Environment International 91 (2016) 122–132

Contents lists available at ScienceDirect

Environment International

journal homepage: www.elsevier.com/locate/envint

Full length article

Eight-year (2007–2014) trends in ambient fine particulate matter

(PM2.5) and its chemical components in the Capital Region of
Alberta, Canada
Md. Aynul Bari ⁎, Warren B. Kindzierski
School of Public Health, University of Alberta, 3-57 South Academic Building, 11405-87 Avenue, Edmonton, Alberta T6G 1C9, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Currently there have been questions about ambient fine particulate matter (PM2.5) levels in the Capital Region of
Received 3 November 2015 Alberta, Canada. An investigation of temporal trends in PM2.5 and its chemical components was undertaken in the
Received in revised form 26 February 2016 City of Edmonton within the Capital Region over an 8-year period (2007–2014). A non-parametric trend detec-
Accepted 28 February 2016
tion method was adopted to characterize trends in ambient concentrations. No statistically significant change
Available online 4 March 2016
was observed for ambient PM2.5 concentrations during 2007–2014, while significant decreasing trends were
Keywords:
found for organic carbon, elemental carbon, oxalate, barium, lead and cadmium. A statistically significant increas-
Fine particulate matter ing trend was observed for sodium chloride indicating an increase of de-icing salt contribution for winter road
Chemical components maintenance in recent years. Concentrations of potassium ion and zinc exhibited strong and significant seasonal
Trends variability with higher concentrations in winter than in summer likely reflecting wood smoke origins more than
Edmonton other potential sources in Edmonton and the surrounding region. No statistically significant changes were ob-
served for all other chemical components examined. Notwithstanding robust population growth that has oc-
curred in Edmonton, these findings reveal that particulate air quality and corresponding trace elements in
Edmonton's air has been unchanged or improved over the investigated period (2007–2014). Longer-term air
quality monitoring at least over several decades is needed to establish whether trends reported here are actually
occurring.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction particle number and mass concentrations (Forsberg et al., 2005;

Exposure to ambient fine particulate matter i.e., PM2.5 (with aerody-
namic diameter b 2.5 μm) is of increasing public concern over the past
decades. A number of epidemiological studies suggest an association be-
tween long-term and short-term exposure to PM2.5 and increased mor-
bidity, mortality and emergency hospital admissions for cardiovascular,
cerebrovascular (stoke) and ischemic heart disease as well as diabetes
and neurological disorders (Pope et al., 2014; Zanobetti et al., 2014;
Burnett et al., 2014; Weichenthal et al., 2014; Lippmann, 2014). Apart
of its deleterious effects on public health, studies also indicate that par-
ticulate matter affects visibility and climate change (Cheung et al., 2005;
Forster et al., 2007).
Characterization of ambient PM2.5 is complex and an understanding
of their sources, sinks, temporal and spatial variation is needed. At pres-
ent, there is not enough evidence at the population level to identify dif-
ferences in the effects of particles with different chemical compositions
or emanating from various sources. Nevertheless, there is a body of ev-
idence that chemical composition of particles may be valuable and ap-
propriate information for evaluating PM health risk in addition to
⁎ Corresponding author.
E-mail address: mdaynul@ualberta.ca (M.A. Bari).
Cassee et al., 2013; Dominici et al., 2015; Peters et al., 2015). From a tox-
icological point of view, some trace elements and their compounds
(e.g., arsenic, cadmium, cobalt, chromium, mercury, manganese, nickel,
lead, antimony, selenium) fall under the list of toxic substances (arsenic,
lead, mercury, cadmium, nickel) as defined by Health Canada under the
Canadian Environmental Protection Act (CEPA, 1999) and are also con-
sidered hazardous under the United States Environmental Protection
Agency Clean Air Act (U.S. EPA, 2009). Some studies indicate that
crustal-derived elements (e.g., silicon, aluminum, calcium, iron) are pre-
dominant in coarse particles (PM10-2.5), and appear to be less potentially
toxic than combustion-derived trace elements (e.g., lead, arsenic chro-
mium, nickel, zinc, cadmium and selenium), which are more abundant
in fine particle fractions (i.e., PM2.5) suggesting that smaller sizes, com-
position and sources of particles are important factors in toxicity (U.S.
EPA, 2009; Dominici et al., 2014).
Canadian Ambient Air Quality Standards (CAAQS) for PM2.5 were de-
veloped in 2012 through a collaborative process involving federal, pro-
vincial and territorial governments and stakeholders and the Canadian
Council of Ministers of the Environment (CCME, 2012). Under the Air
Quality Management System (AQMS), Canadian provinces and terri-
tories will need to determine if a given air zone achieves the CAAQS.
Previously, a 24 h Canada-Wide Standard (CWS) for PM2.5 of 30 μg/m3

http://dx.doi.org/10.1016/j.envint.2016.02.033
0160-4120/© 2016 Elsevier Ltd. All rights reserved.
 M.A. Bari, W.B. Kindzierski / Environment International 91 (2016) 122–132
was established and has been used to assess and manage urban air qual-
ity issues since 2000 (CCME, 2000). The new CAAQS for PM2.5 was de-
veloped in 2012 and it will replace and is more stringent than the
CWS. The new 24 h and annual average CAAQS for PM2.5 for the year
2015 are 28 μg/m3 and 10 μg/m3; while for the year 2020 they are
27 μg/m3 and 8.8 μg/m3, respectively (CCME, 2012).
Recently there has been increasing public concern about the status
of particulate air pollution in the Capital Region (Edmonton and imme-
diate surrounding area) of Alberta. A recent investigation of air quality
event days represented by elevated hourly concentrations of ambient
PM2.5 was undertaken in the Capital Region (AESRD, 2014). Results in-
dicated that elevated hourly PM2.5 concentrations near roadways did
not co-vary with traffic volumes at roadway locations near air monitor-
ing stations and nitrogen dioxide concentrations on ‘high-hourly con-
centration’ event days. AESRD (2014) suggested that air quality
monitoring stations are not impacted by primary fine particulate matter
from vehicle traffic. It was also suggested that fine particulate matter on
event days may be the result of formation of secondary fine particles
from regional precursor sources that may include, but are not limited
to, vehicle emissions. The study concluded that the fine particulate mat-
ter issue in the Capital Region is complex.
Another recent study was conducted in six Alberta air zones to as-
sess potential compliance with the new 2015 CAAQS for PM2.5 for the
time period of 2011–2013 (Alberta Environment and Parks, 2015). It
was reported that in the North Saskatchewan air zone which includes
the Capital Region, the Edmonton East air quality monitoring station
had PM2.5 metric values (24 h average: 38.2 μg/m3, annual average:
11.0 μg/m3) above the new CAAQS in 2013. Recent media releases re-
ported by the Canadian Association of Physicians for the Environment
(CAPE) also claimed that Edmonton had higher levels of harmful air pol-
lutant compared to a big city like Toronto and PM2.5 exceeded CWS of
30 μg/m3 on several winter days in 2010 through 2012 (National Post,
2015). This raises questions about the current state of particulate matter
air quality in the City of Edmonton. It was therefore of particular interest
to further investigate the state of air quality in the region to understand
how poor or good it is for fine particulate matter in Edmonton.
Evaluating ambient air quality monitoring data is important for
characterizing and understanding the state of air quality and whether
it is changing. In this study, we made an in-depth investigation of
trends, monthly and annual concentration profiles of PM2.5 and its
chemical components using available data for last 8 years
(2007–2014). The objective of this investigation was to establish,
whether concentrations of PM2.5 and its chemical composition have
changed over the last 8 years.
2. Methodology
2.1. Study area
Edmonton is the capital and second largest city in Alberta with a
population within its municipal boundaries of 877,926 in 2014
(Municipal Census, 2015). The Capital Region of Alberta (Supplementa-
ry information, Fig. S1) includes the Edmonton Census Metropolitan
Area, Lamont County, and the Elk Island National Park. It covers an
area of 11,993 km2, accounts for 1.9% of Alberta's land mass and 31.8%
of Alberta's population (AESRD, 2014). The Capital Region has a relative-
ly humid continental climate with wide variations in seasonal tempera-
tures. Daily average temperatures range from − 10.4 °C in January to
17.7 °C in July. Prevailing wind directions are west-northwesterly and
early morning surface inversions are frequent throughout the whole
year (Myrick et al., 1994). AESRD (2014) used upper air sounding data
from weather balloons released twice daily at the Environment
Canada Stony Plain weather observation station located 25 km west of
Edmonton to calculate the occurrence of inversion conditions by
month for 2006–2011 for the Capital Region. The monthly percent fre-
quency of inversion conditions for this period is illustrated in Fig. S2
123
and it shows that inversion conditions occurred most frequently in No-
vember through January. The major emission sources of PM2.5 and pre-
cursor gases in the Capital Region include petroleum refining, light-
manufacturing, and fugitive emissions from holding tanks or petroleum
terminals as well as transportation, space heating (e.g., wood burning
fireplaces), coal- and gas-fired electrical power generating units located
60 km to the west, wildfires smoke and biogenic emissions and other
non-specific industrial sources (Myrick et al., 1994; Schulz and
Kindzierski, 2001; AESRD, 2014). Over the 10-year period 2004–2014
the city added 177,270 residents to its population (City of Edmonton,
2015) and 19,800 more registered vehicles each year (average) using
its roadways (Alberta Transportation, 2008, 2012, 2014).
The study was conducted in Edmonton using historical data mea-
sured at Environment Canada's National Air Pollution Surveillance
(NAPS) PM2.5 chemical speciation monitoring station at Edmonton
McIntyre (53° 29′ 9.8″ N, 113° 27′ 52.5″ W). The location of Edmonton
McIntyre station and three other air monitoring stations (Edmonton
south, central and east) is shown in Fig. 1. The station is located to the
south of Edmonton and close to highway 14 (Whitemud Drive). A
wind rose plot was generated (Fig. S3) to show a 8-year (2007–2014)
average prevailing wind direction at the monitoring station using his-
torical hourly meteorological data publically available from the Clean
Air Strategic Alliance (CASA, 2015) data warehouse. The wind rose
plot indicates a general tendency for winds to be from southerly and
west-northwesterly directions at Edmonton McIntyre station. Prevail-
ing winds blew from the south direction 40% of the time and from the
west-northwest 23% of the time over the 8-year period at Edmonton
McIntyre.
2.2. Sampling and analysis
As part of Environment Canada's NAPS program, Alberta Environ-
ment and Parks (AEP) routinely monitors criteria air pollutants includ-
ing PM2.5 in Edmonton air monitoring stations. Up to 2009, hourly
concentrations of PM2.5 had been measured at all Edmonton stations
using Tapered Element Oscillating Microbalances (TEOM). Since May
2009 the Environment Canada NAPS program adopted the United
States Environment Protection Agency's Federal Equivalent Method
(FEM) i.e., TEOM coupled with Filter Dynamics Measurement Systems
(TEOM-FDMS) in order to ensure comparability of continuous monitors
to a reference method and to capture semi-volatile components of the
PM2.5 mass (Government of Canada, 2013; Alberta Environment,
2013). Edmonton McIntyre station has been used for quality assurance
for PM2.5 monitoring methods, where TEOM-FDMS has been deployed
since November 2006. Data for hourly PM2.5 concentrations from
TEOM-FDMS at Edmonton McIntrye station was accessed for the
2007–2014 period via the CASA data warehouse (CASA, 2015).
Data was also accessed for ambient 24 h integrated PM2.5 sampling
that has been performed at Edmonton McIntyre station at a frequency
of once every 3 days from 2007 as part of the NAPS PM2.5 speciation pro-
gram. Samples are collected using co-located Thermo Scientific sam-
plers (Partisol-Plus 2025D sequential dichotomous particle sampler,
Partisol 2300 sequential speciation sampler) and analyzed by the Anal-
ysis and Air Quality Section (AAQS) of Environment Canada in Ottawa.
Data for PM2.5 components including organic carbon (OC), elemental
carbon (EC), major water soluble ions as well as heavy and trace metals
were accessed for the 2007–2014 period from NAPS data products
made available to the public by Environment Canada (2015a). Details
of the data analytical methods and quality control are presented else-
where (Celo et al., 2010; Dabek-Zlotorzynska et al., 2011).
2.3. Data analysis
Data for trend analysis of hourly PM2.5 concentrations from continu-
ous monitoring at Edmonton McIntrye station was obtained in temporal
order of year, month, day, and hour. A cut-off criterion of 80%
124 M.A. Bari, W.B. Kindzierski / Environment International 91 (2016) 122–132
Fig. 1. Location of the Edmonton McIntyre NAPS PM2.5 chemical speciation monitoring station (source: Google Earth Image Landsat © 2015 Google).
completeness will be used as an initial screening step to establish
whether to include an annual dataset in trend analysis. This criterion
represents ~ 7000 hourly values for an annual PM2.5 dataset and is
judged more than adequate for purposes of this study; in addition it is
a criterion similar to that used by others (Blanchard, 1999). A paramet-
ric method was chosen in this study, which tend to be more powerful
than non-parametric tests and has potential to quantify the magnitude
of a trend (McLeod et al., 1991).The method consists of time series linear
regression using various percentiles of a distribution of 1 h concentra-
tions for a pollutant during a year. Benchmarks representing 50th,
65th, 80th, 90th, 95th and 98th percentile PM2.5 concentrations for a
year were identified as response variables from these frequency distri-
butions and used for parametric trend analysis using simple linear re-
gression and hypothesis testing (p b 0.05) to establish the significance
of a trend over the period tested. Zero values and values below the
limit of detection (LOD) were replaced by half of the LOD and missing
values were excluded from the dataset. Further details of trend analysis
method are provided elsewhere (Bari and Kindzierski, 2015a).
In general, data distributions of chemical components derived from
24 h integrated PM2.5 sampling are right skewed and usually not nor-
mally distributed. A Shapiro–Wilks test of normality revealed that
these datasets were log-normally distributed and therefore geometric
means best represented the central tendency of the annual concentra-
tion distributions. A non-parametric method was chosen in this study
to investigate temporal trends in annual geometric mean concentra-
tions of ambient PM2.5 components from 24 h integrated sampling.
The Mann–Kendall test (Mann, 1945; Kendall, 1975) was used to deter-
mine the presence of a monotonic increasing or decreasing trend over
time and the Theil–Sen approach was applied to estimate magnitude
of the rate of change per year (i.e., slope) (Sen, 1968; Gilbert, 1987).
The MAKESENS (Mann–Kendall test for trend and Sen's slope estimate)
template application was used to determine statistical significance
levels (p = 0.001, 0.01, 0.05, and 0.1) of a trend (Salmi et al., 2002)
and results were verified using ProUCL version 5.0 software from the
United States Environmental Protection Agency (U.S. EPA, 2013). The
Mann–Kendall and Theil–Sen approaches have been applied to deter-
mine trends in air pollutants in several studies (Rao et al., 1992;
Ruoho-Airola et al., 2004; Cusack et al., 2012; Saylor et al., 2015; Bari
et al., 2016). Further details of the Mann–Kendall and Theil–Sen ap-
proaches are provided in U.S. EPA (2013). To evaluate year to year
variation in PM2.5 components, a non-parametric Kruskal–Wallis test
(McBean and Rovers, 1998) was performed to determine whether
daily concentrations of PM2.5 and its components were significantly dif-
ferent during at least one year. Monthly and seasonal variations were
also evaluated using the Kruskal–Wallis test.
Application of the Mann–Kendall method is appropriate for detec-
tion of potential trends in air quality data provided that underlining as-
sumptions of the tests are satisfied. The test is suitable for cases where
the trend is monotonic and there is no seasonality and or autocorrela-
tion in the data. A monotonic upward (downward) trend means that
the variable consistently increases (decreases) through time, but the
trend may or may not be linear. Seasonality refers to patterns in data
that depend on or are controlled by season of the year. Seasonal effects
may be related to wind direction, temperature and other meteorological
influences, and emissions disruptions (i.e., planned shutdowns) that
occur the same season each year. Variables used for trending (annual
geometric mean concentrations) were calculated from data drawn
from all seasons of a year. Thus the so-called “effect of season” is the
same for each variable for each year and the non-seasonality assump-
tion should be satisfied. Unplanned emissions disruptions (e.g., upsets
or accidental shutdowns) may occur, however their occurrence is ran-
dom in nature and these events do not violate application of the
Mann–Kendall method for analysis of annual data.
3. Results and discussion
3.1. Ambient levels for PM2.5 components from 24 h sampling
Twenty-four hour ambient PM2.5 speciation data quality was
assessed for validity and species selection for trend analysis. Data qual-
ity for all measured PM2.5 components (n = 50) during the period
2007–2014 is shown in Table S1. Only 34 components, including OC,
EC, secondary pollutants, ions and trace elements, were selected for
trend investigation based on a higher data completeness of at least
50% of samples above the detection limit and source-specific tracers
(e.g., levoglucosan and mannosan for biomass smoke). Data below the
minimum detection limit were replaced by half of the detection limit
and non-detect or missing values were excluded from the dataset.
Table 1 shows descriptive statistics for PM2.5 and selected 34 compo-
nents over the 8-year period (2007–2014). The geometric mean
 M.A. Bari, W.B. Kindzierski / Environment International 91 (2016) 122–132 125

Table 1
Descriptive statistics of PM2.5 and detected 34 chemical components (more than 50% of the samples) for 2007–2014.

Analytical % of samples N % of samples b % of non-detect or No. of Units Gmean Min P10 P25 Median P75 P98 Max
method MDL MDL missing samples

PM2.5 Gravimetric 100 0 0 871 μg/m3 7.22 0.14 3.63 4.95 6.90 10.39 29.42 62.54
Organic carbon (OC) TOR 90 10 0 867 μg/m3 1.68 0.19 0.46 0.99 1.81 3.31 7.57 29.45
Elemental carbon (EC) TOR 94 6 0 867 μg/m3 0.65 0.018 0.18 0.38 0.71 1.31 4.55 10.20
Sulfate (SO2−
4 ) IC 100 0 0 869 μg/m 3
0.61 0.024 0.21 0.33 0.63 1.10 3.82 14.75
Nitrate (NO− 3 ) IC 92 7 1 869 μg/m 3
0.32 0.010 0.06 0.10 0.19 1.03 11.57 20.15
Ammonium (NH+ 4 ) IC 100 0 0 869 μg/m 3
0.32 0.002 0.08 0.14 0.31 0.68 4.17 10.21
Sodium (Na+) IC 67 33 0 869 μg/m3 0.03 0.002 0.01 0.02 0.03 0.06 0.31 2.15
Chloride (Cl−)a IC 47 53 0 869 μg/m3 0.02 0.005 0.01 0.01 0.02 0.05 0.32 2.85
Potassium (K+) IC 85 15 0 869 μg/m3 0.03 0.003 0.01 0.02 0.03 0.05 0.12 0.25
Calcium (Ca+2) IC 88 12 0 869 μg/m3 0.01 0.001 0.00 0.01 0.01 0.01 0.03 0.08
Magnesium (Mg+2) IC 92 7 0 869 μg/m 3
0.04 0.001 0.01 0.02 0.05 0.07 0.20 0.92
Floride (F−) IC 51 17 32 869 μg/m 3
0.00 0.001 0.00 0.00 0.00 0.01 0.02 0.02
Formatea IC 44 43 13 869 μg/m 3
0.02 0.005 0.01 0.01 0.02 0.02 0.06 0.20
Oxalate IC 82 12 6 869 μg/m3 0.04 0.005 0.02 0.03 0.05 0.07 0.16 0.61
Levoglucosan GC–MS 96 0 4 561 ng/m3 40.95 1.75 9.99 18.17 39.74 84.71 416.06 1492
Mannosan GC–MS 60 11 29 561 ng/m3 8.75 0.94 2.84 3.71 8.20 16.81 84.98 347
Aluminum (Al) ICP-MS 73 27 0 833 ng/m3 3.24 0.52 1.04 1.56 3.50 6.74 28.00 488
3
Arsenic (As) ICP-MS 98 2 0 833 ng/m 0.16 0.01 0.06 0.10 0.16 0.26 0.98 12.52
3
Barium (Ba) ICP-MS 100 0 0 833 ng/m 1.56 0.05 0.65 1.01 1.65 2.51 5.16 14.61
3
Cadmium (Cd) ICP-MS 81 19 0 833 ng/m 0.05 0.007 0.02 0.03 0.05 0.09 0.31 1.15
Cobalt (Co) ICP-MS 62 38 0 833 ng/m3 0.05 0.005 0.01 0.02 0.04 0.09 1.16 4.80
Chromium (Cr) ICP-MS 71 28 1 833 ng/m3 0.42 0.026 0.10 0.21 0.43 0.81 2.30 8.97
Copper (Cu) ICP-MS 90 10 0 833 ng/m3 1.49 0.21 0.52 0.99 1.57 2.40 6.47 56
Iron (Fe) ICP-MS 99 1 0 833 ng/m3 12.73 0.52 4.59 7.74 13.57 22.43 49.73 130
Manganese (Mn) ICP-MS 100 0 0 833 ng/m3 2.43 0.05 0.62 1.25 2.93 5.01 11.06 21
Molybdenum (Mo) ICP-MS 74 26 0 833 ng/m3 0.18 0.01 0.04 0.10 0.18 0.34 1.77 10.44
3
Nickel (Ni) ICP-MS 81 18 1 833 ng/m 0.34 0.04 0.07 0.16 0.34 0.74 2.84 11.02
Lead (Pb) ICP-MS 89 11 0 833 ng/m3 0.38 0.05 0.10 0.22 0.37 0.67 2.82 10.44
Antimony (Sb) ICP-MS 99 1 0 833 ng/m3 0.17 0.01 0.08 0.11 0.17 0.25 0.69 1.46
Selenium (Se)a ICP-MS 19 75 6 833 ng/m3 0.08 0.021 0.05 0.05 0.07 0.14 0.32 0.53
Tin (Sn) ICP-MS 61 32 8 833 ng/m3 0.11 0.026 0.03 0.05 0.10 0.18 0.96 3.34
Strontium (Sr) ICP-MS 96 4 0 833 ng/m3 0.26 0.021 0.12 0.18 0.26 0.39 0.85 2.51
3
Titanium (Ti) ICP-MS 55 45 0 833 ng/m 0.32 0.10 0.16 0.16 0.34 0.52 1.15 4.28
3
Vanadium (V) ICP-MS 66 33 1 833 ng/m 0.08 0.01 0.02 0.03 0.06 0.16 2.61 8.66
Zinc (Zn) ICP-MS 93 7 0 833 ng/m3 5.80 0.52 1.47 3.25 6.31 11.37 39.64 226

TOR: Thermal optical reflectance, IC: Ion chromatography, ICP-MS: Inductively-coupled plasma mass spectrometry, GC-MS: Gas chromatography mass spectrometry, P: percentile, MDL:
method detection limit, Gmean: geometric mean, Min: minimum, Max: maximum.
a
Se, Cl− and formate selected due to source specific tracers.

concentration of PM2.5 in Edmonton McIntyre was 7.22 μg/m3 (medi-
an = 6.90 μg/m3, interquartile range, IQR = 4.95–10.39 μg/m3,
range = 0.14–62.54 μg/m3). Concentrations measured at this station
were comparable to other monitoring stations in Edmonton (median
levels: central: 7.0 μg/m3, east 7.4 μg/m3, south 6.9 μg/m3), but lower
than Calgary central (median: 10.1 μg/m3) for the same time period
(CASA, 2015).
The reconstructed mass calculated as the sum of major PM2.5 com-
ponents has been used as a quality assurance tool for gravimetrically
measured mass (Malm et al., 1994; Frank, 2006). PM2.5 components
were grouped as ammonium sulfate-(NH4)2SO4, ammonium nitrate
(NH4NO3), organic matter (OM), EC, crustal matter, trace element ox-
ides (TEO), sodium chloride (NaCl) and particle-bound water (PBW)
and the reconstructed mass was calculated after Malm et al. (1994) as
reported in Dabek-Zlotorzynska et al. (2011) (Table S2). To calculate
OM from measured OC, several methods have been reported in the lit-
erature. OM is a complex mixture of organic compounds originating
from primary sources and secondary formation processes (Seinfeld
and Pandis, 1998) and sources of OM can be attributed to secondary or-
ganic aerosols (SOA) formed from the emissions of fossil fuel combus-
tion, traffic, biomass burning including forest fires and agricultural
burning as well as biogenic volatile organic compounds (VOCs) (Hoyle
et al., 2011). A common practice to represent OM in urban aerosols is
through the use of multiplying factor (k) from the mean value of OM/
OC ratio such as 1.4 (Grosjean and Friedlander, 1975), 1.6 (Malm
et al., 1994), 1.6 ± 0.2 (Turpin and Lim, 2001), and 1.8 (Hand et al.,
2012) for North America. Depending on the sources of organic aerosols
and their degree of aging, the OM/OC ratio can vary by season with sum-
mer maximum and winter minimum values have been reported over
the regions dominated by combustion emissions (Simon et al., 2011;
Xing et al., 2013; Philip et al., 2014). In this study, OM was calculated
using seasonal k values (e.g., summer: 2.1 ± 0.7, winter: 1.8 ± 0.7)
adopted from Dabek-Zlotorzynska et al. (2011) (Table S3). In general,
the higher OM/OC ratio in summer is likely due to the intense photo-
chemistry and higher oxidant levels leading to enhanced SOA formation
in warmer months.
Fig. 2 shows the relative proportions of major groups of PM2.5 com-
ponents for the period of 2007–2014. The sum of major groups
accounted for 81% of the gravimetrically measured PM2.5 mass and
19% of mass was unaccounted. The most dominant component of
PM2.5 at Edmonton was OM, contributing to 45% (3.25 μg/m3) of the
PM2.5 mass. Concentrations of OC ranged from 0.19 to 29.45 μg/m3
with a geometric mean of 1.68 μg/m3 (median 1.81 μg/m3). EC, a prima-
ry pollutant formed in the combustion processes ranged from 0.02 to
10.20 μg/m3 with a geometric mean of 0.65 μg/m3 (median 0.71 μg/m3)
and contributed to 9% of PM2.5 mass.
The second most predominant component in PM2.5 mass was am-
monium sulfates (0.78 μg/m3, 10.8%), which can be present in the atmo-
sphere in the particulate form as secondary sulfate consisting of sulfuric
acid, ammonium bisulfate and ammonium sulfate (Ansari and Pandis,
1998) and can remain airborne for hundreds of kilometers (Tuncel
et al., 1985). Concentrations of SO24 − and NH+ 4 ranged from 0.02 to
14.75 μg/m3 and 0.002–10.21 μg/m3 with geometric means of 0.61 μg/
m3 (median 0.63 μg/m3) and 0.32 μg/m3 (0.31 μg/m3), respectively. In
Alberta, background regional sulfate is found in high abundance due
to oil and gas production (e.g., natural and sour gas extraction, flaring
and processing), other industrial emissions like coal- and gas-fired in-
dustrial boilers for power generation and other non-specific industrial
126 M.A. Bari, W.B. Kindzierski / Environment International 91 (2016) 122–132
Fig. 2. Geometric mean PM2.5 composition (μg/m3) at Edmonton McIntyre for 2007–2014.
sources (Schulz and Kindzierski, 2001). Sulfate was also found as the
abundant mass fraction of ambient fine particles (~11% of PM2.5 mass)
in Edmonton (median 1.0 μg/m3, range 0.01–11.14 μg/m3) and Calgary
(median 1.0 μg/m3, range 0.1–16.20 μg/m3) during a study period of
1985–1995 by Cheng et al. (1998).
Another important contributor was ammonium nitrate, which
accounted for 5.7% (0.41 μg/m3) of the PM2.5 mass. It is assumed that
all nitrates are present as ammonium nitrate (Malm et al., 1994) partic-
ularly in Western Canada, where ammonium ion (NH+ 4 ) is abundant
(Dabek-Zlotorzynska et al., 2011). The geometric mean concentration
of NO− 3 3
3 was 0.32 μg/m (median 0.19 μg/m , range = 0.01–20.15 μg/
m3). Similar levels were found in the Cheng et al. (1998) 1985–1995
study in Edmonton and Calgary with median of 0.20 μg/m3 and maxi-
mum of 15.6 μg/m3, making up almost 5% of PM2.5 mass in both cities.
In Edmonton, the contribution of secondary nitrate to PM2.5 was report-
ed by Jeong et al. (2011) to come from emissions of ammonia and ox-
ides of nitrogen (NOx) from agricultural activities and primary
emissions (vehicle exhaust or industry and local oil refinery sources).
Other components, such as crustal matter (i.e. soil) contributed only
2.1% (0.15 μg/m3) of the PM2.5 mass. Soil components are presumably
emitted from windblown dust sources (soil and road dust resuspension,
construction and agricultural activities) in spring and summer months.
A low inferred contribution of windblown dust would be expected in
winter due to a greater predominance of light winds, and presence of
snow cover and/or wet/frozen ground conditions. The contribution of
NaCl was 0.8% (0.06 μg/m3) of the PM2.5 mass and road salt represents
a likely source during winter months. Median concentrations of K+,
levoglucosan and mannosan (typical markers for biomass burning)
were 0.03 μg/m3 (range 0.003–0.25 μg/m3), 39.74 ng/m3 (range
1.75–1492 ng/m3) and 8.20 ng/m3 (range 0.94–347 ng/m3), respective-
ly and are likely associated with biomass smoke-related sources includ-
ing winter fireplace burning, open pit camp fires and barbeques as well
as wild fires in Alberta and nearby provinces. Concentrations of TEO that
normally occur at very low levels in the environment were found in less
abundance accounting for 0.2% (0.02 μg/m3) of the PM2.5 mass. Concen-
trations of potential carcinogenic elements e.g., As (geomean 0.16 ng/
m3) and Ni (geomean 0.34 ng/m3) were far below the Alberta Ambient
Air Quality Objectives and Guidelines (10 ng/m3 for As and 50 ng/m3 for
Ni, AESRD, 2013). On average, PBW accounted for 7.5% (0.54 μg/m3) of
the PM2.5 mass in Edmonton, similarly found in the earlier study (~6%
of PM2.5 mass) (Cheng et al., 1998).
To evaluate the state of fine particulate air pollution in the City of Ed-
monton relative to national and international metropolitan populations,
levels of PM2.5 and its selected components measured in Edmonton
were compared with major Canadian cities and international
(e.g., United States) urban areas with similar characteristics
(e.g., prairie nature) and/or comparable populations of over 1 million
for the same time period (2007–2014) (Table 2). Our intent was to try
to understand whether PM2.5 levels in Edmonton were large relative
to other urban areas in North America. In general, median PM2.5 concen-
tration in Edmonton was comparable to Montreal in Quebec, Denver in
Colorado and Boston in Massachusetts but lower than Toronto in Ontar-
io, Milwaukee in Wisconsin and Pittsburgh in Pennsylvania. Levels of
secondary pollutants in Edmonton were comparable to Denver, while
lower levels were observed for most trace metals (except for Mn and
Zn) compared to U.S. urban areas. We further evaluated statistically
significant differences in median levels of PM2.5 and its components
between Edmonton and Toronto using paired t-test (Table 3) in order
to assess the accuracy of the media release by the advocacy group
(CAPE). From paired measurements during 2007–2014 time period,
significantly lower levels were observed in Edmonton for PM2.5 (p =
0.001), SO2−4 (p b 0.0001), NO3 (p = 0.036), NaCl and K+ (p b 0.0001)
as well as most trace metals (except Cr, Mn and Ni) compared to levels
found in Toronto. Significantly higher levels of EC (p b 0.0001) were
found in Edmonton than in Toronto, which may possibly be explained
by the higher influence of biomass smoke (including wildfires and
wood-fired heating) and traffic in Edmonton and the surrounding re-
gion during the investigated time period. However, caution is acknowl-
edged when comparing air quality levels in different urban areas across
North America. Factors such as size and density of the population, ex-
tent of development and per capita emissions along with geography
and meteorology and transboundary pollution may influence particu-
late air quality in urban areas. Different methodological factors
e.g., monitoring methods, location of monitoring sites can also influence
results of PM2.5 comparisons across urban areas.
.
3.2. Temporal and seasonal variation for PM2.5 and its components from
24 h sampling
Temporal profiles for 24 h concentrations of PM2.5, OC, EC, secondary
−
pollutants (i.e., SO2− + +
4 , NO3 , and NH4 ), oxalate, K , and trace elements
over the period of 2007–2014 are shown in Figs. S4 and S5 using box–
whisker plots. To evaluate year to year variation in concentrations, a
non-parametric Kruskal–Wallis test was performed to determine
whether daily concentrations of PM2.5 and major components were
 M.A. Bari, W.B. Kindzierski / Environment International 91 (2016) 122–132 127

Table 2
Comparison of median levels of selected PM2.5 components with metropolitan cities in North America for the time period 2007–2014.

Units Edmonton Torontoa Montreala Bostonb Denverb Milwaukeeb Pittsburghb

Alberta Ontario Québec Massachusetts Colorado Wisconsin Pennsylvania

PM2.5 μg/m3 6.90 8.26 7.10 7.55 6.91 9.44 12.74

OC μg/m3 1.81 2.17 1.53 1.59 1.60 1.93 2.60
EC μg/m3 0.71 0.56 0.46 0.56 0.66 0.47 1.21
SO2−
4 μg/m3 0.63 1.30 0.98 1.43 0.69 1.48 3.74
NO−3 μg/m3 0.19 0.22 0.15 1.20 0.54 1.40 1.12
NH+ 4 μg/m3 0.31 0.58 0.37 0.54 0.28 0.75 1.76
Na+ μg/m3 0.03 0.05 0.05 0.08 0.03 0.14 0.07
Cl− μg/m3 0.02 0.04 0.04 na na na na
K+ μg/m3 0.03 0.04 0.04 0.02 0.02 0.03 0.06
As ng/m3 0.16 0.30 0.38 0.42 0.23 0.59 2.61
Ba ng/m3 1.65 1.67 1.16 5.06 5.10 5.06 5.95
Cd ng/m3 0.05 0.07 0.10 1.49 1.83 1.43 1.71
Cr ng/m3 0.43 0.25 0.26 1.13 0.86 1.78 1.57
Cu ng/m3 1.57 1.88 1.46 4.87 2.54 2.29 5.74
Mn ng/m3 2.93 1.03 1.15 0.84 1.80 2.06 2.00
Ni ng/m3 0.34 0.18 0.26 1.03 0.28 0.98 0.92
Pb ng/m3 0.37 1.25 1.41 0.97 0.65 1.33 10.55
Sb ng/m3 0.17 0.25 0.28 19.69 19.50 19.50 19.26
Se ng/m3 0.07 0.23 0.17 1.09 1.15 1.16 4.08
V ng/m3 0.06 0.18 0.37 1.63 1.58 1.63 1.60
Zn ng/m3 6.31 8.15 9.16 5.39 4.62 7.99 22.25

na: not measured.

a
Toronto: Gage Institute station (NAPS ID: 60427), Montreal: Riviére-des Prairies station (NAPS ID: 50129) (Environment Canada, 2015a).
b
United States Environmental Protection Agency's Air Quality System (AQS) data (U.S. EPA, 2015).

significantly different during at least one year (Table 4). We found no
significant variation in year to year concentrations for PM2.5 and ammo-
nium concentrations (p N 0.05). While statistically significant differ-
ences (p b 0.05) in year to year concentrations were observed for OC,
−
EC, SO2− +
4 , NO3 , oxalate, K , levoglucosan and some trace metals (As,
Pb, Se, V, Zn).
Month to month variation in PM2.5, OC, EC, secondary pollutants, ox-
alate, K+ and levoglucosan over the time period of 2007–2014 is shown
in Fig. 3 using box-whisker plots (other components are shown in
Fig. S6). We found statistically significant differences in month to
month concentrations for PM2.5 and all components except for Ni
(p b 0.05) (Table 4). Significantly higher PM2.5 concentrations were ob-
served during winter than in summer months (p b 0.0001), which can
possibly be explained by a combination of atmospheric conditions and
anthropogenic emissions (e.g., vehicle and industrial emissions, road
Table 3
Differences in median levels of PM2.5 and its selected components between Edmonton
(EDM) and Toronto (TOR) for 2007–2014.
de-icing salt). During winter, high atmospheric stability, low wind
speed and morning ground-based temperature inversions with low
mixing layer heights can favor accumulation of pollutants resulting in
higher particle concentrations than in summer (Myrick et al., 1994).
Similar winter patters for PM2.5 have been found in Edmonton in the
earlier Cheng et al. (1998) study of 1985–1995 and during 2004–2006
in Golden, British Columbia (Jeong et al., 2008), where poor atmospher-
ic dispersion conditions due to frequent temperature inversions con-
tributed to elevated PM2.5 concentrations in winter months. The
diurnal hourly average pattern for PM2.5 (Fig. S7) showed very small
maxima occurring between 11:00 to 13:00 and 19:00 to 21:00 during
winter and between 8:00 to 10:00 and 23:00 to 1:00 during summer.
Minima were observed in the early morning hours (between 4:00 and
6:00). In many instances, maximum hourly PM2.5 concentrations
ranged from 100 μg/m3 to 435 μg/m3 occurring between the months
of May and August. A possible explanation is long-range transport of
smoke particles from natural events e.g., wildfires. Several events oc-
curred when maximum 24 h concentrations exceeded the Canada-
Wide Standard of 30 μg/m3 (e.g., 62 μg/m3 on August 21, 2010)

Units Difference in median Median p-Value

Table 4
Paired t-test Yearly and monthly variation (p-value, Kruskal–Wallis test) in PM2.5 and selected tracer
PM2.5 μg/m3 −1.36 EDM b TOR 0.001 components in Edmonton over the period 2007–2014.
OC μg/m3 −0.36 0.487
Annual variation Monthly variation Seasonal variation
EC μg/m3 0.15 EDM N TOR b0.0001
p-Value p-Value (winter vs summer)
SO2−
4 μg/m3 −0.67 EDM b TOR b0.0001
p-Value
NO−3 μg/m3 −0.03 EDM b TOR 0.036
NH+ 4 μg/m3 −0.27 EDM b TOR b0.0001 PM2.5 0.270 b0.0001 b0.0001
Na+ μg/m3 −0.02 EDM b TOR b0.0001 OC b0.0001 b0.0001 0.485
Cl− μg/m3 −0.02 EDM b TOR b0.0001 EC b0.0001 b0.0001 0.012
K+ μg/m3 −0.01 EDM b TOR b0.0001 SO2−
4 0.037 b0.0001 0.069
As ng/m3 −0.14 EDM b TOR b0.0001 NO− 3 b0.0001 b0.0001 b0.0001
Ba ng/m3 −0.02 EDM b TOR 0.025 NH+ 4 0.148 b0.0001 b0.0001
Cd ng/m3 −0.02 EDM b TOR 0.007 Oxalate b0.0001 b0.0001 b0.0001
Cr ng/m3 0.18 EDM N TOR b0.0001 K +
b0.0001 b0.0001 b0.0001
Cu ng/m3 −0.31 EDM b TOR 0.007 Levoglucosan 0.047 b0.0001 b0.0001
Mn ng/m3 1.90 EDM N TOR b0.0001 Mannosan 0.241 b0.0001 b0.0001
Ni ng/m3 0.16 EDM N TOR b0.0001 As 0.047 0.025 0.004
Pb ng/m3 −0.88 EDM b TOR b0.0001 Ni 0.710 0.418 0.967
Sb ng/m3 −0.08 EDM b TOR b0.0001 Pb b0.0001 b0.0001 0.007
Se ng/m3 −0.16 EDM b TOR b0.0001 Se b0.0001 b0.0001 b0.0001
V ng/m3 −0.12 0.293 V b0.0001 b0.0001 0.660
Zn ng/m3 −1.84 EDM b TOR 0.046 Zn 0.013 b0.0001 b0.0001
128 M.A. Bari, W.B. Kindzierski / Environment International 91 (2016) 122–132
Fig. 3. Monthly profiles of 24 h concentrations of PM2.5, OC, EC, secondary pollutants, oxalate, and K+ over the period 2007–2014 and levoglucosan for 2009–2014 using box-whisker plots.
Boxes represent 25th (lower quartile) and 75th (upper quartile) percentile values, with median values as lines across the boxes, geometric mean values as round black ball and minimum
and maximum concentrations as whiskers.
suggesting an influence of long-range transport of wildfire smoke epi-
sodes occurring every year from elsewhere in Alberta and nearby prov-
inces such as British Columbia, Saskatchewan, and Manitoba.
No significant variation was found in OC concentrations between
summer and winter months (p = 0.485). The highest OC concentration
of 30 μg/m3 was found in August 2010, which was due the influence of
out-of-region wildfire smoke. A significant variation was found in EC
concentrations (p = 0.012) with comparatively higher concentrations
in winter (geomean 0.90 μg/m3) than in summer (geomean 0.77 μg/
m3). An apparent seasonal variation was found for SO2− 4 with higher
concentrations in winter months. In general, a summer high pattern
would be expected for SO2− 4 concentrations as observed in eastern Ca-
nadian cities (Dabek-Zlotorzynska et al., 2011) possibly due to en-
hanced photochemical reactions associated with higher temperature
and intense solar radiation. In Edmonton, geometric mean SO−2 4 con-
centrations were 0.66 μg/m3 (median 0.69 μg/m3) in winter and
0.57 μg/m3 (median 0.61 μg/m3) in summer, however the difference
was not statistically significant (p = 0.069). The observed apparent
higher levels in winter months are presumably due to dominant hetero-
geneous oxidation of SO2 from local sources and regional transport of oil
and gas activities coupled with stable weather and inversion conditions
during winter.
NO−3 levels showed a strong and significant seasonal variability
(p b 0.0001) with higher concentrations during colder months in winter
(geomean 1.39 μg/m3) than in summer (geomean 0.07 μg/m3). This is
probably due to low thermal stability of ammonium nitrate in warmer
months (Adams et al., 1999; Schaap et al., 2002).The influence of local
vehicle exhaust and industrial sources are likely associated with higher
winter levels of NO− −
3 in Edmonton (Jeong et al., 2011). Like NO3 , NH4
+
levels also exhibited a similar strong seasonal pattern with elevated
concentrations in winter (geomean 0.59 μg/m3) compared to summer
(geomean 0.21 μg/m3).
Oxalate exhibited strong seasonal variability (p b 0.0001) with 2-
fold higher concentrations during warmer months in summer
(geomean 0.06 μg/m3) than in winter (geomean 0.03 μg/m3). Oxalate
can be emitted from several primary and secondary emission sources
such as fuel oil combustion, biomass burning and biogenic sources
(Chebbi and Carlier, 1996). Elevated levels in summer months suggest
the formation of SOA due to photochemical activity. Concentrations of
K+ exhibited a strong and significant seasonal variability (p b 0.0001)
with higher concentrations in winter (geomean 0.04 μg/m3) particularly
November through February than in summer (geomean 0.02 μg/m3).
The City of Edmonton transportation department does not use potassi-
um in sand/salt mixture that is applied to Edmonton roadways for win-
ter maintenance (Carnegie, 2015). Seasonal patterns with high winter
levels likely reflect wood smoke origins more than other potential
sources (Ming-Yi et al., 2015). Significantly higher concentrations of
biomass smoke tracers e.g., levoglucosan and mannosan were observed
in winter (geomean 82.5 ng/m3, 13.1 ng/m3) than in summer (geomean
26.5 ng/m3, 6.5 ng/m3). This suggests an influence of emissions from
residential wood stove/fireplace heating and/or agricultural slash burn-
ing during winter months in Edmonton and the surrounding region.
Some peak concentrations of K+, levoglucosan and mannosan were
also evident during summer months (e.g., July, August), indicating the
contribution of long range transport of smoke from wildfires episodes
in Alberta and nearby provinces such as British Columbia, Saskatchewan
and Manitoba during the study period. Evidence of one such episode is
presented in Bari et al. (2015), where large wildfires smoke poured
across Edmonton on August 18–21, 2010 from British Columbia
resulting high concentrations of 24-h PM2.5 values (N 150 μg/m3) in Ed-
monton stations.
Similar to K+, concentrations of Zn exhibited strong and significant
seasonal variability (p b 0.0001) with higher concentrations in winter
(geomean 10.39 μg/m3) than in summer (geomean 4.89 μg/m3). Zn is
a major trace element of wood (Sippula et al., 2009; Chalot et al.,
2012) and others have reported high Zn emissions from waste wood
combustion (Wellinger et al., 2012) and that increased concentrations
of Zn occur in ambient air affected by wood combustion (Echalar
et al., 1995). This also supports a winter time influence of emissions
from wood burning in Edmonton and the surrounding region and is
an issue worthy of further investigation. Significant seasonal variations
were also found for trace elements e.g., As, Pb, and Se (Fig. S6). While
 M.A. Bari, W.B. Kindzierski / Environment International 91 (2016) 122–132
Ni and V displayed no significant seasonal variation suggesting an influ-
ence from sources such as local refineries and regional oil and gas activ-
ities throughout the year.
3.3. Trends in ambient PM2.5 concentrations from continuous monitoring
Fig. 4a shows temporal trends in annual hourly PM2.5 concentration
percentiles from continuous monitoring (TEOM-FDMS) at Edmonton
McIntyre station using parametric (linear regression) approach. No
change was observed for PM2.5 concentration percentiles over the 8-
year period (2007–2014) (Table S4). This is consistent with the results
of Bari and Kindzierski (2015b) obtained for a longer time period
(1998–2014) at Edmonton central and east stations, where no change
was also observed for PM2.5. In contrast, a non-parametric approach
(Fig. 4b) from 24 h gravimetric measurements suggested a slight down-
ward trend for the PM2.5 geometric mean concentration with an annual
decrease of − 0.05 μg/m3 (0.7% per year) over the 2007–2014 period,
however the trend was not statistically significant (p = 0.268). Non-
parametric tests are alternatives to parametric tests when assumptions
for parametric tests cannot be met. Although these methods make
fewer assumptions regarding the distribution of data, they tend to be
less powerful than parametric tests, and are not robust to homogeneity
of variance of data (Day and Quinn, 1989; McLeod et al., 1991). It is
noted that if a true temporal trend for any environmental parameter is
strong enough, it will be detected regardless of the statistical approach
(parametric or non-parametric) used.
3.4. Trends in PM2.5 components from 24 h sampling
A summary of trends in annual geometric mean concentrations of
higher detected PM2.5 components (i.e., species detected in N50% of
samples) is shown in Table 5 for the 2007–2014 period. Trend plots
for OC, EC, secondary pollutants, NaCl, K+ and oxalate are shown in
Fig. 5 and selected trace elements are shown in Fig. 6. Trend plots for re-
maining trace elements are shown in Fig. S8. Concentrations of OC and
EC showed downward trends at 95% significance level with annual de-
creases of −0.14 and −0.15 μg/m3 per year, (−7.3% and −12.4%), re-
spectively. As reported previously, statistically significant variation
(p b 0.05) in year to year concentrations were observed for OC, EC. Max-
imum values of OC and EC were recorded in 2008 (4.6 and 3.3 μg/m3, re-
spectively) and minimum were recorded in 2014 (0.93 and 0.08 μg/m3,
respectively). The observed decrease in OC and EC concentrations is
likely due to reduction OC/EC emissions or precursors from primary
emission sources such as vehicle exhaust and consistent with observed
decline of traffic-related gaseous pollutants like NO2 and CO over the
Fig. 4. Temporal trends in PM2.5 concentrations at Edmonton McIntyre using (a) parametric (1 h measurements) and (b) non-parametric approach (from 24 h graviemetric
measurements).
129
last 17 years at other Edmonton stations (Bari and Kindzierski,
2015b). As stated previously, OM and EC accounted for 45% and 9% of
the total PM2.5 mass on average. It is also likely that the observed reduc-
tion of OC, EC may partly be attributed to the reduction in PM2.5 mass.
At Edmonton McIntyre, secondary particles e.g., SO24 − and NH+ 4
showed downward trends with small annual decreases of − 0.01 and
− 0.005 μg/m3 per year, respectively, while NO− 3 showed a upward
trend with a small increase of 0.004 μg/m3 (1.2%) per year. However,
they are not statistically significant. The use of road salt in winter as
de-icing or anti-icing chemicals is common practice in Edmonton for
winter road maintenance. A statistically significant upward trend
(p = 0.05) was observed for NaCl with an annual increase of 0.005 μg/
m3 per year (10.7% per year). Concentrations of a biomass burning trac-
er K+ showed a small non-significant increasing trend of 0.001 μg/m3
(1.8%) per year. The influence of long-range transport of wildfires
smoke frequently occurring in recent years in Alberta and nearby prov-
inces e.g., British Columbia, Saskatchewan is likely associated with this.
Oxalate exhibited a statistically significant downward trend with an an-
nual decrease of −0.005 μg/m3 (−7.9%) per year.
Out of PM2.5 elements Ba showed a statistically significant down-
ward trend (p = 0.01) with annual decrease of −0.07 ng/m3 (−3.8%)
per year. Ba can be emitted from exhaust and non-exhaust emissions
of vehicles (Lee and Hopke, 2006; Lough et al., 2005; Sternbeck et al.,
2002). Most trace elements (As, Co, Cr, Mn, Ni, Sb, Se) considered poten-
tially toxic or hazardous by regulatory agencies, displayed non-
significant trends. Ambient concentrations of other potentially toxic el-
ements e.g., Pb and Cd exhibited significant downward trends (p values
of 0.05 and 0.054, respectively) with small annual decreases of −0.03
and − 0.004 ng/m3 (− 6.1% and − 5.5%) per year, respectively. Other
trace elements e.g., V, Cu and Zn exhibited non-significant downward
trends with annual decreases ranging from − 0.43 to − 0.001ng/m3
(−5.6% to −0.8%) per year.
A number of limitations exist in the current study. We did not inves-
tigate the dependency between emissions and ambient concentrations
over the study period. In order to understand the state of inventory of
emissions to air in Alberta Capital Airshed, we summarized the available
annual industrial emissions for PM2.5 and other criteria air pollutants
and major trace elements reported by Environment Canada's National
Pollutant Release Inventory (NPRI) over the 2003–2014 period. This
data is shown in the Supplemental material. It is noteworthy that no
similar historical emission inventory data is available on other poten-
tially important emission sources in this region e.g., automobile emis-
sions and residential fuel wood combustion. Only province-wide
historical emission inventory data is available for these potentially im-
portant emission sources (Environment Canada, 2015b). Further
130 M.A. Bari, W.B. Kindzierski / Environment International 91 (2016) 122–132

Table 5
Temporal trends (8-year) in ambient concentrations of PM2.5 components at Edmonton McIntyre (Theil–Sen trend test at 95% confidence coefficient).

Unit 2007 Gmean 2014 Gmean 2007–2014 slope unit/year % change per year Confidence level Level of significance
p-Value

OC μg/m3 1.53 0.93 −0.14 −7.3 95% 0.05⁎

EC μg/m3 0.51 0.08 −0.15 −12.4 95% 0.05⁎
SO2−
4 μg/m3 0.59 0.59 −0.01 −1.5 n.s. 0.268
NO−3 μg/m3 0.29 0.50 0.004 1.2 n.s. 0.548
NH+ 4 μg/m3 0.34 0.33 −0.005 −1.5 n.s. 0.268
NaCl μg/m3 0.05 0.09 0.005 10.7 95% 0.05⁎
K+ μg/m3 0.04 0.03 0.001 1.8 n.s 0.548
Oxalate μg/m3 0.05 0.03 −0.005 −7.9 95% 0.05⁎
Al ng/m3 2.77 1.75 −0.16 −4.1 n.s. 0.355
As ng/m3 0.16 0.14 −0.002 −1.3 n.s. 0.268
Ba ng/m3 1.83 1.41 −0.07 −3.8 99% 0.01⁎⁎
Cd ng/m3 0.06 0.04 −0.004 −5.5 90%a 0.054+
Co ng/m3 0.05 0.04 −0.003 −5.2 n.s. 0.089
Cr ng/m3 0.43 0.36 −0.01 −1.3 n.s. 0.451
Cu ng/m3 1.30 1.52 −0.02 −1.3 n.s. 0.355
Fe ng/m3 10.99 10.11 −0.07 −0.5 n.s. 0.452
Mn ng/m3 2.24 2.34 0.05 2.2 n.s. 0.193
Mo ng/m3 0.26 0.20 0.002 0.9 n.s. 0.451
Ni ng/m3 0.31 0.31 0.01 2.1 n.s. 0.193
Pb ng/m3 0.45 0.24 −0.03 −6.1 95% 0.05⁎
Sb ng/m3 0.18 0.16 −0.004 −2.1 n.s. 0.054
Se ng/m3 0.14 0.06 −0.01 −6.9 n.s. 0.113
Sn ng/m3 0.07 0.08 −0.001 −1.1 n.s. 0.451
Sr ng/m3 0.19 0.24 0.01 3.1 n.s. 0.193
Ti ng/m3 0.26 0.22 −0.01 −4.0 n.s. 0.268
V ng/m3 0.07 0.08 −0.001 −0.8 n.s. 0.548
Zn ng/m3 6.30 4.51 −0.43 −5.6 n.s. 0.054

n.s. = not significant.

a
Theil–Sen trend test at 90% confidence coefficient.
⁎⁎ p ≤ 0.01.
⁎ p ≤ 0.05.
+
p ≤ 0.1.

investigation of historical emission for these sources in Edmonton and
the surrounding region is needed to better understand the relation be-
tween emissions and ambient concentrations of PM2.5 and its chemical
components.
It is not surprising that most of the observed trends for PM2.5 and its
components were not statistically significant. A lack of significant statis-
tical trends for PM2.5 and its components may indicate absence of true
trends, changes in ambient levels that are too small to detect using
our methods or a time period that is too short. Indications of statistically
significant trends for datasets that are limited in duration (years) or
where only small changes are occurring from year-to-year (i.e., less a
couple of percent) are unlikely to be truly representative of the trends
that are actually occurring. This would only be captured with a much
longer time period e.g., two to three decades as suggested by
Weatherhead et al. (1998).
4. Conclusion
To understand the state of particulate air pollution in the City of Ed-
monton within the Capital Region of Alberta, temporal trends in ambi-
ent PM2.5 and its chemical components were investigated over a 8-
year (2007–2014). No statistically significant change was observed for
ambient PM2.5 concentrations over the period 2007–2014, while signif-
icant decreasing trends were found for organic carbon, elemental

Fig. 5. Temporal trends in annual geometric mean concentrations of PM2.5, OC, EC, secondary pollutants, NaCl, K+ and oxalate for 8-year period (2007–2014) at Edmonton McIntyre.
 M.A. Bari, W.B. Kindzierski / Environment International 91 (2016) 122–132
Fig. 6. Temporal trends in annual geometric mean concentrations of selected trace components for 2007–2014 at Edmonton McIntyre.
carbon, oxalate, barium, lead and cadmium. A statistically significant in-
creasing trend was observed for sodium chloride indicating an increase
of de-icing salt contribution for winter road maintenance in recent
years. Concentrations of K+ and zinc exhibited strong and significant
seasonal variability with higher concentrations in winter than in sum-
mer. Seasonal patterns with high winter levels of these trace elements
likely reflect wood smoke origins more than other potential sources in
Edmonton and the surrounding region and further research of the
wood smoke contribution to Alberta Capital Region winter time air
quality is recommended. No statistically significant changes were ob-
served for all other chemical components examined. Notwithstanding
robust population growth that has occurred in Edmonton, these find-
ings reveal that particulate air quality and corresponding trace elements
in Edmonton's air has been unchanged or improved over the investigat-
ed period (2007–2014). Longer-term air quality monitoring at least
over several decades is needed to establish whether trends reported
here are actually occurring. However, the data presented here provides
reasonably strong baseline information and key insights about current
levels and observed temporal trends in ambient PM2.5 and its compo-
nents in the City of Edmonton. The methods used in this study are sim-
ple statistical approaches to demonstrate temporal trends in ambient
fine particulate matter and its components in Edmonton and can be ap-
plied to other major urban areas in North America and elsewhere.
Acknowledgement
Both authors contributed equally to the development of this article.
This study was funded by TransAlta Corporation (Calgary, AB) under the
University of Alberta research grant agreement #0027778. The funder
had no role in study design, data collection and analysis, decision to
publish, or preparation of the manuscript. The authors would like to
acknowledge Environment Canada and Clean Air Strategic Alliance
(CASA) for the provision of PM2.5 and its chemical composition data
for this investigation.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.envint.2016.02.033.
References
Adams, P.J., Seinfeld, J.H., Koch, D.M., 1999. Global concentrations of tropospheric sul-
phate, nitrate, and ammonium aerosol simulated in a general circulation model.
J. Geophys. Res. 104, 13791–13823.
131
Alberta Environment, 2013. Monitoring PM2.5. Newer technology, better data, same
air quality. http://environment.gov.ab.ca/info/library/8862.pdf [accessed 15 July
2015].
Alberta Environment and Parks, 2015. Alberta: Air zones report 2011–2013. http://esrd.
alberta.ca/air/management-frameworks/canadian-ambient-air-quality-standards-
for-particulate-matter-and-ozone/documents/AlbertaAirZonesReport-2011-13-
Sep2015.pdf [accessed 15 July 2015].
Alberta Environment and Sustainable Resource Development (AESRD), 2013. Alberta En-
vironment Ambient Air Quality Objectives. http://environment.gov.ab.ca/info/library/
5726.pdf [accessed 19 Jun 2015].
Alberta Environment and Sustainable Resource Development (AESRD), 2014. Capital Re-
gion Fine Particulate Matter Science Report. December 2014. http://esrd.alberta.ca/
focus/cumulative-effects/capital-region-industrial-heartland/documents/
CapitalRegion-PM-ScienceReport-Dec2014.pdf [accessed 15 July 2015].
Alberta Transportation, 2008. Number of Motorized Vehicles Registered in Alberta as of
March 31, 2004–2008. Government of Alberta, Alberta Transportation, Office of Traf-
fic Safety. http://www.transportation.alberta.ca/Content/docType47/Production/
VehReg2008.pdf [accessed 10 September 2015].
Alberta Transportation, 2012. Number of Motorized Vehicles Registered in Alberta as of
March 31, 2008–2012. Government of Alberta, Alberta Transportation, Office of Traf-
fic Safety. http://www.transportation.alberta.ca/Content/docType47/Production/
vehreg2012.pdf [accessed 10 September 2015].
Alberta Transportation, 2014. Number of Motorized Vehicles Registered in Alberta as of
March 31, 2010–2014. Government of Alberta, Alberta Transportation, Office of Traf-
fic Safety. http://www.transportation.alberta.ca/Content/docType47/Production/
VehReg2014.pdf [accessed 10 September 2015].
Ansari, A.S., Pandis, S.N., 1998. Response of inorganic PM to precursor concentrations. En-
viron. Sci. Technol. 32, 2706–2714.
Bari, M.A., Kindzierski, W.B., 2015a. Fifteen-year trends in criteria air pollutants in oil
sands communities of Alberta, Canada. Environ. Int. 74, 200–208.
Bari, M.A., Kindzierski, W.B., 2015b. Investigation of Fine Particulate Matter Characteris-
tics and Sources in Edmonton, Alberta. Final Report, University of Alberta, Edmonton,
Alberta.
Bari, M.A., Kindzierski, W.B., Wallace, L.A., Wheeler, A.J., MacNeill, M., Héroux, M.-E., 2015.
Indoor and outdoor levels and sources of submicron particles (PM1) at homes in Ed-
monton, Canada. Environ. Sci. Technol. 49, 6419–6429.
Bari, M.A., Kindzierski, W.B., Spink, D., 2016. Twelve-year trends in ambient concentra-
tions of volatile organic compounds in a community of the Alberta Oil Sands Region,
Canada. Environ. Int. 91, 40–50.
Blanchard, L.C., 1999. Methods for attributing ambient air pollutants to emission sources.
Annu. Rev. Energy Environ. 24, 329–365.
Burnett, R.T., Pope, C.A., Ezzatt, M., Olives, C., Lim, S.S., Mehta, S., 2014. An integrated risk
function for estimating the global burden of disease attributable to ambient fine par-
ticulate matter exposure. Environ. Health Perspect. 122, 397–403.
Canadian Council of Ministers of the Environment (CCME), 2000. Canada-Wide Standards
for Particulate Matter (PM) and Ozone. Canadian Council of Ministers of the Environ-
ment, Quebec City, Quebec.
Canadian Council of Ministers of the Environment (CCME), 2012. Guidance Document on
Achievement Determination Canadian Ambient Air Quality Standards for Fine Partic-
ulate Matter and Ozone. Canadian Council of Ministers of the Environment, Winni-
peg, Manitoba.
Canadian Environmental Protection Act (CEPA), 1999. Statutes of Canada. Act Assented to
September 14, 1999. Queen's Printer, Ottawa.
Carnegie, L., 2015. City of Edmonton Transportation Department, General Supervisor,
Contracts & Materials, Roadway Maintenance. Personal communication with W.
Kindzierski, 2 November 2015.
Cassee, F.R., He′ roux, M.-E., Gerlofs-Nijland, M.E., Kelly, F.J., 2013. Particulate matter be-
yond mass: recent health evidence on the role of fractions, chemical constituents
and sources of emission. Inhal. Toxicol. 25 (14), 802–812.
132 M.A. Bari, W.B. Kindzierski / Environment International 91 (2016) 122–132
Celo, V., Dabek-Zlotorzynska, E., Mathieu, D., Okonskaia, I., 2010. Validation of simple
microwave-assisted acid digestion method using microvessels for analysis of trace el-
ements in atmospheric PM2.5 in monitoring and fingerprinting studies. Open Chem.
Biomed. Meth. J. 3, 141–150.
Chalot, M., Blaudez, D., Rogaume, Y., Provent, A.S., Pascual, C., 2012. Fate of trace elements
during the combustion of phytoremediation wood. Environ. Sci. Technol. 46,
13361–13369.
Chebbi, A., Carlier, P., 1996. Carboxylic acids in the troposphere, occurrence, sources, and
sinks: a review. Atmos. Environ. 24, 4233–4249.
Cheng, L., Sandhu, H.S., Angle, R.P., Myrick, R.H., 1998. Characteristics of inhalable partic-
ulate matter in Alberta cities — results for 1988–1989 summers. Atmos. Environ. 32,
3835–3844.
Cheung, H.-C., Wang, T., Baumann, K., Guo, H., 2005. Influence of regional pollution out-
flow on the concentrations of fine particulate matter and visibility in the coastal
area of southern China. Atmos. Environ. 39, 6463–6474.
City of Edmonton, 2015. Population History. http://www.edmonton.ca/city_government/
facts_figures/population-history.aspx [accessed 25 August 2015].
Clean Air Strategic Alliance (CASA), 2015. Alberta Ambient Air Data Management System,
CASA Data Warehouse. http://www.casadata.org [accessed 02 June 2015].
Cusack, M., Alastuey, A., Pérez, N., Pey, J., Querol, X., 2012. Trends in particulate matter
(PM2.5) and chemical composition at a regional background site in the Western Med-
iterranean over the last nine years (2002− 2010). Atmos. Chem. Phys. 12,
8341–8357.
Dabek-Zlotorzynska, E., Dann, T.F., Martinelango, P.K., Celo, V., Brook, J.R., Mathieu, D., et al.,
2011. Canadian National Air Pollution Surveillance (NAPS) PM2.5 speciation program:
methodology and PM2.5 chemical composition for the years 2003 to 2008. Atmos.
Environ. 45, 673–686.
Day, R.W., Quinn, G.P., 1989. Comparison of treatments after an analysis of variance in
ecology. Ecol. Monogr. 59 (4), 433–463.
Dominici, F., Greenstone, M., Sunstein, C.R., 2014. Particulate matter matters. Science 344,
257–259.
Dominici, D., Wang, Y., Correia, A.W., Ezzati, M., Pope III, C.A., Dockery, D.W., 2015. Chem-
ical composition of fine particulate matter and life expectancy in 95 US counties be-
tween 2002 and 2007. Epidemiol. 26, 556–564.
Echalar, F., Gaudichet, A., Cachier, H., Artaxo, P., 1995. Aerosol emissions by tropical forest
and savanna biomass burning — characteristic trace-elements and fluxes. Geophys.
Res. Lett. 22 (22), 3039–3042.
Environment Canada, 2015a. NAPS data products. http://maps-cartes.ec.gc.ca/rnspa-
naps/data.aspx [accessed 5 September 2015].
Environment Canada, 2015b. Air Pollutant Emission Inventory (APEI). http://ec.gc.ca/
inrp-npri/donnees-data/ap/index.cfm?lang=En [accessed 25 February 2016].
Forsberg, B., Hansson, H.C., Johansson, C., Areskoug, H., Persson, K., Jarvholm, B., 2005.
Comparative health impact assessment of local and regional particulate air pollutants
in Scandinavia. Ambio 34, 11–19.
Forster, P., Ramaswamy, V., Artaxo, P., Berntsen, T., Betts, R., Fahey, D.W., et al., 2007.
Changes in atmospheric constituents and in radiative forcing. In: Solomon, S. (Ed.),
Climate Change 2007. The Physical Science Basis. Contribution of Working Group I
to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change.
Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA.
Frank, N.H., 2006. Retained nitrate, hydrated sulfates, and carbonaceous mass in federal
reference method fine particulate matter for six U.S. cities. J. Air Waste Manage.
Assoc. 56, 500–511.
Gilbert, R.O., 1987. Statistical Methods for Environmental Pollution Monitoring. Van
Nostrand Reinhold, New York.
Government of Canada, 2013. National Air Pollution Surveillance Program (NAPS). http://
www.ec.gc.ca/rnspa-naps/Default.asp?lang=Enandn=5C0D33CF-1 [accessed 15 Au-
gust 2015].
Grosjean, D., Friedlander, S.K., 1975. Gas-particle distribution factors for organic and other
pollutants in Los Angeles atmosphere. J. Air Pollut. Control Assoc. 25, 1038–1044.
Hand, J.L., Schichtel, B.A., Pitchford, M., Malm, W.C., Frank, N.H., 2012. Seasonal composi-
tion of remote and urban fine particulate matter in the United States. J. Geophys. Res.
Atmos. 117, D05209.
Hoyle, C.R., Boy, M., Donahue, N.M., Fry, J.L., Glasius, M., Guenther, A., et al., 2011. A review
of the anthropogenic influence on biogenic secondary organic aerosol. Atmos. Chem.
Phys. 11, 321–343.
Jeong, C.-H., Evans, G.J., Dann, T., Graham, M., Herod, D., Dabek-Zlotorzynska, E., et al.,
2008. Influence of biomass burning on wintertime fine particulate matter: source
contribution at a valley site in rural British Columbia. Atmos. Environ. 42, 3684–3699.
Jeong, C.-H., McGuire, M.L., Herod, D., Dann, T., Dabek-Zlotorzynska, E., Wang, D., et al.,
2011. Receptor model based identification of PM2.5 sources in Canadian cities.
Atmos. Pollut. Res. 2, 158–171.
Kendall, M.G., 1975. Rank Correlation Methods. Griffin, London, UK.
Lee, J.H., Hopke, P.K., 2006. Apportioning sources of PM2.5 in St. Louis, MO using speciation
trends network data. Atmos. Environ. 40, 360–377.
Lippmann, M., 2014. Toxicological and epidemiological studies of cardiovascular effects of
ambient air fine particulate matter (PM2.5) and its chemical components: coherence
and public health implications. Crit. Rev. Toxicol. 44, 299–347.
Lough, G.C., Schauer, J.J., Park, J.S., Shafer, M.M., Deminter, J.T., Weinstein, J.P., 2005. Emis-
sions of metals associated with motor vehicle roadways. Environ. Sci. Technol. 39,
826–836.
Malm, W.C., Sisler, J.F., Huffman, D., Eldred, R.A., Cahill, T.A., 1994. Spatial and seasonal
trends in particle concentration and optical extinction in the United States.
J. Geophys. Res. 99, 1347–1370.
Mann, H.B., 1945. Nonparametric tests against trend. Econometrica 13, 245–259.
McBean, E.A., Rovers, F.A., 1998. Statistical Procedures for Analysis of Environmental Mon-
itoring Data and Risk Assessment. Prentice Hall PTR, Upper Saddle River, NJ (313 pp.).
McLeod, A.I., Hipel, K.W., Bodo, B., 1991. Trend analysis methodology for water quality
time series. Environmetrics 2, 169–200.
Ming-Yi, T., Hoek, G., Eeftens, M., de Hoogh, K., Beelen, R., Beregszaszi, T., et al., 2015. Spa-
tial variation of PM elemental composition between and within 20 European study
areas — results of the ESCAPE project. Environ. Int. 84, 181–192.
Municipal Census, 2015. Available at http://edmonton.ca/city_government/municipal-
census.aspx [accessed 15 June 2015].
Myrick, R.H., Sakiyama, S.K., Angle, R.P., Sandhu, H.S., 1994. Seasonal mixing heights and
inversions at Edmonton, Alberta. Atmos. Environ. 28, 723–729.
National Post, 2015. Edmonton's Air Quality Is often Worse than Toronto's, which Has
Five Times More People. National Post, Toronto, ON. http://news.nationalpost.com/
news/canada/edmontons-air-quality-is-often-worse-than-torontos-which-has-five-
times-more-people.
Peters, A., Hampel, R., Breitner, S., Geruschkat, U., Kraus, U., Zareba, W., et al., 2015. Elevat-
ed particle number concentrations induce immediate changes in heart rate variabili-
ty: a panel study in individuals with impaired glucose. Particle Fibre Toxicol. 12 (7),
1–11.
Philip, S., Martin, R.V., Pierce, J.R., Jimenez, J.L., Zhang, Q., Canagaratna, M.R., et al., 2014.
Spatially and seasonally resolved estimate of the ratio of organic mass to organic car-
bon. Atmos. Environ. 87, 34–40.
Pope III, C.A., Ezzati, M., Dockery, D.W., 2014. Fine particulate air pollution and life expec-
tancies in the United States: the role of influential observations. J. Air Waste Manage.
Assoc. 63, 129–132.
Rao, S.T., Sistla, G., Henry, R., 1992. Statistical analysis of trends in urban ozone air quality.
J. Air Waste Manage. Assoc. 42, 104–1211.
Ruoho-Airola, T., Anttila, P., Salmi, T., 2004. Airborne sulfur and nitrogen in Finland —
trends and exposure in relation to air transport section. J. Environ. Monit. 6, 1–11.
Salmi, T., Maatta, A., Anttila, P., Ruoho-Airola, T., Amnell, T., 2002. Detecting trends of an-
nual values of atmospheric pollutants by the Mann–Kendall test and Sen's slope
estimates—the Excel template application MAKESENS. Publications on Air Quality
No. 31, Finnish Meteorological Institute, Helsinki, Finland.
Saylor, R., Myles, L., Sibble, D., Caldwell, J., Xing, J., 2015. Recent trends in gas-phase am-
monia and PM2.5 ammonium in the Southeast United States. J. Air Waste Manage.
Assoc. 65, 347–357.
Schaap, M., Muller, K., ten Brink, H.M., 2002. Constructing the European aerosol nitrate
concentration field from air quality analysed data. Atmos. Environ. 36, 1323–1335.
Schulz, J., Kindzierski, W.B., 2001. Characterization and source apportionment of regional
background inhalable particulate matter in Alberta. SFM Network Project: Analysis of
Particulate Matter Loading in Community Ambient Air. Department of Civil and Envi-
ronmental Engineering, University of Alberta, Edmonton, AB (26 pp.).
Seinfeld, J.H., Pandis, S.N., 1998. Atmospheric Chemistry and Physics: From Air Pollution
to Climate Change. John Wiley and Sons, New York, NY.
Sen, P.K., 1968. On a class of aligned rank order tests in two-way layouts. Ann. Math. Stat.
39, 1115–1124.
Simon, H., Bhave, P.V., Swall, J.L., Frank, N.H., Malm, W.C., 2011. Determining the spatial
and seasonal variability in OM/OC ratios across the US using multiple regression.
Atmos. Chem. Phys. 11, 2933–2949.
Sippula, O., Hokkinen, J., Puustinen, H., Yli-Pirila, P., Jokiniemi, J., 2009. Particle emissions
from small wood-fired district heating units. Energy Fuel 23, 2974–2982.
Sternbeck, J., Sjödin, A., Andréasson, K., 2002. Metal emissions from road traffic and the
influence of resuspension results from two tunnel studies. Atmos. Environ. 36 (30),
4735–4744.
Tuncel, S.G., Olmez, I., Parrington, J.R., Gordon, G.E., Stevens, R.K., 1985. Composition of
fine particle regional sulfate component in Shenandoah Valley. Environ. Sci. Technol.
19, 529–537.
Turpin, B.J., Lim, H.-J., 2001. Species contributions to PM2.5 mass concentrations: revisiting
common assumptions for estimating organic mass. Aerosol Sci. Technol. 35, 602–610.
U.S. Environmental Protection Agency (U.S. EPA), 2009. Integrated Science Assessment
for Particulate Matter (Final Report). U.S. Environmental Protection Agency,
Washington, D.C.
U.S. Environmental Protection Agency (U.S. EPA), 2013. ProUCL version 5.0 Software. EPA/
600/R-07/041. Office of Research and Development. Washington D.C.
United States Environmental Protection Agency (U.S. EPA), 2015. Air Quality System
(AQS) data. http://aqsdr1.epa.gov/aqsweb/aqstmp/airdata/download_files.html
[accessed 20 January 2016].
Weatherhead, E.C., Reinsel, G.C., Tiao, G.C., Meng, X.L., Choi, D., Cheang, W.K., et al., 1998.
Factors affecting the detection of trends: statistical considerations and applications to
environmental data. J. Geophys. Res. 103 (D14), 17149–17162.
Weichenthal, S., Villeneuve, P.J., Burnett, R.T., van Donkelaar, A., Martin, R.V., Jones, R.R., et al.,
2014. Long-term exposure to fine particulate matter: association with nonaccidental
and cardiovascular mortality in the Agricultural Health Study Cohort. Environ. Health
Perspect. 112, 609–615.
Wellinger, M., Wochele, J., Biollaz, S.M.A., Ludwig, C., 2012. Online elemental analysis of
process gases with ICP-OES: a case study. Waste Manag. 32, 1843–1852.
Xing, L., Fu, T.-M., Cao, J.J., Lee, S.C., Wang, G.H., Ho, K.F., et al., 2013. Seasonal and spatial
variability of the OM/OC mass ratios and high regional correlation between oxalic
acid and zinc in Chinese urban organic aerosols. Atmos. Chem. Phys. 13, 4307–4318.
Zanobetti, A., Dominici, F., Wang, Y., Schwartz, J.D., 2014. A national case-crossover anal-
ysis of the short-time effect of PM2.5 on hospitalizations and mortality in subjects
with diabetes and neurological disorders. Environ. Health 13, 38. http://dx.doi.org/
10.1186/1476-069X-13-38.