Sustainable Cities and Society 32 (2017) 475–485

Contents lists available at ScienceDirect

Sustainable Cities and Society

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

Pedestrian exposure to traffic PM on different types of urban roads: A case T

study of Xi’an, China
⁎
Zhaowen Qiu , Xiaoqin Xu, Jianhua Song, Yaping Luo, Ruini Zhao, Bihai Xiang, Wencai Zhou,
Xiaoxia Li, Yanzhao Hao
School of Automobile, Chang’an University, Chang’an Road, Xi’an 710064 Shaanxi, China

A R T I C L E I N F O A B S T R A C T

Keywords: To investigate the pedestrian exposure to fine particulate on various typical urban roadways, mass concentration
Pedestrian exposure of PM10, PM2.5, PM1 and particle number concentrations in the given size distribution were obtained at four
Typical urban roadways types of urban roads in Xi’an City: urban expressway, arterial road, collector road, and local road. The principal
PM mass and particle number concentrations component analysis (PCA) was used. Measurement and analysis results showed that the mean particle number
Meteorological parameters
concentration for the size range of 0.25–32 μm on the urban expressway was 9%, 29% and 32% higher than
Principal component analysis
those on the arterial road, the collector road and the local road, respectively. However, the mass concentration of
particles, especially for PM2.5 and PM1, varied little. Traffic volumes on the urban expressway, the collector
road and the local road all had important influences on PM particle-number concentration, but at different size
ranges. Additionally, the results discussion indicated that PM10, PM2.5 and PM1 were generated from the same
emission source(i.e. vehicles)on the urban expressway, yet on the other types of roads PM (especially PM10 and
PM1) emissions were generated from different sources (i.e. vehicles, pedestrians, or road dust).

1. Introduction
On urban roadways, the amount of pollution generated by vehicle
traffic varies widely, depending on the day of the week, the time of day,
and the type of roadway. During rush hours, for example, most urban
roadways are congested, often with stop-and-go traffic; hence there is a
high fluctuation in vehicle speeds, and more pollutants are produced
than on other types of roads such as rural highways or speedways.
Moreover, there are various types of urban roadways such as urban
expressways, arterial roads, collector roads, and local roads, and each of
these has its own unique characteristics of traffic pollution, because of
the differences in vehicle operation modes (acceleration versus cruis-
ing, and how they are proportioned) and exhaust emission rates.
However, most of the existing researches related to traffic pollution
have been carried out on a single type of urban roadway, or even at a
fixed monitoring station (Karner, Eisinger, & Niemeier, 2010; Rakowska
et al., 2014; Sabaliauskas, Jeong, Yao, Jun, & Evans, 2013; Schneider,
Teixeira, Oliveira, & Wiegand, 2015), limiting the value of these
studies, in their analysis of pedestrian exposure to traffic pollutants.
Such exposure is often affected by characteristics more complex than
just the ambient conditions at a fixed monitoring station (Adams,
Nieuwenhuijsen, & Colvile, 2001).
Fixed monitoring stations have traditionally been used for the
establishment of air quality guidelines and policies, but these stations
are not designed to represent micro-scale impacts and as a result may
not adequately describe local conditions in close proximity to traffic
(Gulliver & Briggs, 2004; Moore, Figliozzi, & Monsere, 2012). There-
fore, real-time measurements of pedestrian exposure concentrations
near different types of urban roadways can help better characterize
traffic pollution and its impact on human health (Wang & Gao, 2011).
Particulate Matter (PM) is one of the exhaust pollutants emitted by
road vehicles. Traffic-related PM causes people living and working near
roadways to be at increased risk of respiratory problems, birth and
developmental defects, premature mortality, cardiovascular problems,
and cancer (Finn et al., 2010). Traffic-related PM commonly includes
PM10 (coarse particles), PM2.5 (fine particles), and PM1 (very fine
particles) in units of mass concentration. Toxicological studies have
suggested that ultrafine particles (UFPs) may be more harmful to health
than large particles, because small particles have a much larger
adsorbed or condensed surface area than large particles with the same
mass, and can be deposited in alveoli, where they interact with
epithelial cells (Cheng, Chang, & Hsieh, 2011). PM’s ultrafine fraction
is characterized by particle number concentration as opposed to mass
concentration. Hence, it is of significance to investigate the behaviors of
fine-particle concentrations in both number and mass, on different
types of urban roads.

⁎
Corresponding author.
E-mail address: qzw_chd@qq.com (Z. Qiu).

http://dx.doi.org/10.1016/j.scs.2017.04.007
Received 24 January 2017; Received in revised form 17 March 2017; Accepted 13 April 2017
Available online 18 April 2017
2210-6707/ © 2017 Elsevier Ltd. All rights reserved.
Z. Qiu et al.
Although there have been several studies of PM pollution at
roadsides (Bereitschaft, 2015; Boarnet et al., 2011; Cepeda et al.,
2017; Charron & Harrison, 2005; Qiu & Peng, 2015; Wang et al., 2008;
Zhu, Hinds, Kim, & Sioutas, 2002), these have focused on the quantifi-
cation and correlation of the major influencing parameters (wind,
temperature, traffic density, etc.) at single types of roads, and have
lacked a comprehensive comparison of pedestrian PM exposure con-
centrations on different types of urban roads. Additionally, since the
level of PM pollution is affected by the entire urban road network, the
relationships among the different PM concentrations along the links in
these networks need to be identified. Pedestrian exposure concentra-
tions to traffic-related PM are influenced not only by the traffic
emission intensity and the meteorological conditions, but also by travel
routes.
In this study, pedestrian exposure to mass concentration of PM10,
PM2.5, PM1 and particle number concentrations in the 0.25–32 μm
range on the various types of urban roadways—including urban
expressways, arterial roads, collector roads, and local roads—were
investigated in Xi’an City. The objectives of this study were: 1) to reveal
the distribution characteristics of fine particulate matter in mass and
particle number concentrations on the different types of urban roads;
and 2) to identify the effects of traffic, meteorological conditions and
road type on fine particulate matter concentration for the different
types of urban roads.
2. Materials and methods
2.1. Sampling locations
For the background concentration to be similar for the chosen
sampling sites, four sampling sites were chosen, at the four types of
urban roads in the same region, Xiaozhai—the busiest commercial area
Fig. 1. Map showing the sampling locations.
476
Sustainable Cities and Society 32 (2017) 475–485
in the southern part of Xi’an City. An aerial view of the monitoring site
is provided in Fig. 1. Point A is located at South Two-Ring Road, an
urban expressway, and Points B, C and D are located at a typical arterial
road (Chang’an Road), a collector road (Cuihua Road), and a local road
(Yucai Road), respectively. Each sampling point is placed on the
sidewalk, about 1 m away from the edge of the road.
2.2. Instrumentation and schedule
Mass concentrations of PM10, PM2.5, PM1 and particle number
concentrations in different size ranges were measured with two Portal
Aerosol Spectrometers (GRIMM 1.109). The measuring principle of
GRIMM 1.109 is the light scattering of single particles with a
semiconductor laser as the light source. This instrumentation can detect
very small particles, down to 0.25 μm, and can record at 31 size
channels (over a very wide size range, from 0.25 to 32 μm). This way
the particle size distribution can be measured, providing the basis for
the calculation of the dust mass. The time interval analyzed was set as
6 s for each measurement period. Before data collection, both sets of
instruments were run side by side in the laboratory, to ensure that their
measurements were highly correlated (r2 = 0.996). Device intake
points were set at 1.5 m above the ground, and the height of the
instrumentation was selected to simulate the exposure concentration in
the breathing zone of a typical adult pedestrian (Buonanno,
Fuoco, & Stabile, 2011; Moore et al., 2012). Field sampling was
conducted between February 27 and April 5 in 2016. PM concentration
data were measured during the morning (7:00 to 9:00 a.m.) and
evening (5:00 to 7:00) peak hours.
Meteorological parameters (wind speed and direction, relative
humidity, temperature) around urban road areas can have significant
effects on micro measurements of air quality. To control for these
effects, a micro weather station (a Kestrel 4500 Pocket Weather
Z. Qiu et al. Sustainable Cities and Society 32 (2017) 475–485

Table 1 analysis: 1) meteorological characteristics; 2) 5-min averaged traffic

Meteorological and traffic conditions during sampling time. flow rate and composition; 3) minutely averaged PM1, PM2.5, and
PM10 mass concentration and particle number concentrations for seven
Date Average wind Temperature Relative Traffic
speed (°C) humidity volume size intervals. The PM and PN concentrations on the ground and the
(m/s) (%) (vehicle/h) bridge were compared after accounting for the hourly averaged back-
ground concentrations.
Feb 27a.m. 0.4 12 38 2775
The size of particles is an important parameter, in order to
Feb 27p.m. 0.3 20 13 2788
Feb 28a.m. 1.1 13 29 2272 determine the particle sources and behavior in air. Therefore, measured
Feb 28p.m. 1.9 16 24 2733 particle number concentrations were divided into seven size intervals:
Mar 2a.m. 0.4 10 45 3248 PN0.25–0.39, PN0.40–0.59, PN0.60–0.99, PN1.0–1.99, PN2.0–4.99, PN5.0–9.99,and
Mar 2p.m. 0.4 20 22 2719 PN≥10, based on the particle number concentration orders of magni-
Mar 5a.m. 0.6 9 22 3414
tude.
Mar 5p.m. 0.6 17 24 3441
Mar 6a.m. 0.8 10 34 2415 Principal component analysis (PCA) was used in the data analysis
Mar 6p.m. 0.6 17 24 3261 process. PCA is a projection method that transforms an original set of
Mar 7a.m. 0.8 13 39 3182 correlated variables into a new set, with an equal number of indepen-
Mar 7p.m. 0.4 14 34 2736
dent uncorrelated variables or principal components (PCs) that are in
Mar 12a.m. 0.3 8 42 534
Mar 12p.m. 0.3 15 33 935 linear combinations with the original variables. The main results of PCA
Mar 14a.m. 0.6 7 33 2762 are factor loading, which reflects how much a given variable con-
Mar 14p.m. 0.6 16 29 3016 tributes to a particular PC, and how well the PC describes the variation
Mar 19a.m. 0.5 15 55 3742 of that specific variable. The higher the loading of a variable, the more
Mar 27a.m. 0.4 10 43 534
that variable contributes to the variation by the particular principal
Mar 27p.m. 0.7 21 25 836
Mar 28a.m. 0.9 16 40 3044 component. In practice, only loadings with absolute values greater than
Mar 28p.m. 1.1 23 23 2884 0.5 are selected for principal component interpretation (Jolliffe, 1986).
Mar 30p.m. 1.1 24 28 3467 A correlation and linear regression analysis were performed to explore
Apr 4a.m. 0.3 19 61 796
the potential effects of the road type, meteorology, traffic volume, and
Apr 4p.m. 0.6 26 39 735
Apr 5p.m. 1.0 28 36 623
fleet composition on pedestrian exposure to fine particulates. The
statistical software package SPSS 20.0 for Windows was used for the
data analysis.
Tracker) capable of providing measures of the meteorological para-
meters at a 5-s resolution, was placed next to each particulate monitor.
This device consists of a wind-speed sensor with a range of 0.4–40 m/s 3. Results and discussion
and an accuracy of ± 3%, and a wind-direction sensor with an accuracy
of ± 2% degrees for wind speeds of up to 30 m/s. The hourly averaged 3.1. Traffic characteristics of different types of roads
PM2.5 concentrations at the regional air pollution monitoring station
(Xiaozhai Station) were recorded as PM2.5 background concentrations. Air pollution at the roadside strongly depends on the traffic volume
The PM2.5 measuring principle of National monitoring station is also and vehicle fleet composition. The traffic data, which were collected
light scattering theory which is same with the field instrumentation continuously, showed that the variation of traffic volume within
GRIMM 1.109. So, the measurement results are comparable. daytime hours is very small. However, the traffic volumes among the
Because the PM and PN concentrations in the vicinity of urban different types of urban roads showed great differences. The 5-min
roadways depend strongly on the traffic flow, traffic data needs to be averaged traffic statistics on each type urban roadway are shown in
taken into account as well. Hence, during measurements, traffic volume Fig. 2, which shows that vehicle flow varied from 623 to 3190 vehicles/
data, including fleet composition, were collected and aggregated to h on the various urban roads.
five-minute intervals. The specific meteorological data and traffic For the detailed investigation of vehicle fleet composition effects,
conditions during sampling are listed in Table 1. these motor vehicles were further classified into four categories: private
cars, fueled with gasoline; taxis, fueled with CNG (Compressed Natural
Gas); buses, fueled with CNG; and other types of vehicles, fueled with
2.3. Data analysis diesel. The other types of vehicles referred to are vehicles such as road
sprinklers, ambulances, etc. From the observations in this study, the
The following parameters were further processed in the data traffic fleet composition proportions at the various types of urban roads

Fig. 2. Hourly average traffic flow on different types of urban roads.

477
Z. Qiu et al.
Fig. 3. Percentages of the four motor vehicle types, on the different types of urban roads.
are similar. Gasoline-fueled private cars are the most numerous, CNG-
fueled vehicles second, followed by a small number of diesel-fueled
vehicles. The corresponding percentages of the four types of motor
vehicles are shown in Fig. 3.
3.2. Wind speed and direction
Wind speed and direction play important roles in the transportation
of the particles. As is well known, if the wind blows from the roadway
to the measuring points, the observed concentrations of particles will
increase, showing that the diffusion is enhanced. The urban expressway
(South Two-Ring Road) and the local road (Yucai Road) are east-west
direction road, and the samplings were carried out on the southern side
of the road. The arterial road (Chang’an Road) and the collector road
(Cuihua Road) are south-north direction road, and the samplings were
carried out on the eastern side of the arterial road and western side of
the local road, respectively. When combined with wind rose showed in
Fig. 4, it could be found that sampling sites were all basically located at
the downwind. In this way, the roadway effects on particles could be
better indicated.
3.3. Particle number concentrations on different types of roads
3.3.1. Basic statistics
The particle number concentrations were measured using a
GRIMM1.109 model, for particles ranging in size from 0.25 to 32 μm
(for full-size particles within the instrumentation measurement capa-
city), and the average number concentration of the particles for the
other different size ranges within the overall range (0.25–2.5 μm (for
fine particles) and 2.5–10 μm (for coarse particles)) for the four types of
roads are shown in Table 2. The particle number concentrations in the
size range of 0.25–2.5 μm on the urban expressway were 9%, 28% and
478
Sustainable Cities and Society 32 (2017) 475–485
32%, respectively, more than those on the arterial road, the collector
road and the local road. In addition, the number concentrations of
particles with a size range of 2.5–10 μm on the urban expressway were
42%, 75% and 90% more than those on the arterial road, the collector
road and the local road, respectively. It could also be found that the
number concentrations of particles with a size range of 0.25–32 μm on
the urban expressway were found to be 9%, 29% and 32% higher than
those on the arterial road, the collector road and the local road,
respectively, similar to the particles in the size range of 0.25–2.5 μm.
The differences among the particle number concentrations on the
different types of roads may be due to the differing geometries of the
roads, the road facility configurations, and/or the vehicle operational
environment, including the number of lanes, traffic flow and vehicle
speeds (Wang et al., 2008).
Although the concentration level is different on four types of roads,
it was found that the distribution of particles with different size range
was essentially the same. For further study, Fig. 5(a) and (b) shows the
percentages of the particle number concentrations for each particle size.
Particles with a size range of 0.25–0.59 μm accounted for 98% of the
total number concentration of particles, but particles which size larger
than 0.6 μm contributed only 2% to the total, indicating that the
smaller the particle size, the greater its contribution to the environ-
ment. In other words, particulate pollution beside urban roads consists
mainly of fine or ultrafine particles, which do the greatest harm to
human health (Wang & Gao, 2011).
3.3.2. PCA of temperature, relative humidity and wind speed
In this study, PCA was applied, using factor extraction with an
eigenvalue larger than 1 to explore the interactions among meteorolo-
gical parameters (temperature, relative humidity and wind speed) and
traffic, based on data obtained by measurements at four different types
of roads. We selected the data measured in relatively lower background
Z. Qiu et al. Sustainable Cities and Society 32 (2017) 475–485

Fig. 4. Wind rose diagrams on studied roads.

Table 2 high loading of particles in the size range of that larger than 5 μm,
Average particle number concentrations for the sampling sites (particle/cm3). indicating that the behaviors of particles in these size ranges are similar
to each other. The PC3 factor was characterized by significant loading
Range of particle size 0.25–2.5 μm 2.5–10 μm 0.25–32 μm
Urban expressway 725,306 2586 727,907
only on the meteorological variables of temperature and relative
Arterial road 657,636 1487 659,135 humidity, and these showed a negative correlation. This result can be
Collector road 519,072 640 519,720 interpreted to imply that an increase in temperature probably makes
Local road 492,947 259 493,209 the surrounding moisture evaporate and then results in a decrease in
relative humidity. On the arterial road, the PC1 factors exhibited high
loading of particles in the size range of 0.6–10 μm and a negative
concentrations. Meanwhile, much higher concentrations appeared
loading on wind speed, which may indicate that particles in this size
sometimes resulting from the garbage truck and pedestrian passing
range are negatively correlated to wind speed. It can be inferred that
by, so the outliers were eliminated to insure the accuracy of the results.
high-speed wind dilutes vehicle emissions into the surrounding air. The
Then the number concentrations are normalized by using the Z-score
PC2 factor had a high loading on particles smaller than 0.6 μm and a
standardized methods.
negative loading on relative humidity, suggesting that relative humidity
We first analyzed the relationship between meteorological para-
has some influence on aerosol dynamics (Nikolova et al., 2011),
meters and the particle number concentration. As shown in Table 3, on
especially for the particles size range of 0.25–0.59 μm. Similar analysis
the urban expressway, the PC1 factor was dominated by high loading of
results were found for the collector road and the local road.
particles in the size ranges of 0.25–0.39 μm and 0.4–0.59 μm; the PC2
Therefore, the effects of meteorological parameters on the particle
factor showed a high loading of particles in the size ranges of
number concentration are different for the four types of roads. On the
0.6–0.99 μm, 1–1.99 μm, and 2–4.99 μm; and the PC4 factor had a
urban expressway, the meteorological variables had no significant

479
Z. Qiu et al. Sustainable Cities and Society 32 (2017) 475–485

Fig. 5. Percentages of particle number concentrations for each particle size (on March 14th, at urban expressway).

relationship to the particle number concentrations. On the arterial road,
the wind speed had a negative correlation with the concentration of
particles in the size range of 0.6–9.99 μm, and the relative humidity had
a negative correlation with the concentration of particles in the size
range of 0.25–0.59 μm. For the collector road, relative humidity had the
same influence on particle number concentration as it did for the
arterial road. It also can be seen that on the local road, relative
humidity was negatively related with the concentration of particles in
the size range of 2–4.99 μm, and wind speed had a negative correlation
with the concentration of particles in the size range of 1–1.99 μm.
Overall, it was found that wind speed has a large negative influence
on the particle number concentration on arterial road and local road,
especially for the particles in the size range of 0.6–9.99 μm, relative
humidity has negative influence on particles size range from
0.25–0.59 μm on the three types of roads except expressway. In other
words, meteorological parameters have no obvious impacts on particle
number concentration on the urban expressway, it could be explained
by that the road being much wider, and there are more than the many
branches of roads on both sides of expressway results in complex
atmospheric variations.
In addition, an analysis of correlation coefficients between particle
number concentration, temperature, relative humidity, and wind speed
was performed, to verify the PCA results above. It was found that the
particle number concentrations of particles in the size ranges of
0.25–0.39 μm and 0.4–0.59 μm for these other road types have high
coefficients of correlations with each other, corresponding to the high
loading of PC1 in the same ranges for the urban expressway. The same
result is applicable to particle sizes in the ranges of 0.6–0.99 μm,
1–1.99 μm and 2–4.99 μm. Furthermore, wind speed had a negative
correlation coefficient with particles in the size range of 0.6–9.99 μm,
which is in accordance with the results of PC1 for the arterial road.
Finally, relative humidity was negatively correlated with temperature,
also consistent with the PCA results of PC3 for the urban expressway.
3.3.3. PCA of traffic volume
As we noted in the Methods section, every five-minute traffic flow

Table 3
Factor loading of PCA with temperature, relative humidity and wind speed.

Urban expressway Arterial road

PC1 PC2 PC3 PC4 PC1 PC2

PN0.25–0.39 0.902 0.105 0.002 0.016 0.038 0.942

PN0.4–0.59 0.879 0.274 0.125 -0.016 0.144 0.926
PN0.6–0.99 0.422 0.763 0.074 0.043 0.737 0.377
PN1–1.99 0.115 0.814 0.040 0.035 0.758 0.050
PN2–4.99 0.018 0.876 -0.048 0.077 0.661 0.023
PN5–9.99 -0.040 0.177 -0.059 0.781 0.714 -0.094
PN≥10 0.035 -0.053 0.015 0.841 0.210 0.160
Wind speed -0.297 -0.029 0.734 -0.042 -0.606 0.284
Temperature -0.231 -0.041 -0.891 -0.025 -0.658 -0.154
relative humidity 0.499 0.054 0.694 -0.063 -0.077 -0.873
Eigenvalue 3.116 1.882 1.320 1.169 3.201 2.528
Proportion 21.710 21.388 18.438 13.338 29.371 27.921
Cumulative 21.710 43.098 61.536 74.874 29.371 57.292

Collector road Local road

PC1 PC2 PC3 PC1 PC2 PC3 PC4

PN0.25–0.39 0.885 0.307 0.086 0.946 0.014 0.022 -0.003

PN0.4–0.59 0.743 0.507 0.105 0.954 -0.059 -0.002 0.011
PN0.6–0.99 0.261 0.854 0.133 0.816 0.121 -0.013 -0.049
PN1–1.99 -0.043 0.839 0.152 0.003 0.024 -0.325 -0.511
PN2–4.99 -0.080 0.842 0.024 0.058 0.608 -0.136 -0.013
PN5–9.99 -0.127 0.362 0.652 -0.077 0.078 0.747 -0.054
PN≥10 0.240 0.024 0.811 0.074 -0.021 0.676 0.029
Wind speed 0.374 0.013 -0.457 -0.034 -0.029 0.181 0.861
Temperature 0.894 -0.116 -0.026 0.006 0.780 0.094 -0.280
relative humidity -0.903 0.229 0.161 -0.003 -0.780 0.181 0.192
Eigenvalue 3.506 2.700 1.105 2.499 1.697 1.155 1.087
Proportion 32.405 26.916 13.786 24.866 16.128 12.138 11.241
Cumulative 32.405 59.321 73.108 24.866 40.994 53.133 64.374

480
Z. Qiu et al. Sustainable Cities and Society 32 (2017) 475–485

Table 4
Factor loading of PCA with traffic volume.

Urban expressway Arterial road

PC1 PC2 PC3 PC1 PC2

PN0.25–0.39 0.102 -0.444 0.770 0.205 0.911

PN0.40–0.59 0.832 -0.108 0.025 0.299 0.889
PN0.60–0.99 0.904 0.207 0.052 0.813 0.479
PN1.00–1.99 0.601 0.210 0.516 0.930 0.147
PN2.0–4.99 0.532 0.169 0.546 0.925 -0.089
PN5.0–9.99 0.372 0.798 0.103 0.962 -0.014
PN≥10.00 -0.070 0.684 0.489 0.550 0.247
Traffic volume 0.756 0.027 -0.006 0.322 -0.465
Eigenvalue 2.308 1.773 1.639 3.843 2.157
Proportion 28.844 22.158 20.486 48.041 26.967
Cumulative 28.844 51.002 71.489 48.041 75.008

Collector road Local road

PC1 PC2 PC3 PC1 PC2 PC3 PC4

PN0.25–0.39 0.003 0.977 -0.024 0.940 0.137 0.189 -0.055

PN0.40–0.59 0333 0.927 0.055 0.942 0.041 0.048 -0.154
PN0.60–0.99 0.848 0.310 0.204 0.528 0.724 -0.200 0.007
PN1.00–1.99 0.930 0.064 0.065 -0.076 0.597 0.327 0.551
PN2.0–4.99 0.808 0.323 0.315 0.030 0.859 0.114 0.177
PN5.0–9.99 0.582 0.144 0.682 0.050 0.126 0.735 -0.103
PN≥10.00 -0.035 -0.046 0.878 0.123 -0.054 0.803 0.269
Traffic volume 0.620 -0.050 -0.317 -0.220 0.201 0.195 0.893
Eigenvalue 3.073 2.044 1.485 2.118 1.697 1.421 1.242
Proportion 38.413 25.549 18.563 26.480 21.208 17.764 15.521
Cumulative 38.413 63.962 82.525 26.480 47.688 65.451 80.973

rate and composition was measured, and the volume data were
averaged to per-minute values, to fit with the simultaneous data related
to particle size. Table 4 shows the factor loading of PCA with traffic
volume. On the urban expressway, PC1 factors had a high loading on
particles in the size range of 0.4–4.99 μm, suggesting that the behaviors
of particles in these size ranges were similar and related to each other.
PC2 factors exhibited a high loading on particles in the size ranges of
larger than 5.0 μm and traffic flow, which may indicate that motor
vehicle can lead to an increase in the number of coarse particles. PC3
was characterized by significant loading on particles in the size ranges
of 0.25–0.39 μm, indicating that these particles represented behaviors
different from those in other size ranges. On the collector road, PC1 was
characterized by significant loading on traffic volume and particles in
the size range of 0.6–10 μm, with 38.41% of the variance, PC2 factors
are dominated with high loading on particles in the size range of
0.25–0.59 μm, these account for 25.55% of the total variance. Particles
in the size ranges between 5.0–9.99 μm and larger than 10 μm
represented high loading factors, with 18.56% variance in PC3. On
the local road, PC1 factors exhibited high loading on particles in the
size range of 0.25–0.59 μm, PC2 factor had a high loading for particles
in the size ranges of 0.6–4.99 μm, and PC3 factors showed high loading
on particles in the size larger than 5 μm, which may imply that the
number concentration of particles for these three size range were
different. PC4 factors had a high loading on particles in the size range of
1.0–1.99 μm and traffic flow. Analogously, measured traffic volume as
well as the particle number concentrations on arterial road was also
analyzed, and revealed some differences between arterial road and
other types of roads. There wasn’t obvious relationship between traffic
volume and particle number concentrations on the arterial road. The
explanation for this could be that there are more vehicles on the arterial
road, the roads are relatively narrower and the buildings are relatively
higher, which influence the atmospheric turbulence and dispersion of
particles.
Overall, it was found that traffic flows on urban expressways,
collector roads and local roads all have an important influence on the
particle number concentration, especially for the particles in the size
range of 1.0–9.99 μm, Whereas on arterial roads, traffic volume has no
obvious impacts on particle number concentrations due to the different
street configurations between the different types of roads.
In order to better understand the correlations between particle
number concentrations and parameters above, such as average wind
speed, linear regression was used. Average wind speed and the
corresponding average PM number concentrations on the arterial road
are depicted in Fig. 6. It also can be found that number concentration of
particles size greater than 0.6 μm is negatively related to the wind
speed, whereas correlation between particles size less than 0.6 μm and
wind speed is not high. This is because the unstable airflow caused by
traffic flow and atmospheric environment makes particles size less than
0.6 μm decrease along with the increase of mechanical turbulence.
Particles size greater than 0.6 μm because of its own gravity and inertia
force is highly negative correlated with wind speed.
3.4. PM mass concentrations for different types of roads
The minutely averaged mass concentrations of PM10, PM2.5, PM1,
which were measured at four types of roads, are shown in Fig. 7. The
data were selected under good weather conditions, and averaged to
represent data for one day. The results indicate that the average
concentration of PM10 was almost 2.5 times and 3.6 times higher than
those of PM2.5 and PM1, respectively, for every type of road.
According to the data displayed in Fig. 7, the PM10 mass concen-
tration on the urban expressway was obviously the highest of all the
road types. However, the PM2.5 and PM1 values do not reflect
significant differences from each other. In addition, when taking
background concentration into consideration (Fig. 8), it could be found
that the PM differences among the four road types was probably caused
by background concentration, not the characteristic of the road type
itself, for the ratio of background concentration of PM2.5 to measured
PM2.5 were all more than 60%. In a word, the result may imply that
only the PM2.5 concentration indicator cannot reflect the PM2.5
emissions of urban roads completely.
Fig. 9 represents the change trend of PM over a period of time. It can

481
Z. Qiu et al. Sustainable Cities and Society 32 (2017) 475–485

Fig. 6. Profiles of particle concentration (in number) for average wind speed on the arterial road.

482
Z. Qiu et al.
Fig. 7. PM mass concentrations for the four road types.
be seen that the PM1 mass concentration varied little with time,
remaining at around 30 μg m−3. For the PM2.5 and PM10 concentra-
tions, however, fluctuation occasionally appeared, especially for PM10.
This result may possibly be due to the dust and fly ash generated when
pedestrians pass along the road and garbage trucks pass by.
The relationship between PM mass concentrations and meteorolo-
gical parameters was investigated using correlation analysis (Table 5).
It was found that PM10 concentration, versus PM2.5 and PM1
concentrations, showed significant correlation. Similar results showed
up for PM2.5 versus PM1 on the urban expressway. This indicates that
the PM10, PM2.5 and PM1 were all from the same emission source.
On the other types of roads, PM10 versus PM2.5 showed significant
correlation, as did PM2.5 versus PM1. However, PM10 showed poor
correlation with PM1. Therefore, PM10 and PM1 may have been
contributed by different sources for this road type. On roads other than
the expressway, there are more pedestrians, and this additional
pedestrian traffic may result in the re-suspense of existing particles
(Li et al., 2016), forming coarser particles.
When the meteorological factors were analyzed, PM2.5 and PM1
were positively correlated with temperature and negatively correlated
with relative humidity, implying that temperature and relative humid-
ity had a great impact on PM2.5 and PM1 on the urban expressway.
However, on the arterial road and the collector road, only PM1 was
highly influenced by relative humidity, and on the local road, PM10
was obviously influenced by temperature. In other words, meteorolo-
gical parameters (temperature and relative humidity) have obvious
impacts on mass concentrations of coarse particles on the local road,
which is different from other three types of roads. It could be explained
Fig. 8. The ratio of background PM2.5 to measured PM2.5.
483
Sustainable Cities and Society 32 (2017) 475–485
by the road being much narrow, less vehicles and many buildings here,
then resulting in coarse particles increased on the local road.
4. Summary and conclusions
The pedestrian exposure to PM10, PM2.5, and PM1 mass and
particle number concentrations, in several particle-size ranges, at
various typical urban roadways (urban expressway, arterial road,
collector road, and local road), were investigated in this research.
Correlations between the PM mass and particle number (of different
sizes) concentrations, traffic volume and meteorological parameters
were analyzed, using the principal component analysis (PCA) method.
The following conclusions were drawn:
• The selected urban expressway showed the highest mean particle
number concentration for the size range of 0.25–32 μm, of all the
urban road types. However, the mass concentration of particles,
especially for PM2.5 and PM1, varied little among the four different
road types.
• The effect of meteorological parameters on particle number con-
centration differed among the four road types. Wind speed has a
large negative influence on the particle number concentration on
arterial road and local road, especially for the particles in the size
range of 0.6–9.99 μm, relative humidity has negative influence on
particles size range from 0.25–0.59 μm on three types of roads
except expressway. In other words, meteorological parameters have
no obvious impacts on particle number concentration on the urban
expressway. It could be explained by that the road being much
Z. Qiu et al. Sustainable Cities and Society 32 (2017) 475–485

Fig. 9. Variation of mass concentration of PM with time (on March 14th, at urban expressway).

Table 5 and Technology Project of the Ministry of Housing and Urban-rural

Correlation between PM concentrations and temperature, relative humidity and wind Development of the People’s Republic of China (No. 2016-K2-032), and
speed. National Natural Science Foundation of China under Projects
PM10 PM2.5 PM1 Wind speed temperature Relative humidity
#21607014.

Urban expressway References

PM 10 1 0.893 0.895 0.379 0.295 −0.348
PM 2.5 0.893 1 0.979 0.409 0.503 −0.510
Adams, H. S., Nieuwenhuijsen, M. J., & Colvile, R. N. (2001). Determinants of fine particle
PM 1 0.895 0.979 1 0.427 0.536 −0.545
(PM2.5) personal exposure levels in transport microenvironments, London, UK.
Arterial road Atmospheric Environment, 35, 4557–4566. http://dx.doi.org/10.1016/S1352-
PM 10 1 0.863 0.478 −0.480 −0.454 −0.227 2310(01)00194-7.
PM 2.5 0.863 1 0.788 −0.483 −0.269 0.075 Bereitschaft, B. (2015). Pedestrian exposure to near-roadway PM2.5 in mixed-use urban
PM 1 0.478 0.788 1 −0.264 −0.043 −0.557 corridors: a case study of Omaha, Nebraska. Sustainable Cities and Society, 15, 64–74.
http://dx.doi.org/10.1016/j.scs.2014.12.001.
Collector road Boarnet, M. G., Houston, D., Edwards, R., Princevac, M., Ferguson, G., Pan, H., et al.
PM 10 1 0.691 0.309 −0.142 −0.160 0.181 (2011). Fine particulate concentrations on sidewalks in five Southern California
PM 2.5 0.691 1 0.818 0.127 0.305 0.435 cities. Atmospheric Environment, 45, 4025–4033. http://dx.doi.org/10.1016/j.
PM 1 0.309 0.818 1 0.355 0.643 −0.619 atmosenv.2011.04.047.
Buonanno, G., Fuoco, F. C., & Stabile, L. (2011). Influential parameters on particle
Local road exposure of pedestrians in urban microenvironments. Atmospheric Environment, 45,
PM 10 1 0.658 0.567 0.155 0.538 0.499 1434–1443. http://dx.doi.org/10.1016/j.atmosenv.2010.12.015.
PM 2.5 0.658 1 0.927 −0.052 0.347 0.236 Cepeda, M., Schoufour, J., Freak-Poli, R., Koolhaas, C. M., Dhana, K., Bramer, W. M., et al.
PM 1 0.567 0.927 1 0.028 0.164 0.121 (2017). Levels of ambient air pollution according to mode of transport: a systematic
review. The Lancet, 2, e23–34. http://dx.doi.org/10.1016/S2468-2667(16)30021-4.
Charron, A., & Harrison, R. M. (2005). Fine (PM2.5) and coarse (PM2.5-10) particulate
matter on a heavily trafficked London highway: sources and processes. Environmental
wider, and that there are more than many branches of roads on both Science & Technology, 39, 7768–7776. http://dx.doi.org/10.1021/es050462i.
sides of the expressway results in complex atmospheric variations. Cheng, Y., Chang, H., & Hsieh, C. (2011). Short-term exposure to PM10, PM2.5, ultrafine
• Traffic flows on urban expressways, collector roads and local roads particles and CO2 for passengers at an intercity bus terminal. Atmospheric
Environment, 45, 2034–2042. http://dx.doi.org/10.1016/j.atmosenv.2011.01.061.
all have an important influence on the particle number concentra-
Finn, D., Clawson, K. L., Carter, R. G., Rich, J. D., Eckman, R. M., Perry, S. G., et al.
tion, especially for the particles in the size range of 1.0–9.99 μm. (2010). Tracer studies to characterize the effects of roadside noise barriers on near-
Whereas on arterial roads, traffic volume has no obvious impacts on road pollutant dispersion under varying atmospheric stability conditions. Atmospheric
Environment, 44, 204–214. http://dx.doi.org/10.1016/j.atmosenv.2009.10.012.
particle number concentrations due to the different street config-
Gulliver, J., & Briggs, D. J. (2004). Personal exposure to particulate air pollution in
urations between the different types of roads. transport microenvironments. Atmospheric Environment, 38, 1–8. http://dx.doi.org/
• PM10, PM2.5 and PM1 were all generated from the same emission 10.1016/j.atmosenv.2003.09.036.
Jolliffe, I. T. (1986). Principal component analysis. New York: Springer.
source on the urban expressway. On the other types of roads,
Karner, A. A., Eisinger, D. S., & Niemeier, D. A. (2010). Near-roadway air quality:
however, PM (especially PM10 and PM1) was generated from synthesizing the findings from real-world data. Environmental Science & Technology,
different sources (i.e. vehicles, pedestrians, or areas surrounding 44, 5334–5344. http://dx.doi.org/10.1021/es100008x.
the roads), and may be attributed to re-suspense of the existing Li, Y., Zhang, D., Jin, X., Yang, L., Du, X., & Yang, Y. (2016). Transportation
characteristics of motor vehicle pollutants near Beijing typical expressway. Science
particles (caused by the movement of pedestrians) to form coarser China Technological Sciences, 59, 457–468. http://dx.doi.org/10.1007/s11431-015-
particles. Meteorological parameters (temperature and relative 5997-7.
humidity) have obvious impacts on mass concentrations of coarse Moore, A., Figliozzi, M., & Monsere, C. M. (2012). Air quality at bus stops, empirical
analysis of exposure to particulate matter at bus stop shelters. Transportation Research
particles on the local road, which is different from other three types Record: Journal of the Transportation Research Board, 2270, 76–86. http://dx.doi.org/
of roads. It could be explained by the road being much narrow, less 10.3141/2270-10.
vehicles and many buildings here, then resulting in coarse particles Nikolova, I., Janssen, S., Vos, P., Vrancken, K., Mishra, V., & Berghmans, P. (2011).
Dispersion modelling of traffic induced ultrafine particles in a street canyon in
increased on the local road. Antwerp, Belgium and comparison with observations. Science of the Total
Environment, 412, 336–343. http://dx.doi.org/10.1016/j.scitotenv.2011.09.081.
Acknowledgments Qiu, Z., & Peng, X. (2015). Investigating the impact of urban grade-separation on
pedestrian PM2.5 exposure. Clean Technologies and Environmental Policy, 17,
1917–1927. http://dx.doi.org/10.1007/s10098-015-0909-x.
This study was supported by the Fundamental Research Funds for Rakowska, A., Wong, K. C., Townsend, T., Chana, K. L., Westerdahl, D., Ng, S., et al.
the Central Universities (No. 310822173702, 310822152006), Science (2014). Impact of traffic volume and composition on the air quality and pedestrian

484
Z. Qiu et al.
exposure in urban street canyon. Atmospheric Environment, 98, 260–270. http://dx.
doi.org/10.1016/j.atmosenv.2014.08.073.
Sabaliauskas, K., Jeong, C., Yao, X., Jun, Y., & Evans, G. (2013). Cluster analysis of
roadside ultrafine particle size distributions. Atmospheric Environment, 70, 64–74.
http://dx.doi.org/10.1016/j.atmosenv.2012.12.025.
Schneider, I. L., Teixeira, E. C., Oliveira, L. F. S., & Wiegand, F. (2015). Atmospheric
particle number concentration and size distribution in a traffic–impacted area.
Atmospheric Pollution Research, 6, 877–885. http://dx.doi.org/10.5094/APR.2015.
097.
Wang, X., & Gao, H. O. (2011). Exposure to fine particle mass and number concentrations
485
Sustainable Cities and Society 32 (2017) 475–485
in urban transportation environments of New York City. Transportation Research Part
D, 16, 384–391. http://dx.doi.org/10.1016/j.trd.2011.03.001.
Wang, Y., Zhu, Y., Salinas, R., Ramirez, D., Karnae, S., & John, K. (2008). Roadside
measurements of ultrafine particles at a busy urban intersection. Journal of the
Air & Waste Management Association, 58, 1449–1457. http://dx.doi.org/10.3155/
1047-3289.58.11.1449.
Zhu, Y., Hinds, W. C., Kim, S., & Sioutas, C. (2002). Concentration and size distribution of
ultrafine particles near a major highway. Journal of the Air & Waste Management
Association, 52, 1032–1042. http://dx.doi.org/10.1080/10473289.2002.10470842.