and consequences for atmospheric model predictions 910 of carbon monoxide. Journal of Geophysical Research-Atmospheres 116:D24303; doi: 911
10.1029/2011JD016245. 912
Murray LT, Jacob DJ, Logan JA, Hudman RC, Koshak WJ. 2012. Optimized regional and 913 interannual variability of lightning in a global chemical transport model constrained by LIS/OTD 914 satellite data. Journal of Geophysical Research-Atmospheres 117:D20307; doi: 915
10.1029/2012JD017934. 916
Olivier JGJ, et al. 2005. Recent trends in global greenhouse gas emissions: regional trends 1970-917 2000 and spatial distribution of key sources in 2000. Env. Sc., 2 (2-3), 81-99. DOI: 918
10.1080/15693430500400345. 919
Park RJ, Jacob DJ, Chin M, Martin RV. 2003. Sources of carbonaceous aerosols over the united 920 states and implications for natural visibility. Journal of Geophysical Research-Atmospheres 921
108(D12):4355; doi: 10.1029/2002JD003190. 922
Park RJ, Jacob DJ, Field BD, Yantosca RM, Chin M. 2004. Natural and transboundary pollution 923 influences on sulfate-nitrate-ammonium aerosols in the united states: Implications for policy. 924
Journal of Geophysical Research-Atmospheres 109(D15):D15204; doi: 10.1029/2003JD004473. 925
45
Park RJ, Jacob DJ, Kumar N, Yantosca RM. 2006. Regional visibility statistics in the united 926
states: Natural and transboundary pollution influences, and implications for the regional haze 927
rule. Atmos Environ 40(28):5405-5423; doi: 10.1016/j.atmosenv.2006.04.059. 928
Pey J, Querol X, Alastuey A. 2009. Variations of levels and composition of PM10 and PM2.5 at 929 an insular site in the western mediterranean. Atmos Res 94(2); doi: 930
10.1016/j.atmosres.2009.06.006. 931
Price C, Rind D. 1992. A simple lightning parameterization for calculating global lightning 932
distributions. Journal of Geophysical Research-Atmospheres 97(D9):9919-9933. 933
Pye HOT, Liao H, Wu S, Mickley LJ, Jacob DJ, Henze DK et al. 2009. Effect of changes in 934
climate and emissions on future sulfate-nitrate-ammonium aerosol levels in the united states. 935
Journal of Geophysical Research-Atmospheres 114:D01205; doi: 10.1029/2008JD010701. 936
Qu WJ, Zhang XY, Arimoto R, Wang D, Wang YQ, Yan LW et al. 2008. Chemical composition 937
of the background aerosol at two sites in southwestern and northwestern china: Potential 938 influences of regional transport. Tellus Series B-Chemical and Physical Meteorology 60(4):657-939
673; doi: 10.1111/j.1600-0889.2008.00342-x. 940
Qu W, Zhang X, Arimoto R, Wang Y, Wang D, Sheng L et al. 2009. Aerosol background at two 941 remote CAWNET sites in western china. Sci Total Environ 407(11); doi: 942
10.1016/j.scitotenv.2009.02.012. 943
Rattigan OV, Felton HD, Bae M, Schwab JJ, Demerjian KL. 2011. Comparison of long-term 944 PM2.5 carbon measurements at an urban and rural location in new york. Atmos Environ 945
45(19):3228-3236; doi: 10.1016/j.atmosenv.2011.03.048. 946
Ridley DA, Heald CL, Ford B. 2012. North african dust export and deposition: A satellite and 947 model perspective. Journal of Geophysical Research-Atmospheres 117:D02202; doi: 948
10.1029/2011JD016794. 949
Sarnat JA, Moise T, Shpund J, Liu Y, Pachon JE, Qasrawi R et al. 2010. Assessing the spatial 950
and temporal variability of fine particulate matter components in israeli, jordanian, and 951
palestinian cities. Atmos Environ 44(20); doi: 10.1016/j.atmosenv.2010.04.007. 952
Sauvage B, Martin RV, van Donkelaar A, Ziemke JR. 2007. Quantification of the factors 953
controlling tropical tropospheric ozone and the south atlantic maximum. Journal of Geophysical 954
Research-Atmospheres 112(D11):D11309; doi: 10.1029/2006JD008008. 955
Sheppard SC, Bittman S, Swift ML, Tait J. 2011. Modelling monthly NH3 emissions from dairy 956
in 12 ecoregions of canada. Canadian Journal of Animal Science 91(4):649-661; doi: 957
10.4141/CJAS2010-005. 958
Simon H, Bhave PV, Swall JL, Frank NH, Malm WC. 2011. Determining the spatial and 959
seasonal variability in OM/OC ratios across the US using multiple regression. Atmospheric 960
Chemistry and Physics 11(6):2933-2949; doi: 10.5194/acp-11-2933-2011. 961
46
So KL, Guo H, Li YS. 2007. Long-term variation of PM2.5 levels and composition at rural, 962
urban, and roadside sites in hong kong: Increasing impact of regional air pollution. Atmos 963
Environ 41(40); doi: 10.1016/j.atmosenv.2007.08.053. 964
Soluri DS, Godoy MLDP, Godoy JM, Roldao LA. 2007. Multi-site PM2.5 and PM2.5-10 aerosol 965
source apportionment in rio de janeiro, brazil. Journal of the Brazilian Chemical Society 18(4). 966
Stone E, Schauer J, Quraishi TA, Mahmood A. 2010. Chemical characterization and source 967 apportionment of fine and coarse particulate matter in lahore, pakistan. Atmos Environ 44(8); 968
doi: 10.1016/j.atmosenv.2009.12.015. 969
Streets DG, Bond TC, Carmichael GR, Fernandes SD, Fu Q, He D et al. 2003. An inventory of 970
gaseous and primary aerosol emissions in asia in the year 2000. Journal of Geophysical 971
Research-Atmospheres 108(D21):8809; doi: 10.1029/2002JD003093. 972
Thornton JA, Jaegle L, McNeill VF. 2008. Assessing known pathways for HO2 loss in aqueous 973
atmospheric aerosols: Regional and global impacts on tropospheric oxidants. Journal of 974
Geophysical Research-Atmospheres 113(D5):D05303; doi: 10.1029/2007JD009236. 975
Turpin BJ, Lim HJ. 2001. Species contributions to PM2.5 mass concentrations: Revisiting 976
common assumptions for estimating organic mass. Aerosol Science and Technology 35(1):602-977
610; doi: 10.1080/02786820152051454. 978
van der Werf GR, Randerson JT, Giglio L, Collatz GJ, Mu M, Kasibhatla PS et al. 2010. Global 979 fire emissions and the contribution of deforestation, savanna, forest, agricultural, and peat fires 980 (1997-2009). Atmospheric Chemistry and Physics 10(23):11707-11735; doi: 10.5194/acp-10-981
11707-2010. 982
van Donkelaar A, Martin RV, Leaitch WR, Macdonald AM, Walker TW, Streets DG et al. 2008. 983 Analysis of aircraft and satellite measurements from the intercontinental chemical transport 984
experiment (INTEX-B) to quantify long-range transport of east asian sulfur to canada. 985
Atmospheric Chemistry and Physics 8(11). 986
van Donkelaar A, Martin RV, Brauer M, Kahn R, Levy R, Verduzco C et al. 2010. Global 987 estimates of ambient fine particulate matter concentrations from satellite-based aerosol optical 988
depth: Development and application. Environ Health Perspect 118(10):847-855; doi: 989
10.1289/ehp.0901623. 990
van Donkelaar A, Martin RV, Pasch AN, Szykman JJ, Zhang L, Wang YX et al. 2012. 991 Improving the accuracy of daily satellite-derived ground-level fine aerosol concentration 992
estimates for north america. Environ Sci Technol 46(21):11971-11978; doi: 10.1021/es3025319. 993
van Donkelaar A, Martin RV, et al. 2013. Optimal estimation for global ground-level fine 994 particulate matter concentrations. Journal of Geophysical Research-Atmospheres, Accepted, doi: 995
10.1002/jgrd.50479. 996
47
Vinken GCM, Boersma KF, Jacob DJ, Meijer EW. 2011. Accounting for non-linear chemistry of 997
ship plumes in the GEOS-chem global chemistry transport model. Atmospheric Chemistry and 998
Physics 11(22):11707-11722; doi: 10.5194/acp-11-11707-2011. 999
Walker JM, Philip S, Martin RV, Seinfeld JH. 2012. Simulation of nitrate, sulfate, and 1000 ammonium aerosols over the united states. Atmospheric Chemistry and Physics 12(22):11213-1001
11227; doi: 10.5194/acp-12-11213-2012. 1002
Wang X, et al. 2013. Top-down estimate of China's black carbon emissions using surface 1003
observations: Sensitivity to observation representativeness and transport model error. Journal of 1004
Geophysical Research-Atmospheres, doi: 10.1002/jgrd.50397. 1005
Wang Q, Jacob DJ, Fisher JA, Mao J, Leibensperger EM, Carouge CC et al. 2011. Sources of 1006 carbonaceous aerosols and deposited black carbon in the arctic in winter-spring: Implications for 1007
radiative forcing. Atmospheric Chemistry and Physics 11(23):12453-12473; doi: 10.5194/acp-1008
11-12453-2011. 1009
Wang Q, Jacob DJ, Fisher JA, Mao J, Leibensperger EM, Carouge CC et al. 2011. Sources of 1010
carbonaceous aerosols and deposited black carbon in the arctic in winter-spring: Implications for 1011 radiative forcing. Atmospheric Chemistry and Physics 11(23):12453-12473; doi: 10.5194/acp-1012
11-12453-2011. 1013
Wang YH, Jacob DJ, Logan JA. 1998. Global simulation of tropospheric O-3-NOx-hydrocarbon 1014 chemistry 1. model formulation. Journal of Geophysical Research-Atmospheres 103(D9):10713-1015
10725; doi: 10.1029/98JD00158. 1016
Wesely ML. 1989. Parameterization of surface resistances to gaseous dry deposition in regional-1017
scale numerical-models. Atmos Environ 23(6):1293-1304; doi: 10.1016/0004-6981(89)90153-4. 1018
White WH. 2008. Chemical markers for sea salt in IMPROVE aerosol data. Atmos Environ 1019
42(2):261-274; doi: 10.1016/j.atmosenv.2007.09.040. 1020
Yang F, Tan J, Zhao Q, Du Z, He K, Ma Y et al. 2011. Characteristics of PM2.5 speciation in 1021
representative megacities and across china. Atmospheric Chemistry and Physics 11(11); doi: 1022
10.5194/acp-11-5207-2011. 1023
Yevich R, Logan JA. 2003. An assessment of biofuel use and burning of agricultural waste in the 1024
developing world. Global Biogeochem Cycles 17(4):1095; doi: 10.1029/2002GB001952. 1025
Yienger JJ, Levy H. 1995. Empirical-model of global soil-biogenic nox emissions. Journal of 1026
Geophysical Research-Atmospheres 100(D6):11447-11464; doi: 10.1029/95JD00370. 1027
Zhang L, Jacob DJ, Knipping EM, Kumar N, Munger JW, Carouge CC et al. 2012. Nitrogen 1028 deposition to the united states: Distribution, sources, and processes. Atmospheric Chemistry and 1029
Physics 12(10):4539-4554; doi: 10.5194/acp-12-4539-2012. 1030
Zhang L, Vet R, Wiebe A, Mihele C, Sukloff B, Chan E et al. 2008. Characterization of the size-1031
segregated water-soluble inorganic ions at eight canadian rural sites. Atmospheric Chemistry and 1032
Physics 8(23):7133-7151. 1033
48
Zhang Q, Streets DG, Carmichael GR, He KB, Huo H, Kannari A et al. 2009. Asian emissions in 1034
2006 for the NASA INTEX-B mission. Atmospheric Chemistry and Physics 9(14):5131-5153. 1035
Zhang XY, Wang YQ, Niu T, Zhang XC, Gong SL, Zhang YM et al. 2012. Atmospheric aerosol 1036 compositions in china: Spatial/temporal variability, chemical signature, regional haze 1037 distribution and comparisons with global aerosols. Atmospheric Chemistry and Physics 12(2); 1038
doi: 10.5194/acp-12-779-2012. 1039
Zhang XY, Wang YQ, Zhang XC, Guo W, Gong SL. 2008. Carbonaceous aerosol composition 1040
over various regions of china during 2006. Journal of Geophysical Research-Atmospheres 1041
113(D14):D14111; doi: 10.1029/2007JD009525. 1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
49
Table S1: Global in situ data collected from publications 1063
1064
1065
Note: NA represents data not available or filtered out. 1066
Sl. No. Site/City Country Latitude Longitude Study Period SO42- NO3
- NH4+
OC BC Size Source
degree degree ug/m3
ug/m3
ug/m3
ug/m3
ug/m3
1 Beijing China 40.3 116.3 Mar 2005 - Feb 2006 15.8 10.1 7.3 34.4 8.3 PM2.5 (Yang et al. 2011)
2 Miyun Resevoir China 40.5 116.8 Mar 2005 - Feb 2006 13.0 6.4 6.1 21.9 3.8 " "
3 Chongqing China 29.6 106.5 Mar 2005 - Feb 2006 25.5 5.3 7.9 42.2 6.5 " "
4 Dadukou China 29.5 106.5 Mar 2005 - Feb 2006 23.4 5.1 7.6 47.2 6.4 " "
5 Jinyun China 29.8 106.4 Mar 2005 - Feb 2006 24.0 4.8 7.3 41.2 4.7 " "
6 Hok Tsui Hong Kong 22.2 114.3 Nov 2004 - Oct 2005 11.9 0.8 3.1 4.3 2.1 " (So et al. 2007)
7 Tsuen Wan Hong Kong 22.4 114.1 Nov 2004 - Oct 2005 13.2 1.6 4.1 7.4 6.0 " "
8 Mong Kok Hong Kong 22.3 114.1 Nov 2004 - Oct 2005 12.8 2.4 4.4 11.9 13.7 " "
9 Delhi India 28.4 77.1 Mar 2001 - Jan 2002 10.9 6.1 6.4 40.3 10.5 " (Choudhary 2004)
10 Mumbai India 22.6 88.3 Mar 2001 - Jan 2002 6.9 1.6 2.0 12.6 4.6 " "
11 Kolkata India 18.7 72.8 Mar 2001 - Jan 2002 7.2 2.9 3.6 37.1 12.1 " "
12 Lahore Pakistan 31.5 74.3 Jan 2007 - Jan 2008 10.5 6.6 3.6 64.4 11.2 " (Stone et al. 2010)
13 Brazil Brazil 23.0 43.2 Sep 2003 - Sep 2004 1.5 0.3 0.6 NA 2.1 " (Soluri et al. 2007)
14 Brazil Brazil 22.9 43.2 Sep 2003 - Sep 2004 1.9 0.4 0.7 NA 3.1 " "
15 Brazil Brazil 22.9 43.2 Sep 2003 - Sep 2004 2.2 0.5 0.7 NA 2.9 " "
16 Brazil Brazil 23.0 43.4 Sep 2003 - Sep 2004 1.4 0.3 0.4 NA 1.3 " "
17 Brazil Brazil 23.0 43.3 Sep 2003 - Sep 2004 1.8 0.5 0.5 NA 2.7 " "
18 Brazil Brazil 22.8 43.4 Sep 2003 - Sep 2004 1.5 0.4 0.5 NA 3.2 " "
19 Brazil Brazil 23.0 43.6 Sep 2003 - Sep 2004 1.5 0.4 0.5 NA 1.2 " "
20 Brazil Brazil 22.9 43.6 Sep 2003 - Sep 2004 1.7 0.4 0.6 NA 2.4 " "
21 Brazil Brazil 22.9 43.4 Sep 2003 - Sep 2004 1.7 0.4 0.6 NA 2.5 " "
22 Brazil Brazil 22.9 43.7 Sep 2003 - Sep 2004 1.7 0.3 0.5 NA 1.7 " "
23 Tel Aviv Israel 32.1 34.8 Jan 2007 - Dec 2007 5.2 0.7 NA 4.7 1.5 " (Sarnat et al. 2010)
24 Haifa Israel 32.8 35.0 Jan 2007 - Dec 2007 5.2 1.4 NA 3.0 1.0 " "
25 W. Jerusalem Israel 31.8 35.2 Jan 2007 - Dec 2007 4.6 1.0 NA 4.1 1.1 " "
26 Hebron Palestine 31.5 35.1 Jan 2007 - Dec 2007 3.8 0.9 NA 5.3 1.8 " "
27 E. Jerusalem Palestine 31.8 35.2 Jan 2007 - Dec 2007 4.4 0.9 NA 5.2 2.2 " "
28 Nablus Palestine 32.2 35.2 Jan 2007 - Dec 2007 4.3 1.0 NA 8.2 5.6 " "
29 Amman Jordan 32.0 35.8 Jan 2007 - Dec 2007 4.5 1.0 NA 6.2 2.4 " "
30 Balearic islands Spain 39.6 2.6 Jan 2004 - Feb 2005 3.9 0.9 2.0 2.9 0.5 " (Pey et al. 2009)
31 Kuwait Kuwait 29.3 48.0 Feb 2004 - Jan 2005 9.9 1.8 NA 3.7 2.5 " (Brown et al. 2008)
32 Chengdu China 30.7 104.0 Jan 2006 - Dec 2007 40.5 NA 14.0 36.3 10.8 PM10 (Zhang XY et al. 2012)
33 Dalian China 38.9 121.6 Jan 2006 - Dec 2007 23.3 NA 7.7 20.2 5.3 " "
34 Dunhuang China 40.2 94.7 Jan 2006 - Dec 2007 6.6 NA 0.4 26.7 3.6 " "
35 Gaolanshan China 36.0 105.9 Jan 2006 - Dec 2007 16.7 NA 6.5 19.1 3.8 " "
36 Gucheng China 39.1 115.8 Jan 2006 - Dec 2007 35.5 NA 14.4 38.5 11.0 " "
37 Jinsha China 29.6 114.2 Jan 2006 - Dec 2007 26.6 NA 7.6 15.3 3.0 " "
38 Lhasa China 29.7 91.1 Jan 2006 - Dec 2007 2.9 NA 0.2 21.7 3.8 " "
39 LinAn China 30.3 119.7 Jan 2006 - Dec 2007 21.7 NA 6.8 15.1 4.3 " "
40 Longfengshan China 44.7 127.6 Jan 2006 - Dec 2007 10.0 NA 2.5 15.9 2.3 " "
41 Nanning China 22.8 108.4 Jan 2006 - Dec 2007 21.6 NA 5.8 17.9 4.0 " "
42 Panyu China 23.0 113.4 Jan 2006 - Dec 2007 26.8 NA 8.6 22.3 7.9 " "
43 Taiyangshan China 29.2 111.7 Jan 2006 - Dec 2007 28.8 NA 7.9 13.8 2.7 " "
44 Xian China 34.4 109.0 Jan 2006 - Dec 2007 46.7 NA 14.4 42.6 12.7 " "
45 Zhengzhou China 34.8 113.7 Jan 2006 - Dec 2007 45.0 NA 16.5 29.2 9.2 " "
46 Akdala China 47.1 88.0 Jul 2004 - Mar 2005 3.3 NA 0.6 2.9 0.4 " (Qu et al. 2008; 2009)
47 Shangri-La, Zhuzhang China 28.0 99.7 Jul 2004 - Mar 2005 1.6 NA 0.2 3.1 0.3 " "
