CHARACTERIZATION OF AIR QUALITY AND FINE PARTICULATE MATTER SOURCES IN

THE TOWN OF HINTON, ALBERTA-2017

With concern in recent years about dust issues and fine particulate matter (PM2.5) levels approaching a new Canadian Ambient Air Quality Standards
(CAAQS), an investigation of air quality characteristics and potential sources influencing PM2.5 concentrations was undertaken in the community of
Hinton, Alberta. The study was conducted for the period November 2013 to February 2016 using hourly concentrations of criteria air pollutants.
Comparatively higher concentrations of PM 2.5 were observed in summer (mean: 12.5 mg/m3, median: 8.0 mg/m3) than in winter (mean: 7.5 mg/m3,
median: 6.0 mg/m3).
The 3-year averages of annual average PM2.5 concentrations (8.1e8.9 mg/m3) were below the 2015 annual CAAQS value of 10 mg/m3. Exceedances of
a 1 h Alberta Ambient Air Quality objective (58 times > 80 mg/m3) and a 24 h CAAQS (16 times > 28 mg/m3) were observed at Hinton for the study
period and occurred during summer months primarily due to occurrence of forest fire episodes. A multivariate model positive matrix factorization (PMF)
revealed five sources. Background dust and secondary aerosol was identified as the largest source contributing 68% to PM2.5 mass. Other sources included
traffic (13.4%), an O3-rich source (12.7%), industry (3.1%) and a mixed source (3.1%). These findings offer preliminary information about contributions
of different sources to PM2.5 at Hinton; and this information can support policy makers in developing appropriate management initiatives for reducing
dust and secondary particulate matter pollution.
2017 Turkish National Committee for Air Pollution Research and Control.

1. Introduction
Ambient monitoring of criteria air pollutants e.g., fine particulate matter (PM2.5), ground-level ozone (O3), nitrogen dioxide
(NO2), sulfur dioxide (SO2) is of interest due to their potential link to adverse human health outcomes from exposure
In addition, high concentrations of these pollutants are reported to contribute to acid deposition, photochemical smog and reduced
atmospheric visibility. New air quality standards for PM2.5 e i.e., Canadian Ambient Air Quality Standards (CAAQS) e were
developed in Canada in 2012 through a collaborative process with federal, provincial and territorial governments and other
stakeholders (CCME, 2012). The new CAAQS for PM2.5 for the year 2015 (annual: 10 mg/m3, 24 h: 28 mg/m3) and 2020 (annual:
8.8 mg/ m3, 24 h: 27 mg/m3) replace the former 24 h Canada-Wide Standard (CWS) of 30 mg/m3 established in 2000 (CCME,
2000). The Valley-area of Hinton (population 9,882, area 33.52 km2) (Statistics Canada, 2016) is a small community in the Upper
Athabasca air zone in west-central Alberta (Fig. S1a). The Government of Alberta (Alberta Environment and Parks, 2015) recently
evaluated PM2.5 and O3 in Alberta air zones for the time period of 2011e2013, including the Upper Athabasca airshed where Hinton
is situated. It reported that an air quality monitoring station at Hinton had annual and 24 h PM2.5 metric values of 8.1 mg/m3 and 19
mg/m3. had established four-color coded air quality management thresholds for 2015 PM2.5 and O3 in Alberta (Table S1), and it
assigned the Hinton station to an orange management level for PM2.5 e indicating that PM2.5 concentrations were approaching the
new CAAQS and proactive planning and/or action may need to be considered to prevent exceedances. Historical reports had also
existed about air quality concerns related to nuisance and other complaints from dust, odor and wood burning smoke in Hinton
It was therefore of interest to undertake a study to examine PM2.5 levels and to identify different emission sources that affect PM2.5
levels in Hinton.
Under the National Air Pollution Surveillance (NAPS) monitoring program, Environment Canada only performs PM2.5 speciation
measurements in selected major urban centers of Canada (e.g., Vancouver, Edmonton, Toronto, Montreal). Local airshed monitoring
organizations licensed by the environment department in Alberta (Alberta Environment and Parks) operate air quality monitoring
networks and monitoring stations throughout Alberta (in cities, small towns and rural areas) and measure real-time concentrations of
gaseous pollutants and PM2.5. Receptor models like principal component analysis (PCA) and positive matrix factorization (PMF)
have been widely used to identify and apportion emission sources using PM2.5 chemical speciation data. However, receptor
modeling using PM2.5 speciation data is often timeconsuming and cost-intensive. It is not surprising that many of the monitoring
organizations do not have capabilities and financial resources to routinely monitor for chemical species in particulate matter. This is
the case for the organization responsible for air quality monitoring in Hinton.
Given the fact that chemical composition of particulate matter data is not available for most air quality monitoring stations
worldwide, numerous studies have been carried out to investigate sources and origins of ambient fine particulate matter,
nanoparticles, particle size distribution using real-time concentrations of gaseous pollutants data and meteorological parameters
PCA has been used to identify and characterize emission sources of PM2.5, particle size distribution and atmospheric mercury
using gaseous pollutants data. However, compared to PCA, PMF provides a better fit to ambient data, offers nonnegative source
factors, error estimates and better data treatment including handling of missing values and values below the detection limit (Paatero
and Tapper, 1994). To the best of our knowledge, numerous source apportionment studies have been done to date by applying the
PMF model to determine emission sources using realtime gaseous pollutant data
The approach of using real-time gaseous pollutant data in receptor models may provide a small number of source factors and may
not be able to identify some specific sources (e.g., road dust, secondary organic aerosol, biogenic) and distinguish different industry-
related sources (e.g., metallurgy, refinery, cement kiln) that may be important sources in air sheds. However, in the absence of
chemical speciation information it can offer useful preliminary results highlighting potential major emission sources types that can
affect air quality at a receptor location in urban or rural areas. In a recent study, we applied the PMF model followed by multiple
linear regression (MLR) to investigate PM2.5 sources in the City of Red Deer, Alberta using gaseous pollutant data. In this present
study, we characterized air quality in Hinton and investigated sources of PM2.5 using PMF and available real-time continuous air
monitoring data from November 2013 to February 2016 with an objective to identify local and potential long-range sources of
PM2.5.
2. Methodology
2.1. Ambient measurements of air pollutants
Alberta is oil and natural gas-rich and the fourth largest province located in western Canada. It has well-established conventional oil
and gas extraction, refining and upgrading activities (Supplemental Information-SI, Fig. S1b) in addition to unconventional oil sands
development in the northeast area of the province. Hinton is located 81 km northeast of Jasper National Park in the Rocky
Mountains and about 284 km west of Alberta's capital city, Edmonton. Hinton is also located within the edge of established oil and
gas extraction activities (Fig. S1b) and within the Alberta Energy Regulator Drayton Valley Field Centre (Fig. S1c). During 2014,
more than 5 millionm3 of oil and 3.5(103) millionm3 of solution gas was produced in the Drayton Valley Field Centre region where
Hinton is located (AER, 2016). As part of Environment Canada's NAPS initiatives, the West Central Airshed Society (WCAS,
http://www.wcas.ca) has been responsible for regional air quality monitoring and providing results to Alberta Environment and
Parks in west central Alberta. The WCAS airshed zone is approximately 46,000 km2 and spans from the western boundary of
Edmonton city limits to the British Columbia border (Fig. 1a). The topographical map of the town of
Hinton and immediate surrounding area is shown in Fig. S2. Hinton
has a humid, subarctic continental climate with wide fluctuations
in temperatures e.g., long cold winters and short, cool summers.
Seasonal average temperatures typically vary from _8.9 _C in
January to 15 _C in July and the average annual precipitation is in
terms of rain 45 cm and snow 168 cm (www.hinton.ca). Early
morning ground-based temperature inversions are common
throughout the whole year in Alberta with deeper and stronger
inversions observed during winter months (Myrick et al., 1994),
thus limiting the dispersion of air pollutants and potentially
increasing pollutants levels in winter.
The study was performed using historical air quality data
collected from Hinton air monitoring station (AMS) (53.4273011_ N,
117.544067_ W) for the period of January 2011eFebruary 2016
(Fig. 1b). The station is located to the northeast of Hinton near busy
roadways (within 100 m East River Road and to the northwest
Willow Creek Road). These roadways are influenced by transportation
of heavy haul trucks. There is a parking lot and a gas
station (Fas Gas) located within 30 m south of the monitoring
station, where heavy trucks park and idle in the morning hours
(Greg Swain, personal communication). A small asphalt paving
plant (Hinton Batch Plant-Border Paving Ltd.) is located within
0.5 km northwest of Hinton AMS and an adjacent parking lot exists
that is mostly used for haul trucks. Highway 16 (Yellowhead
Highway) is a major east-west transportation route from Manitoba
to British Columbia and is located within 2 kmfrom the monitoring
station. National Pollutant Release Inventory (NPRI) (Environment
Canada, 2016) reported annual releases of sulfur dioxide (SO2),
oxide of nitrogen (NOX), and PM2.5 to the air from major industrial
facilities within 50 km surrounding the Hinton AMS is shown in
Table S2. Hourly concentration data for criteria air pollutants i.e.,
nitric oxide (NO), nitrogen dioxide (NO2), NOX, SO2, ozone (O3),
total reduced sulfur (TRS), and meteorological parameters were
accessed via the Alberta Environment and Parks airdata warehouse
(AEP, 2016). Hourly concentrations of TRS and PM2.5 had been
measured at Hinton AMS since 2004 and February 2010, respectively;
while measurements of other pollutants (NO, NO2, NOX, SO2,
O3) started from November 2013. Measurement instruments used
for real-time air quality monitoring at Hinton AMS are shown in
Table S3. Based on data availability of all criteria pollutants, the
study period selected was from November 2013 to February 2016.
Seasonal wind roses at Hinton AMS were generated for the selected
study period (Fig. S3). Prevailing wind directions during winter and
fall were strongly aligned with the valley from the south-southwest
blowing 71% and 78% of the time, respectively with minor contribution
from the northeast representing 18% of the time during fall.

Fig. 1. Location of Hinton (town) and Steeper (background) in Alberta (Map data 2017 Google), sky blue boundary represents West Central Airshed Society
airshed zone and
triangles represent air monitoring stations (a) and location of Hinton Air Monitoring Station (AMS) (yellow stick pin) and industries in and surrounding the
monitoring station that
report to NPRI during 2015 (Image Landsat 2016 Google) (b).

In spring and summer months south-southwest winds were also

predominant blowing ~50% of the time and a minor contribution
was observed from the northwest direction representing ~20% of
the time. To represent background concentrations of PM2.5 in
Alberta air, hourly PM2.5 concentrations were also accessed via AEP
(2016) from one NAPS background monitoring station Steeper
within 60 km southeast of Hinton (Fig. 1a).
2.2. Data analysis and source apportionment modeling
Monthly average concentrations of criteria air pollutants,
diurnal hourly profiles, and exceedances of CWS, CAAQS and
Alberta Ambient Air Quality Objectives (AAAQOs) were investigated.
To characterize local emission sources, a source identification
tool conditional bivariate probability functions (CBPF)
developed by Uria-telelaetxe and Carslaw (2014) were calculated

for major air pollutants in order to provide information on the

nature of emission sources and potentially identify hidden contributions
from different source types through their wind speed
dependence. The CBPF includes conditional probability function
(CPF) with wind speed as a third variable and it allocates the
observed pollutant concentration to a cell defined by ranges of
wind direction and wind speed and can be defined as:
where mDq;Du is the number of samples in the wind sector Dq with
wind speed interval Du having concentration C greater than a
threshold value x, and nDq;Du is the total number of samples in that
wind direction-speed interval. The threshold criterion was set at
the highest 25% of the concentrations to define the directionality of
local sources.
The United States Environmental Protection Agency's (USEPA's)
multivariate receptor model positive matrix factorization (EPA
PMF5.0) (USEPA , 2014) was used to determine possible emission
sources of measured PM2.5 concentrations in the town of Hinton
using 24 h concentrations of gaseous pollutants. The model
description and data treatment procedures are described in the SI.
Available data for six pollutants (NO, NO2, NOX, SO2, O3, TRS) from
November 2013 to February 2016, including 851 daily (24 h) concentrations,
was used for PMF analysis. In this study, PM2.5 was
used as an independent input variable in the PMF model to directly
obtain the source contributions to PM2.5 instead of using post-hoc
regression analysis. This approach, including PM2.5 in the PMF
model, has been implemented in several studies with PM2.5
speciation only data (e.g., Kim et al., 2003; Mooibroek et al., 2011;
Gibson et al., 2013; Bari and Kindzierski, 2016; Landis et al.,
2017), and gaseous pollutants, mercury and particle number concentration
data (e.g., Ogulei et al., 2007; Kasumba et al., 2009;
Cheng et al., 2009; Sun et al., 2014; Sowlat et al., 2016). In this
study, PM2.5 concentrations measured at a background monitoring
site Steeper were also evaluated (Fig. S4) and included as an
additional input variable e PM2.5 (background) e in the PMF model
to identify the influence of background contributions. Eight 24 h
PM2.5 values measured during winter at Steeper influenced by
biomass burning (e.g., residential heating) and were excluded from
the dataset. The estimated uncertainty of PM2.5 background mass
was set at four times their values to down weight their influence in
the model solution.
To better understand uncertainty of the PMF solution, including
effects of random errors and rotational ambiguity in the data, three
error estimation methods i.e., bootstrapping (BS), displacement
(DISP) and bootstrapping with displacement (BS-DISP) analysis
were applied. Details of these methods are described elsewhere
(USEPA, 2014; Brown et al., 2015). Different approaches were then
used to verify source assignments from the PMF analysis. This
involved calculating Pearson correlation coefficients between PMFderived
factor contribution concentrations and criteria air pollutant
concentrations and meteorological data in order to investigate their
relationship with identified sources and to assist in interpretation
of sources. In addition, to better determine and characterize potential
local emission sources, CBPF plots were generated using
PMF-derived factor contributions along with wind direction and
wind speed values measured at Hinton AMS. In addition, to identify
the influence of potential long-range sources, backward trajectory
analysis was conducted using the National Oceanic and Atmospheric
Administration (NOAA) Hybrid Single Particle Lagrangian
Integrated Trajectory (HYSPLIT) model (Draxler and Rolph, 2003)
with results plotted using the concentration-weighted trajectory
(CWT) method (Seibert et al., 1994; Hsu et al., 2003). Further details
on trajectory analysis are provided in the SI.

3. Results and discussion

3.1. Ambient levels of criteria air pollutants
Descriptive statistics of hourly concentrations of criteria air pollutants at Hinton for different seasons over the study period
November 2013eFebruary 2016 are shown in Table 1 and monthly
profiles of hourly concentrations are presented in Fig. S5. The
overall mean and median concentrations of PM2.5 were higher in
summer (mean: 12.5 mg/m3, median: 8.0 mg/m3) compared to other
seasons (e.g., winter mean: 7.5 mg/m3, median: 6.0 mg/m3). Elevated
hourly 98th percentile concentrations of PM2.5 were observed between
the months of April and October ranging from 30 mg/m3
(April) to 73 mg/m3 (June). The observed higher PM2.5 concentrations
during spring and summer months are likely associated with
spring dust episodes and long-range smoke transport from forest
fire events. Several high-PM2.5 event days occurred in summer and
spring months when maximum 24 h concentrations exceeded the
CWS of 30 mg/m3 (e.g., July 16, 2014: 69 mg/m3;May 24, 2015: 69 mg/
m3), suggesting an influence of forest fire smoke from Alberta and
nearby provinces such as British Columbia, and Saskatchewan as
well as from the western United States (Fig. S6).
Median concentrations of NO, NO2 and NOX were higher in
winter and fall compared to spring and summer months. NO2
higher 98th percentile concentrations were observed from October
to March (29e34 ppb) compared to summer months (11e15 ppb)
(Fig. S5). This seasonal effect is likely due to a combination of the
role of reactive nitrogen gases (e.g., NO, NO2) in photochemical O3
production and subsequent oxidation reactions during summer,
which removes these gases from the atmosphere. In addition,
temperature prolongs the residence time of these reactive gases
during colder months compared to summer. For O3, 98th percentile
concentrations were highest between the months of March
through June, with median peaks recorded in April (38 ppb)/May
(36 ppb) (Fig. S5). The observed spring high pattern is possibly due
to inputs of O3 originating from the tropospheric reservoir and
brought down to the surface from atmospheric boundary layer

mixing (Singh et al., 1978; Taylor and Hanson, 1992; Lovett, 1994;
Aneja et al., 2000; Steinbacher et al., 2004). No seasonal pattern
was observed for SO2, while TRS showed elevated levels with 98th
percentile concentrations ranging from 13 ppb to 17 ppb (July to
October).
Average diurnal (hourly) concentrations of criteria air pollutants
during winter and summer were evaluated at Hinton AMS for
November 2013eFebruary 2016 (Fig. S7). The CBPF plots (Fig. 2) of
criteria air pollutants show the wind direction locations and wind
speed dependence to identify the probable influence of emission
sources located within 1 h travel time (approximately 15e30 km
away depending on wind speed) from the Hinton monitoring station.
During winter, NOX concentration maxima were observed
between 8:00 to 10:00 (local time) and 19:00 to 21:00. Higher NOX
levels in morning hours and increasing levels in evening hours are
typical patterns due to daily traffic intensity, as the monitoring
station is close to a nearby (within 100 m) busy road East River
Road influenced by the frequent transportation of haul trucks and
also relatively close to highway 16-Yellowhead Highway (within
2 km). During summer, NOX concentration maxima occurred between
7:00 to 9:00 and concentration minima occurred in late afternoon
between 18:00 to 19:00. The winter CBPF plot of NOX
revealed the northeast and south-southwest as dominant source
directions both under low wind speeds indicating an influence of
traffic-related emissions, and a moderate influence was observed
from southwest direction suggesting local industry emissions such
as Hinton pulp and wood products (Fig. 1b). While the summer
CBPF plot of NOX indicated the southwest direction as predominant
suggesting an influence from local industries. Peak concentrations
of O3 were observed during mid-afternoon due to the expected role
of daily atmospheric boundary layer mixing and minima occurred
during morning hours. The CBPF plot for O3 revealed dominant
northwest and southwest directions during winter and southeast
directions during summer months.
The diurnal hourly average pattern for PM2.5 exhibited maxima
between 7:00 to 9:00 and 20:00 to 23:00 during summer and
between 10:00 to 12:00 during winter, while minima occurred
during afternoon hours (15:00 to 17:00). The winter CBPF plot for
PM2.5 pointed southwest (industry) and northeast (e.g., idling haul
trucks in nearby parking lot) as dominant and moderate directions,
respectively. While the summer CBPF plot for PM2.5 indicated
northeast close to the station, and southeast and southwest as
probable source directions. The diurnal concentration patterns for
SO2 (Fig. S7) showed maxima between 14:00 to 17:00 during
winter and between 11:00 to 14:00 during summer and are likely
due to the influence of daytime heating and atmospheric boundary
layer mixing. SO2 minima were observed in early morning hours
corresponding to the absence of atmospheric boundary layer
mixing. Both winter and summertime CBPF plots clearly indicated
southwest as dominant directions pointing to local industrial
emissions. The diurnal pattern of TRS during summer (Fig. S7)
closely resembled PM2.5 maxima being observed between 10:00 to
12:00. TRS maxima were observed in morning hours of 7:00 to 9:00
during summertime possibly suggesting an influence from vehiclerelated
emissions. TRS is a mixture of several reduced sulfur compounds
including hydrogen sulfide (H2S) and similar diurnal patterns
for H2S have been reported in other studies influenced by
traffic emissions (Kourtidis et al., 2004, 2008). Like SO2, TRS and
H2S are generally emitted from industrial activities and during both
winter and summer the TRS CBPF plots indicated southwest as a
prevailing direction also suggesting a possible influence from local
industry.
Daytime hourly maxima for NO2, PM2.5 and TRS tended to occur
earlier than for O3. Whereas the upper atmosphere is a reservoir for
ozone (Singh et al., 1978; Taylor and Hanson, 1992; Zhang and Rao,
1999; Aneja et al., 2000; Steinbacher et al., 2004), the same is not
true for other air pollutants, i.e., NO2, PM2.5 and TRS. Thus the
diurnal behavior of these air pollutants at ground level as influenced
by atmospheric boundary layer mixing differs from that of
ozone.
Fig. 3a shows time series of 24 h PM2.5 concentrations at Hinton
AMS over the period from January 2011 to February 2016. All of the

Fig. 3. Time series of 24 h PM2.5 concentrations (a) and the 3-year average of the annual 98th percentile of the 24 h PM2.5 concentrations and the annual average
PM2.5 (b) at Hinton
AMS for January 2011eFebruary 2016.

24 h exceedances were recorded during summer months, suggesting

that forest fire events play a major role in the exceedances
of PM2.5 objectives and guidelines in Hinton. Several forest fire
episodes occurred in Alberta and nearby provinces and territories
such as British Columbia (July 11, 2012), Northwest Territories (July
7, 2014), Saskatchewan (May 21, June 30, 2015) and from the
western United States (August 28, 2015) during the summer
months in recent years (CWFIS, 2015). However, the 3-year average
of the annual 98th percentile of the daily 24 h average PM2.5 concentrations
(Fig. 3b) showed that no exceedances of the CWS and
CAAQS occurred for 2011e2016.
No exceedance of the CWS/CAAQS and AAAQO was observed for
NO2, O3, and SO2 (Table S4), while several exceedances of 1 h
guidelines and 24 h objectives were observed for PM2.5. For the
study period November 2013eFebruary 2016, the 1 h AAAQO for
PM2.5 (80 mg/m3) was exceeded 58 times and the 24 h CAAQS of
28 mg/m3 was exceeded 16 times at Hinton AMS (Table S4). As
mentioned previously, the observed 24 h PM2.5 exceedances that
occurred during summer months is due to the occurrence of forest
fire episodes.
3.2. PM2.5 source identification at Hinton
PMF was performed on gaseous pollutants and a 5-factor solutionwas
chosen at Hinton for the study period based on the results
of PMF error estimates, and physical meaningfulness of the factors.
Model input data statistics, performance criteria and error estimates
are shown in the SI (Tables S5eS7). The relationship between
modeled and all measured pollutant concentrations (except PM 2.5
background and TRS) (Table S6) showed good fits with square of the
correlation coefficient (r) ranging from 0.98 to 1.0. Fig. 4 shows
source profiles i.e., chemical composition of emission sources of

Fig. 4. Source profile and daily contributions of four identified PMF-derived sources at Hinton AMS for November 2013eFebruary 2016

Table 2
Average contribution of PMF-derived sources to PM2.5 (mg/m3, %) at Hinton for November 2013eFebruary 2016.

each of the identified source factors from the base and bootstrap
runs along with a time series plot of daily contributions of the
sources. The average contribution of PMF-derived sources to PM2.5
for the study period and seasonal average contributions (estimated
after regressing measured PM2.5 concentrations against PMFderived
factor contributions with r2 0.986, Fig. S8) are shown
in Table 2. The correlation of sources with criteria air pollutants and
meteorological parameters is reported in Table S8. CBPF plots of the
five identified sources are shown in Fig. 5 and winter and summertime
CBPF plots are reported in the SI (Fig. S9). The spatial
distribution ofCWT values for the contribution of two specific longrange
sources, i.e., background dust and secondary aerosol and O3-
rich, are presented in Fig. 6 (other CWT plots are shown in SI
Figs. S10eS13).
3.2.1. Factor 1-background dust and secondary aerosol
Factor 1 was identified as the largest source at Hinton accounting
for almost two-thirds or 67.7% (6.3 mg/m3) of the PM2.5
mass concentration on average. It was characterized by high
abundances of PM2.5 and background PM2.5 explaining 69% and 77%
of the variation, respectively. Some small mass fractions of TRS
(10%) and O3 (5%) were also present in this factor. Background PM2.5
concentrations in Hinton may arise from local origins together with
background sources brought into the area via regional transport
from both natural (e.g., biogenic) and anthropogenic origin (e.g.,
upstream oil and gas development in Alberta). Secondary inorganic
(e.g., sulfate, nitrate) and organic aerosol (SOA) may also contribute
to PM2.5 in Hinton. In Alberta, background regional sulfate is found
in relative 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 industrial boilers for power
generation and other non-specific industrial sources (Schulz and
Kindzierski, 2001). In an earlier investigation of rural particulate
matter in Alberta during 1996e1997, Cheng et al. (2000) reported
that sulfate content in PM2.5 varied from 11% to 40% (mean 22%) at
rural-remote sites i.e., regional background (e.g., Hightower Ridge,
Swan Hills) and rural-influenced sites (e.g., Violet Grove, Tomahawk),
similar to urban Alberta areas (e.g., Edmonton, Calgary) and
those observed at non-urban areas in the northwestern United
States.
In general, secondary aerosol formation is largely enhanced
during summer month photochemical activity (Turpin and
Huntzicker, 1995) and can be linked to various local and longrange
transport of anthropogenic and natural emissions such as
traffic exhaust, fossil fuel combustion, biomass burning and
biogenic emissions (Hoyle et al., 2011; Skyllakou et al., 2014). In this
study, factor 1 was positively correlated with O3, ambient temperature,
and standard deviation of wind direction (a measure of atmospheric
turbulence) and showed a negative association with
relative humidity (Table S8). From the time series plot of daily
contributions of this factor, several peak concentrations (e.g., July
14, 2014; June 7 and August 27, 2015) were clearly visible during
summer months, indicating an influence from wildfires that
occurred in Alberta and other western provinces such as British
Columbia, Saskatchewan and in the Western United States during
the period 2014e2016. Some elevated levels were also observed
during spring (April 2014 and 2015) and summer months (June 23,
2014), possibly suggesting an influence of dust episodes. The CBPF
plot of this factor revealed northeast and east as dominant local
source directions under low wind speeds (Fig. 5). Winter and
summer CBPF plots indicated highest concentrations with winds
from the northeast and from the southwest (only summertime)
(Fig. S9). The nearest roadway (within 100 m e East River Road)
influenced by haul trucks and an adjacent parking lot in the
northeast may be associated with this factor. Due to dry weather
conditions during summer days, dust can be generated from heavy
haul truck and large rig movement from industrial users of nearby
roadways (e.g., Willow Creek Road and East River Road) as well as
from ongoing road development and land development/construction
activities (since 2014). Several high dust episodes were
observed at Hinton station on June 23, 2014 (e.g., hourly maximum
of PM2.5: 341 mg/m3 at 8:00 a.m.) and on June 24, 2017 (e.g., hourly maximum of PM 2.5: 307 mg/m3 at11:00 a.m.) (Fig. S14a).
Hourly
PM2.5 levels were elevated between midnight and noon over these
days. The CBPF plot for factor 1 during the month of June 2014
(Fig. S14b) suggests that these episodes likely originated from some
type of isolated event/activity (e.g., construction) that occurred
relatively close to the Hinton station under low speed (<10 km/h)
during these days.
The time-series plot of this factor (Fig. 4) also indicated some
elevated levels during winter months. It is plausible that biomassburning
appliance (wood stove) use during winter among residences
within and surrounding Hinton (Town of Hinton, 2014)
contributed to this factor. From backward trajectory analysis, the
CWT plot indicated higher potential long-range contributions to
the immediate east of Hinton, and from southern Alberta and
British Columbia (long-range) (Fig. 6). The winter CWT plot
(Fig. S11) indicated large potential source regions immediate
northeast and south of Hinton and British Columbia (long-range)
suggesting residential wood-burning and open wood burning (e.g.,
agricultural slash burning) may have been occurring during winter.
The summer CWT plot (Fig. S11) also indicated an influence of
wildfire smoke coming from central and southern Alberta (long
range). Therefore, factor 1 was assigned to background dust and
secondary aerosol source. Fugitive dust and secondary sulfate
sources were also identified in a recent investigation (Bari and
Kindzierski, 2017b) for the study period 2010e2013 at an oil
sands community Fort McKay in northeastern Alberta and
contributed > 60% to PM2.5.
3.2.2. Factor 2-traffic
Factor 2 was represented by high concentrations of trafficrelated
gaseous pollutants i.e., NO2, and NOX explaining 65% and
45% of the variation, respectively. Some contributions of NO, TRS
and PM2.5 (21%, 20% and 13% of their total mass, respectively) were
also observed. This factor was well correlated with NO, NO2, and
NOX (Pearson correlation coefficient, r 0.74e0.98, p 0.01),
positively associated with TRS and relative humidity and showed
negative correlation with temperature (r _0.64, p 0.01)
(Table S8). This factor showed a clear seasonality with high contributions
in winter (2.0 mg/m3, 27%) and fall (1.3 mg/m3, 15%). The
CBFP plot for this factor (Fig. 5) revealed northeast as dominant
source direction and moderate influences from the south and
southwest directions, suggesting roadway traffic influences. The
winter CBPF plot of NOX showed similar directions of local (likely
traffic) sources (Fig. 2). The CBPF plot of this factor for winter
(Fig. S9) pointed northeast as dominant direction and southwest as
moderate source directions. While the summer CBPF plot of this
factor underlined north/northwest as major source directions and
moderate influences from the southwest. As reported previously,
busy roadways near the Hinton station influenced by haul trucks
are likely associated with this factor. The presence of highway 16
located approximately 2 km to the east-southwest directions may
also contribute to this factor. Therefore, factor 2 was interpreted as
traffic source and the contribution of this factor was 13.4% (1.3 mg/
m3) to measured PM2.5 mass on average (Table 2).
3.2.3. Factor 3-O3-rich
Factor 3 was represented by high abundance of O3 explaining
90% of the variation and had notable associations with TRS (54%)
and NO2 (15%). Some contributions of PM2.5 and its background
levels (11% and 20% of their total mass, respectively) were also
present in this factor. This factor was interpreted as a general O3-
rich source (i.e., correlated with O3 levels) and it contributed 12.7%
(1.2 mg/m3) to the PM2.5 mass concentration on average (Table 2).
This factor showed significant correlation with O3 (r 0.99,
p < 0.01) and SO2 (r 0.17, p < 0.01), and was positively associated
with wind speed and ambient temperature (Table S8). The time
series plot of this factor indicated a strong seasonality (Fig. 4) with
higher contributions in spring months (19%) compared to summer
and fall (10%). As reported previously, the spring high-O3 pattern is
possibly due to an influence of O3 originating from the tropospheric
reservoir and brought down to the surface from atmospheric
boundary layer mixing. Concentrations of O3 can be enhanced
during late winter (with snow cover still on the ground) thru late
spring due to photochemical reactions along with the presence of
its precursors like NOX, volatile organic compounds (VOCs) (Carter
and Seinfeld, 2012), leading to formation of secondary organic
aerosol (SOA). Thus observed late winter thru late-spring O3 concentrations
in Hinton may also, in part, be associated with various
local and long-range transport of anthropogenic precursor
emissions.
The CBPF plot (Fig. 5) indicated probable dominant local source
directions from the northwest under low wind speed (10 km/h) and
southwest under high wind speed (>20 km/h), while a strong influence
was observed from the east direction. The winter CBPF plot
(Fig. S9) revealed south/southwest as dominant source direction
under high wind speed (>15 km/h) possibly indicating an influence
from local industry, while the summer CBPF showed north and
north/northwest directions under low wind speeds suggesting an influence from near-field sources. The backward trajectory analysis
(CWT plot) indicated higher potential long-range contributions
from the surrounding of Hinton, southwest and southeast Alberta
and British Columbia (Fig. 6 and S12). Industry emissions (upstream
oil and gas in Alberta and mining, pulp mill and wood processing in
British Columbia) are possible long-range sources for this factor.
Using real-time monitoring data a factor classified as O3-rich was
also identified in other studies conducted in Rochester, New York
(Kasumba et al., 2009), Prague, Czech Republic (Thimmaiah et al.,
2009), Kuala Lumpur, Malaysia (Khan et al., 2015) and in Red
Deer, Alberta (Bari and Kindzierski, 2017a).
3.2.4. Factor 4-industry
Factor 4 was identified as a minor source and characterized by
high concentrations of SO2 representing 89% of the explained
variation along with some small contributions of TRS (5%). SO2 is a
known marker for industrial emissions. The time-series plot of
daily source contributions of this factor showed some peak concentrations
during winter and fall (Fig. 4). Contributions of this
factor exhibited significant correlation with SO2 (r 0.99), positive
correlations with O3, TRS, wind speed and temperature, and
negative association with relative humidity (Table S8). For local
(short-range) source influences, the CBPF plot of this factor (Fig. 5)
clearly revealed the southwest direction as a dominant contributor
and moderate influence from the south under wind speed >15 km/
h. The seasonal (winter and summer) CBPF plots also indicated
southwest as a dominant source direction (Fig. S9). The directionality
patterns of this factor were similar to the CBPF plots for SO2
and TRS. Local industries located to the southwest are plausible
sources associated with this factor. The CWT plots of this factor
(Figs. S10 and S13) indicated potential long-range source regions
originating from the north and from British Columbia. Therefore,
factor 4 was represented by industry and it only contributed 3.1%
(0.3 mg/m3) to measured PM2.5 mass concentration on average
(Table 2).
3.2.5. Factor 5-mixed source
Factor 5 was another minor source characterized by high levels
of NO, and NOX explaining 79% and 39% of the variation, respectively
and also showed association with NO2, and TRS (12% and 11%
of the variation, respectively). Daily contributions of this factor
showed significant associations with NO, NO2, and NOX
(r 0.66e0.99), and positive association with TRS (r 0.20)
(Table S8). A clear winter-high trend was observed (Fig. 4). The
CBFP plots (Figs. 5 and S9) showed northeast, northwest and
southwest (during winter) and southeast (during summer) as
dominant directions suggesting both traffic and local industry as
possible sources. Therefore, factor 5 was assigned as a mixed source
and it only contributed 3.1% (0.3 mg/m3) to measured PM2.5 mass
concentration on average.
Role of wildfire smoke e Wildfires are a common occurrence in
the Canadian boreal forest, which makes up almost 50% of the land
area in Alberta (Stocks et al., 2002; Natural Resources Canada,
2017). In this study, PMF re-analysis was also done after
excluding seven high PM2.5 values influenced by wildfire smoke
events during the study period in an attempt to understand the role
of strong wildfire biomass smoke-influenced days (Fig. S15).
Exclusion of these wildfire days did not show any notable change in
contributions of originally derived PMF sources. After exclusion,
background dust and secondary aerosol (70.6%) remained the
dominant contributor to PM2.5, followed by traffic (12.9%), O3-rich
(9.8%), mixed source (3.7%) and industry (3%). As wildfires are
common natural events in Alberta, inclusion of wildfire days in
receptor models provides a more realistic understanding of the
contribution of different natural and other anthropogenic emissions
to PM2.5.
3.3. Origin of PM2.5 pollution days at Hinton
Potential source origins of selected PM2.5 pollution days were
further analyzed during the study period. For this analysis, 24 h
PM2.5 values _ 28 mg/m3 (2015 CAAQS) in the dataset were classified
as high PM2.5 pollution days and values > 19 mg/m3 (orange
threshold value) but < 28 mg/m3 were classified as notable PM2.5
pollution days, where planning and/actions may be needed for
preventing CAAQS exceedances. Daily PM2.5 values > 10 mg/m3
(yellow threshold value but <19 mg/m3) were classified as low
pollution days, where planning and/or actions may be needed for
preventing air quality deterioration. Selected PM2.5 pollution days
that occurred during different seasons in Hinton, the dominant
PMF-derived background dust and secondary aerosol source and
associated potential source regions based on HYSPLIT modeling
observation are shown in Table 3. A detailed list of 24 h PM2.5 values
with different pollution categories is shown in Table S9. HYSPLIT
backward trajectories for 24 h PM2.5 values from Table 3 are shown
in Figs. S16eS21.
From the whole study period 15 days were identified as high
PM2.5 pollution days, where PMF model-derived background dust
and secondary aerosol source was identified as a dominant
contributor. As reported previously, most of all CAAQS exceedances

Table 3
Potential source regions of background dust and secondary aerosol on PM2.5 pollution days in Hinton.

were observed during summer (except February 24, 2016)

(Table S9). This suggests that long-range transport of background
dust and secondary aerosol, including biomass burning is an
important contributor to high PM2.5 pollution days at Hinton. For
example, HYPLIT modeling (Figs. S16 and S17) and NASA's Moderate
Resolution Imaging Spectroradiometer (MODIS) observations
(Fig. S6) clearly show the influence of a biomass-smoke related
source i.e., wildfires during summer months; and potential longrange
source regions were in British Columbia (16-Jul-14, 14-Aug-
14, 7-Jun-15), Saskatchewan (24-May-15), and Western United
States (27-Aug-15).
Dust contributions from local sources during spring, summer
and fall may also be important on specific days for exceedances of
the 2015 CAAQS and orange threshold values. Examples of local
sources include roadways within 100 m influenced by heavy haul
trucks as well as from ongoing road development and construction
activities since 2014 in the vicinity. During winter, PM2.5 contributions
from local residential wood combustion (RWC) appliances
for heating (e.g., wood stoves, fireplaces) may also be important on
specific days.
4. Conclusion
An evaluation of air quality and sources of PM2.5 concentrations
was undertaken in the town of Hinton, Alberta for the period
November 2013 to February 2016. Summer PM2.5 concentrations
were relatively higher compared to other seasons. The 3-year
average of the annual average PM2.5 concentrations (8.1e8.9 mg/
m3) was below the 2015 annual CAAQS value of 10 mg/m3. Observed
1 h and 24 h exceedances of the Canadian air quality objectives and
standards at Hinton occurred during summer months primarily
due to the influence of forest fires. PMF analysis revealed background
dust and secondary aerosol as a dominant contributor to
PM2.5 mass (68%) followed by traffic (13.4%), an O3-rich source
(12.7%); and a minor industry source (3.1%) and minor mixed
source (3.1%). An investigation of the origin of PM2.5 pollution days
revealed dust, local and long-range biomass burning sources as
important contributors to PM2.5 at Hinton.
We acknowledge some limitations in the current study. We
could not separate the contribution of background dust, secondary
aerosol and biomass burning sources to PM2.5 as well as potential
inputs from different long-range industrial emission sources
throughout Alberta, e.g. upstream oil and gas activities, due to lack
of chemical speciation data. PM2.5 chemical speciation monitoring
would be needed at rural-influenced sites and/or small towns like
Hinton to better understand the contribution of these sources to
PM2.5 in rural Alberta. Notwithstanding these limitations, these
findings offer preliminary information about different emission
sources that influence PM2.5 levels at Hinton; and this information
can be used by decision makers for consideration of appropriate
plans and/or actions for fine particulate matter.