lable at ScienceDirect

Atmospheric Environment 98 (2014) 260e270

Contents lists avai

Atmospheric Environment

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

Impact of traffic volume and composition on the air quality andpedestrian exposure in urban street canyon

Agata Rakowska a, Ka Chun Wong a, Thomas Townsend a, Ka Lok Chan a,Dane Westerdahl a, b, Simon Ng c, Gri�sa Mo�cnik d, Luka Drinovec d, Zhi Ning a, b, *

a School of Energy and Environment, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong Special Administrative Regionb Guy Carpenter Climate Change Centre, City University of Hong Kong, Tat Chee Avenue, Kowloon, Hong Kong Special Administrative Regionc Civic Exchange, Hong Kong Special Administrative Regiond Aerosol d.o.o., Ljubljana, Slovenia

h i g h l i g h t s

� Mobile measurements were used to assess urban air quality and pedestrian exposure.� Traffic impact on street canyon and roadside air quality was investigated.� Formation of ultrafine particles in roadside environments was observed.� Transport policy needs to consider traffic volume and composition in urban canyon.

a r t i c l e i n f o

Article history:Received 5 May 2014Received in revised form27 August 2014Accepted 28 August 2014Available online 28 August 2014

Keywords:Urban transportStreet canyonBlack carbonUltrafine particlesRoadside pedestrian

* Corresponding author. School of Energy and EnvHong Kong, Tat Chee Avenue, Kowloon, Hong Kong S

E-mail address: zhining@cityu.edu.hk (Z. Ning).

http://dx.doi.org/10.1016/j.atmosenv.2014.08.0731352-2310/© 2014 Published by Elsevier Ltd.

a b s t r a c t

Vehicle emissions are identified as a major source of air pollution in metropolitan areas. Emission controlprograms in many cities have been implemented as part of larger scale transport policy interventions tocontrol traffic pollutants and reduce public health risks. These interventions include provision of traffic-free and low emission zones and congestion charging. Various studies have investigated the impact ofurban street configurations, such as street canyon in urban centers, on pollutants dispersion and roadsideair quality. However, there are few investigations in the literature to study the impact of change of fleetcomposition and street canyon effects on the on-road pollutants concentrations and associated roadsidepedestrian exposure to the pollutants. This study presents an experimental investigation on the trafficrelated gas and particle pollutants in and near major streets in one of the most developed businessdistricts in Hong Kong, known as Central. Both street canyon and open roadway configurations wereincluded in the study design. Mobile measurement techniques were deployed to monitor both on-roadand roadside pollutants concentrations at different times of the day and on different days of a week.Multiple traffic counting points were also established to concurrently collect data on traffic volume andfleet composition on individual streets. Street canyon effects were evident with elevated on-road pol-lutants concentrations. Diesel vehicles were found to be associated with observed pollutant levels.Roadside black carbon concentrations were found to correlate with their on-road levels but with reducedconcentrations. However, ultrafine particles showed very high concentrations in roadside environmentwith almost unity of roadside/on-road ratios possibly due to the accumulation of primary emissions andsecondary PM formation. The results from the study provide useful information for the effective urbantransport design and bus route reorganization to minimize the impact of traffic emissions on the urbanair quality and public health. Observations on the elevated ultrafine particle concentrations in roadsidepedestrian levels also demonstrate the urgent need to improve roadside air quality to reduce pedestrians’health risks especially inside street canyon.

© 2014 Published by Elsevier Ltd.

ironment, City University ofpecial Administrative Region.

1. Introduction

Extensive epidemiological evidence has shown the associationbetween human exposures to atmospheric pollution with various

A. Rakowska et al. / Atmospheric Environment 98 (2014) 260e270 261

adverse health effects and overall mortality (Pope et al., 2002;Kunzli et al., 2010). In urban areas, emissions from roadwaytransport are a major source of air pollution, especially particulatematter (PM), despite extensive emission control measures targetingmotor vehicles (Pant and Harrison, 2013; Yuan et al., 2013).Recently, theWorld Health Organization classified diesel exhaust asGroup I Carcinogen (WHO, 2012) and various studies have showntheir adverse health impact, for example lung cancer risks (Beelenet al., 2008) and worsening respiratory health of children(Rosenlund et al., 2009). This is of particular concern in denselypopulated cities, where large volume of traffic is often in closeproximity to the population to enhance the people's mobility, but atthe same time increases the population exposure to the traffic-induced pollutants, posing a great threat to the public health.

Roadside air quality in cities has drawn increasing attentionespecially in areas where many roadways are lined by dense andhigh-rise buildings forming street canyon that greatly limits thedispersion of mobile emissions (Weber et al., 2006). Differences incanyon geometry, traffic intensity and mixture, and ambientmeteorology result in variable air pollutants concentrations insidethe street canyon (Kim and Baik, 2004; Liu et al., 2005). Resultsfrom field measurements and computational fluid dynamics sim-ulations report that street canyons are generally hot-spots of airpollution in urban areas (Van Dingenen et al., 2004). For example,Zwack et al. (2011a, b) found significant increases of ultrafine par-ticle number and PM2.5 concentrations in street canyon versus ur-ban background in Manhattan, New York; Liu et al. (2005)simulated deep street canyons and found enhanced pollutantentrainment with larger aspect ratios (building-height-to-street-width) worsening street level air quality. Secondary pollutant for-mation inside street canyons was also suggested in a few recentmodeling studies (Baik et al., 2012; Kim et al., 2012; Bright et al.,2013).

To improve urban air quality, many local governments haveresponded by introducing various mitigation strategies and trafficpolicies intended to reduce the impact of traffic on roadside airpollution levels on busy urban roads. For example, low emissionszones (LEZ) designed to limit high emitters in city centers havebeen implemented in over 150 cities in EU countries (Wolff andPerry, 2010). Congestion Charging Schemes (CCS) directed torestrict certain types of cars from entering the inner city have beenimplemented in many cities, such as Milan, London and New York.Various studies have also demonstrated the effectiveness ofdifferent mitigation measures in the reduction of air pollutionlevels (Invernizzi et al., 2011; Boogaard et al., 2012). However, forcities and areas with complex urban built environment androadway infrastructure, there was also a report that showed nosubstantial benefits on the change of certain pollution levels(Ruprecht and Invernizzi, 2009). The limited understanding of thequantitative influence of marginal changes in traffic volume andmix on roadside air quality complicates cost-effective transportpolicy formulation and implementation. This is especially true forurban street canyons where there are both high traffic and pedes-trian flows in close proximity, and evaluation of potential reductionof roadside pedestrian exposure to the traffic pollutants is of greatsignificance and special interests for environmental and healthbenefits analysis.

There are few studies in the literature that have investigated theimpact of change in traffic volume/composition on both air qualityand pedestrian exposure to the traffic pollutants. This study pre-sents an experimental investigation on the traffic-related gas andparticulate pollutants in Central district, one of the most developedbusiness districts in Hong Kong. Both street canyon and openroadway configurations were covered in the study design andmobile measurement techniques were deployed to investigate both

on-road and roadside pedestrian area pollutants concentrations.Multiple traffic counting points along the sampling routes wereimplemented concurrently to monitor the change of traffic volumeand mix in individual streets and to investigate their relation withthe measured pollutants concentrations.

2. Methodology

2.1. Experimental site

The study was conducted in Central district, Hong Kong, locatedalong the north shore of Hong Kong Island as shown in Fig. 1. It isone of the busiest commercial and business districts in the city,characterized with heavy traffic conditions, high roadside pedes-trian usage and dense, high-rise buildings along the streets. Thestudy area encompasses the busy westeeast traffic corridor con-necting Island West and Causeway Bay. To investigate the streetlevel and roadside pedestrian area air pollutant concentrations, thesampling routes were designed to include streets with differenttraffic and built environmental conditions. Three main streetsincluded in the study are Connaught Road Central (CRC), Des VoeuxRoad Central (DVRC) and Queen's Road Central (QRC). A detaileddescription of the streets and their features is included below:

� The section of Connaught Road Central in this study is a wideroadway of 8 lanes e 4 in each direction-parallel to the northshore as shown in Fig. 1. The north side of Connaught Road isopen to the harbor coast with only scattered single tall buildingspresent, allowing easier dispersion of air pollutants than themore inland roads. The south side is dominated by dense high-rise buildings. The traffic is dominated by private vehicles andtaxis;

� The section of Des Voeux Road Central is parallel to ConnaughtRoad with 4 traffic lanes, 2 in each direction, and two tramtracks as shown in Fig. 1. Both sides of the street are lined withdense high rise buildings forming a typical urban street canyonof the Central area. The traffic mainly consists of buses anddiesel goods vehicles;

� The section of Queen's Road Central has similar street canyonfeatures as Des Voeux Road with only two to three lanes in totalas shown in Fig. 1. The very narrow streets with high risebuildings along both sides greatly confine the dispersion of on-road pollutants. Traffic mainly consists of private vehicles andtaxis.

2.2. Design of sampling protocol

Four full sampling days were investigated in summer and fall2013, with threeweekdays of July 31, August 3 and October 27 and aSunday of October 28, 2013. Prior to the actual sampling campaign,one day of trial sampling was carried out (June 30, 2013) but thedata were not included in the analysis due to instrument mal-functioning and incomplete pollution data. The traffic patternobserved on Saturday of October 27 is similar to other weekdays soit is counted as aweekday. This is consistent with business practicesin Hong Kong, where Saturday commerce is very much as what isseen during the five conventional weekdays. For each sampling day,two to three matched time periods were selected for investigationcovering morning rush hour from 0730 to 0930 h, noon time from1200 to 1400 h, and evening rush hour from 1830 to 2030 h. Duringeach sampling period, nine traffic counting points for traffic volumeand fleet composition recording were deployed along the threeroadway sections. Two mobile sampling routes were developed toinclude major overlaps along the three mains streets for

Fig. 1. The map of study area with two sampling routes and nine traffic counting points. Red dash line is the on-road measurement route and blue solid line is the roadsidemeasurement route. The traffic volume and composition were recorded at nine counting points marked as AA1, BB1, CC1 on Connaught Road Central, AA2, BB2, CC2 on Des VoeuxRoad Central, AA3, BB3, CC3 on Queen's Road Central; Bottom pictures show street view of the streets. (For interpretation of the references to color in this figure legend, the reader isreferred to the web version of this article.)

A. Rakowska et al. / Atmospheric Environment 98 (2014) 260e270262

comparison of on-road and roadside pollutant concentrations wereperformed concurrently (Fig. 1). Typically, 4e6 runs were repeatedfor the on-road mobile route and 2e3 repeated runs for roadsideroute with return trips in each two-hour sampling session.

2.2.1. On-road mobile route for street level measurementsThe on-road measurements were performed by the modified

“On-road Plume Chasing and Analysis System (OPCAS)” which is amobile platform with high time resolution air monitors on-board,

designed for plume chasing on roadways (Ning et al., 2012). TheOPCAS used in the study was reconfigured with the inlet installedone meter above the platform roof to limit impact from individualvehicle plume and sample the on-road pollutant concentration.While driving along the preset sampling route, the mobile platformfollowed the traffic and kept a distance of at least 10e15 m awayfrom any preceding vehicle. We used two separate sampling lineseone for gaseous pollutant of NOx; and one for particulate pollutantsincluding PM2.5, ultrafine particle (UFP) number and black carbon

A. Rakowska et al. / Atmospheric Environment 98 (2014) 260e270 263

(BC). The details of the measuring devices are listed in Table 1. TheAethalometer (Model AE33, Magee Scientific) used in the study tomeasure BC features a parallel light absorption measurement ontwo sampling channels (from the same inlet stream) with differentrates of aerosol accumulation and hence different aerosol loadingon the filter. Using measurements of light attenuation throughthese two sample spots, filter loading effects are quantified and acompensated BCmeasurement is obtained using a mass absorptioncross-section 7.77 m2/g (Drinovec et al., 2014). Teflon tubing andfittings were used for the gas lines and conductive tubing andstainless steel fittings were used for particulate measurements.Additionally, a high resolution Global Positioning System (GPS) wasused to collect vehicle location data and speed. A camera wasinstalled in front window to record the on-road conditions forquality assurance while post-processing the air pollution data.Video files were useful to assess on-road traffic conditions andidentify individual emission events. All equipment was operated athigh time resolution (1 se4 s). A Matlab script was developed fordata acquisition through RS232 communication for all equipment.The simultaneous collection of the real-time data from all themeasuring devices reduces misalignment of data and hastens datareview. Data synchronization for each pollutant channel wasfurther validated by insertion and removal of an HEPA filter at theinlet and identification of concentration peaks during on-roadmeasurements.

2.2.2. Roadside route for pedestrian level measurementRoadside pollutants measurement was performed using the

Mobile Exposure Measurement System (MEMS), which is a wheel-equipped suitcase configured with portable air monitors and a dataacquisition system. The MEMS and its use for personal exposureassessment is described in a prior study (Yang et al., 2014). Table 1shows the list of equipment to measure black carbon and UFPnumber concentrations. The GPS device employed in the currentstudy included a portable weather station with an automaticheading sensor (Airmar Model 150WX, Milford, USA) to record thestreet level wind speed. Monitors in the suitcase were connected toan on-board miniPC (NUC, Intel), while all the real-time data weretransferred via Bluetooth to a mobile phone carried by the operator.Pollutants concentrations and corresponding location informationwere also recorded through a customized mobile application.Fig. S1 shows the setup of the suitcase and interface of the App. Theroadside measurements were carried out on October 27 and 28 andthe suitcase carrier walked along the designed pedestrian routerepeatedly during each sampling period. While performing theroadside measurement along the route, special attention was

Table 1The equipment of OPCAS and MEMS for on-road and roadside measurement.

Pollutants Model and manufacturer Des

OPCAS for on-road measurementsNO, NO2, NOx CLD66, Ecophysics CheCO2 Carbocap GMP343,

Vaisala Inc.No

UFP number CPC 3007, TSI Confor

PM2.5 mass ES-642, Metone Inc. LigBlack carbon AE33, Magee Scientific BlaMEMS for roadside pedestrian level pollutantsUFP number CPC 3007, TSI Con

forBlack carbon AE51, AethLabs MicMeteorological

conditionsAirmar Model 150WX,Milford, USA

Str

CO, CO2 Q-Trak 7575, TSI COCOTem

dedicated to the possible smoking event surrounding the fieldoperator and a tag of smoking was marked in the mobile app asshown in Fig. S1 for data screening prior to data analysis. In order toachieve consistency of the roadside measurements for cross com-parison, the operator walked along the middle line of pedestriansidewalk in different streets. Two or three cycles were completedwithin the 2-h period.

2.2.3. Traffic counting and classificationConcurrent with on-road and roadside measurements, traffic

volume and fleet composition were counted at nine countingpoints along the route as shown in Fig. 1. Traffic surveys wereconducted during all the sampling sessions and included the clas-sification of vehicles into eight categories: private cars, motorcy-cles, taxis, light and heavy goods vehicles, public light buses, doubledecker buses and non-franchised buses. Each counting point wasassigned two observers and each observer counted one lane oftraffic. For multi-lane streets, the traffic conditions were alsorecorded by video cameras for post-analysis and off-site trafficcounting. Each of the 2-h periods was divided into eight 15-min segments and the vehicle number was counted in alternating15-min sessions, followed by averaging the counts in the sessionsfor hourly traffic volumes of each vehicle type. The calculated trafficvolume data were then re-grouped into major groups, includingPetrol Private Cars, LPG Vehicles (including taxi and public lightbuses), Diesel Buses, Diesel Trucks and Vans, and the remainingfraction as Other. The grouping reflects the fleet-based functions forprivate and public transport or goods transport, and also differentfuels used. The traffic volume and fleet composition analysis followthe same classification hereafter in the study.

2.3. Data analysis

2.3.1. Data quality assurance and quality controlPrior to the sampling campaign, the gas analyzerswere calibrated

including zero and span using either standard gas (CO2, Linde) for theCO2 analyzer and a gas calibrator (Model 408 NO Calibration Source,2B technologies) for the NOx analyzer. The particle instruments un-derwent standard zero and flow checks. During the field campaign,intensive quality assurance measures were implemented includingcontinuous monitoring of instrument operation conditions throughOPCAS data acquisition software, and maintenance of on-board in-struments in between the sampling sessions.

For the two CPCs used for OPCAS and MEMS, a side by sidecomparisonwas performed and the correlation coefficient was 0.95with a slope of 0.98 on the basis of one-minute averaged data.

cription Time resolution

miluminescence 4 sn-dispersive infrared detector 2 s

densational particle (dp > 10 nm) counterUFP number concentration

1 s

ht-scattering laser nephelometer 1 sck carbon measurement by light absorption 1 s

densational particle (dp > 10 nm) counterUFP number concentration

1 s

roAethalometer for black carbon 1 seet level wind speed 1 s

2:non-dispersive infrared sensor 1 s: electro-chemical sensor 1 sperature and relative humidity 1 s

Fig. 2. Traffic volume and fleet composition in the streets during different time periods(AM: 0730e0930 h; Noon: 1200e1400 h; PM: 1830e2030 h).

A. Rakowska et al. / Atmospheric Environment 98 (2014) 260e270264

During the field campaign, UFP number concentrations higher than100,000 particles cm�3 were corrected for coincidence errorfollowing the approach by Westerdahl et al. (2005). In further dataprocessing, the UFP number concentration data from the two CPCswere compared directly. For the black carbon measurements, on-road AE33 data were used as it is for further analysis and the rawBC data from roadside AE51were compensated for the filter loadingeffects, following the equation (Ning et al., 2012).

BC ¼ BC00:88 expð�ATN=100Þ þ 0:12

(1)

where BC and BC0 represent the corrected and raw BC concentra-tions, respectively, and ATN is the attenuation of light through thesample laden filter. After attenuation correction, the on-road rackmount AE33 in OPCAS and handheld microAethalometer AE51 inMEMS were validated concurrently against Sunset OCEC analyzerusing NIOSH method 5040 prior to the campaign both with veryhigh correlation coefficients of 0.95 as shown in Fig. S2. The twospot design of AE33 efficiently corrects the loading effect auto-matically as demonstrated by the high correlation with theelemental carbon data. The slopes derived from the regression inFig. S2 were applied separately to the two Aethalometers andpresented as the BC (black carbon equivalent to EC) data in thestudy.

The on-road PM2.5 concentration measured by ES-642 waspresented without further correction. It is important to note thatthe compensation factors of photometers with gravimetric mea-surements are highly dependent on the physical and chemicalnature of aerosols, and the values may be different with changingenvironments. The PM2.5 concentration data are presented only forcross comparison of different streets with similar aerosol charac-teristics, and not intended to comparison with gravimetric PM2.5mass concentration.

2.3.2. On-road data processing proceduresThe on-road pollutant concentration data fromOPCAS first went

through data completeness and integrity check to remove timeperiods of malfunctioning instruments or missing GPS data, fol-lowed by reduction to 10 s moving average for further data review.Only very few data (<5%) were tagged as invalid based on afore-mentioned criteria with intensive quality assurance and qualitycontrol measures. The concentration data were then integrated bygeo-tagging the GPS coordinates for each time period, followed bysegregating them into sections covering the three streets underinvestigation. In order to link the measured on-road data with thetraffic volume and composition, the geo-tagged pollutants con-centration data on each street were further separated into threesections, each centered by the traffic counting point, e.g., AA1, BB1,… , CC3 as shown in Fig. 1. At this stage the on-board videorecording was reviewed to confirm the reported pollutants con-centration data were not from individual vehicle plume. The re-ported pollutants concentrations in each street section were theaverage of multiple runs (2e5 times) within each two-hour sam-pling period. In the analysis of individual fleet contribution to on-road air pollution, the measured pollutants concentrations werealso analyzed by linear regression with counted vehicle numberthrough each street section.

2.3.3. Roadside data processing proceduresThe roadside pollutants data from MEMS went through the

same quality assurance steps as on-road data. The measured pol-lutants concentrations with corresponding GPS coordinates wereused to identify the time periods when the measurements werecarried out along the three studied streets. Different from on-road

data processing, the roadside data were not further segregated bystreet sections due to the less frequent coverage of the sections bywalking to simulate a pedestrian's typical exposure. The reportedroadside pollutant concentrations along different streets were theaverage of the repeated runs performed during each time period.

3. Results and discussion

3.1. Traffic and air pollution pattern in weekdays

During the measurement campaign, a total of 8 sampling pe-riods were successfully conducted 6 during weekdays and 2 duringa Sunday. Fig. 2 shows the weekday traffic volume and fleetcomposition of CRC, DVRC and QRC streets at different time periods,each representing the average of traffic counts from the threecounting points on the individual street. Consistent traffic patternswere observed for the weekdays on each of the three streets, inwhich CRC has the highest hourly total traffic counts ranging from5500 to 7000, while DVRC and QRC had only 550e720 vehicles perhour. There was no substantial diurnal variation observed in thetraffic volume for all the streets during weekdays confirming thenature of busy business district of the study area. The dominantfleets in the streets are petrol private car or LPG vehicles (mostlytaxis) for CRC (34e46%), diesel franchised bus for DVRC (36e43%)and petrol car or LPG vehicles for QRC (24e46%) as shown in Fig. 2.The dominance of bus activity on DVRC is reflected by the largenumber of diesel bus stops setup along this road as shown in Fig. S3.

Fig. 3 shows the further breakdown of the traffic volume andfleet composition in the different sections of each street as anaverage of all time periods in weekdays, together with themeasured on-road pollutants concentrations. The total traffic vol-ume on CRC is consistently higher than the other two streets withan average ratio of 9.9. The composition of the fleets in differentsections on the same street is similar while the total traffic volumevaries slightly due to the diverging and merging traffic flowcrossing streets. The traffic related pollutants concentrations of BC,PM2.5, UFP number and NOx showed dramatic difference betweendifferent streets. Although CRC has substantially higher total trafficflow, the pollutants concentrations peak on the less traffickedDVRC, especially for section of CC2 with concentrations of67.8 mg m�3, 60,273 particles cm�3, 53.0 mg m�3, 542 ppbv for BC,

Fig. 3. Traffic volume and fleet composition versus street level NOx and black carbon,ultrafine particles number and PM2.5 concentration in weekdays. The error barsrepresent the standard deviation of the measured average concentration from differenttime periods in each roadway section.

Table 2Pearson correlation coefficients (R) between traffic fleets and on-road pollutants.

Street Pollutants Allvehicles

Petrolvehicles

LPGvehicles

Dieselvehicles

Dieselbus

Dieseltruck/van

CRC BC �0.29 0.27 �0.21 0.00 �0.15 �0.41NOx 0.33 0.37 0.38 0.00 0.45 �0.34

DVRC BC 0.66 0.54 0.54 0.61 0.57 0.35NOx 0.29 0.18 0.32 0.30 0.37 �0.27

QRC BC �0.41 0.65 �0.37 �0.25 �0.05 �0.37NOx 0.12 �0.01 0.37 �0.08 0.01 �0.15

Positive correlations with R higher than 0.35 are highlighted as bold.

A. Rakowska et al. / Atmospheric Environment 98 (2014) 260e270 265

PN, PM2.5 and NOx, respectively, clearly indicating the substantialimpact of the street canyon on the accumulation of on-road pol-lutants concentrations. Previous studies have also shown thatstreet canyons significantly increase the pollutant concentrationsrelative to the immediate areas around the roadway (Britter andHanna, 2003; Weber et al., 2006). In the present study, due to thedifference in traffic volume, the on-road pollutants concentrationsin the same street also vary. Section CC2 in DVRC has 4.2,1.8, 1.9 and1.4 times higher concentrations than section AA2 for BC, PN, PM2.5and NOx, respectively. Compared to DVRC with similar streetcanyon features, although QRC has similar total traffic volume, theon-road pollutants concentrations were much lower. A further lookat the traffic composition shows the dominant vehicle in DVRC arediesel franchised buses with an average fraction of 39%, followed bypetrol and LPG vehicles, while QRC is dominated with LPG andpetrol vehicles, each representing 40% and 34%. Previous in-vestigations of the on-road vehicles in Hong Kong showed dieselfleets are the major contributor to the PM emissions especiallyblack carbon, and to a lesser extent for NOx emissions (Ning et al.,2012). Additionally, the street canyon in DVRC and QRC signifi-cantly restricts the dispersion of traffic emissions. The portablemeteorological measurements along the roadside of differentstreets suggests that the road surface wind speeds in DVRC and QRCwere reduced compared with CRC as shown in Fig. S4. This isconsistent with the numerical simulation results showing verysmall ground-level wind speeds inside street canyon, making itextremely difficult for pollutants to disperse (Li et al., 2009). Inaddition, the meteorological data at Central and Central/Westernsites (http://envf.ust.hk/dataview/metplot/current/) within thestudy region recorded predominant and consistent easterly andnortheasterly wind directions during the measurement campaign.This would serve to move ground level pollutants along the road-ways, but the canyon effects of the buildings would restrict mixingof air between the major roadways that run in east to west andnorthwest to southeast direction. Significant difference in theobserved on-road pollutants concentrations is also a clear indica-tion of the impact of diesel fleets on the street level air quality andurban street canyon effects to the heterogeneity of spatial distri-bution of pollutants (Ducret-Stich et al., 2013).

Table 2 shows the Pearson correlation coefficients (R) betweenindividual fleets and observed on-road pollutants concentrationsduring the weekday sampling sessions for black carbon and NOx,respectively. For CRC, there is no significant correlation betweendiesel traffic fleets and BC concentrations and modest correlations

were found for NOx with diesel bus and major traffic fleets of petroland LPG vehicles in CRC. Thewide and open roadway feature of CRCfacilitates the dispersion of traffic pollutants from on-road trafficand also possibly brings pollutants from upwind harbor areas.Although diesel fleets are the major contributors to BC emissionsurban areas in general, dispersion seems to dominate the variationof BC concentrations, while NOx emissions are attributed to othermajor fleets as well (Ning et al., 2012), resulting in increased on-road concentrations with traffic volume (R ¼ 0.45 for diesel bus,0.38 for LPG and 0.37 for petrol vehicles). On the contrary, DVRCshowed a different pattern with modest and high correlations be-tween both pollutant concentrations and diesel fleets, especiallydiesel buses (R ¼ 0.57 for BC and 0.37 for NOx) which are thedominant fleet in the street with 36e43% of total traffic volume(Fig. 2), due to the strong canyon effect that suppresses thedispersion of traffic emissions (Weber et al., 2006) and accumula-tion of pollutants is evident with increased diesel vehicle numbers.Good correlations were also found for other fleets of petrol and LPGvehicles, although they are not major contributors to BC emissions,

possibly because their increased traffic volume results in slowertraffic or traffic congestion that extends the time for the highemitters to pass the street canyon (Daniel and Bekka, 2000). Similarobservation was also found in QRC that dominant petrol and LPGvehicles (24e46% in total traffic volume) have modest to goodcorrelation with on-road pollutants, while others didn't showpositive correlations. The contrast of the relation between trafficfleets and pollutant concentrations inside street canyon and openroadway indicates the importance of the transport policies inreducing traffic emissions and improving urban air quality. Forexample, identifying and diverting or removing the heavy emittersfrom streets with hot-spots may be a cost effective pollution controlmeasure as evidenced by high correlation between diesel fleets andon-road pollutant concentration inside street canyon. Many Euro-pean cities have implemented low emission zones in city centers inrecent years to control roadway pollutants, however, different de-grees of success have been reported on their environmental ben-efits (Wolff and Perry, 2010; Boogaard et al., 2012). In this study,although DVRC and CRC have similar roadside built environmentwith high rise buildings, their different correlation patterns be-tween traffic and pollutants concentrations suggests the necessityof comprehensive transport and air quality assessment prior to theimplementation of local transport policies.

3.2. Comparison of weekday and weekend observations

Fig. 4 shows the comparison of total traffic volume and fleetcomposition on Sunday and Monday for both morning and noontime periods. The traffic counts are the sum of three main streets toshow the change of overall traffic pattern in the district. Individualstreet traffic composition is discussed in the following section. Thetotal traffic volume Sunday morning decreased by 45% compared

Fig. 4. Comparison of total traffic volume and composition between Sunday, 27th, Oct, 2013 and Monday, 28th, Oct, 2013 for a) Morning and b) Noon periods. Traffic change in %represent the volume difference from Weekday (Monday) to Weekend (Sunday) in the same time period.

Fig. 5. Comparison of weekend (Sunday, 27 Oct, 2013) and weekday (Monday, 28 Oct, 2013) traffic pattern and on-road pollutants concentrations at different time periods: a)Sunday morning; b) Monday morning; c) Sunday noon; d) Monday noon.

A. Rakowska et al. / Atmospheric Environment 98 (2014) 260e270266

with that of Monday from 9035 to 4970 vehicles/h due to theregular day-off of the business activities that lowers the volume ofpublic transport by buses, daily commuting traffic by private carsand commercial goods delivery service with diesel vans and trucksduring regular morning rush hours. The traffic pattern duringSunday noon time, however, showed a very different trend, inwhich the total traffic volume remained almost the same (only 4%difference from Monday to Sunday), while the fleets compositiondemonstrated a dramatic change with volume of commercial diesel

vans and trucks decreased by 59%while all other fleets increased bydifferent extent, i.e., diesel bus for 10%, petrol private vehicles for14% and LPG taxi/buses for 3%, because of increased typical week-end activities in the central business district attracting passengervolume via private cars and public transport. The contrast in trafficvolume and composition also offers a unique opportunity for theinvestigation of their influence on air quality. In order to confirmthe consistency of urban background pollution between theconsecutive weekday and weekend for comparison between the

Fig. 7. On-road and roadside pollutants concentrations of: a) BC and b) ultrafineparticles number.

A. Rakowska et al. / Atmospheric Environment 98 (2014) 260e270 267

two days, the ambient air quality monitoring data from nearbystation were taken from the Hong Kong Environmental ProtectionDepartment. The station in Central Western, less than 1 km fromthe study area in Central district, recorded a daily average PM2.5concentration of 55.3 and 50.3 mg m�3, and NO2 concentration of25.4 and 25.1 ppbv, for Sunday and Monday, respectively, as shownin Fig. S5. Overall, there was no substantial difference in theambient air pollutants concentrations between the two daysshowing relatively stable urban background.

Fig. 5 presents on-road pollutant concentrations (NOx and BC)measured on different streets and corresponding traffic volumeand composition on weekdays and Sunday. Significantly lowertraffic volume on Sunday morning compared with Monday shows aconsistent trend with on-road pollutants concentrations with anaverage reduction of 41%, 40% and 18% for NOx, and 52%, 39% and36% for BC as measured on CRC, DVRC and QRC respectively. On theother hand, although noon time total traffic volume had only slightchange by 4%, the on-road pollutants concentrations recorded asubstantial reduction on Sunday compared with Monday. Takingsection CC2 of DVRC as one example as shown in Fig. 6, Sundaynoon time traffic was characterized by increased private and publictransport via buses and passenger cars, the difference in measuredpollutants concentrations is likely due to the much reduced num-ber of diesel goods vehicles suggesting their significant contribu-tion to the on-road pollutants concentrations although there wereno significantly high associations as shown in Table 2 due to thechange of traffic composition. Wind speed and wind directionwerenearly identical during these two days of sampling as reported fromthe twometeorology sites operated in the Central area. The findingsin the study suggest controlling high emitter fleets instead oflimiting total traffic volume is potentially a cost-effective measureof improving urban local air quality. Particularly, the reduction ofdiesel goods vehicles from the districts with narrow urban canyonscould have a positive impact on micro-scale air quality in thoseplaces.

3.3. On-road and roadside black carbon and ultrafine particleconcentrations

Fig. 7 shows on-road and roadside concentrations of black car-bon and ultrafine particle number measured on the differentstreets from concurrent measurements. It is worth noting that on-road concentrations represent the average of street level pollutantsconcentration from repeatedmeasurements along the entire street,while the roadside concentrations represent the pedestrian expo-sure to the pollutants walking along the same streets during the

Fig. 6. Comparison of fleets composition and pollutants concentrations in CC2, DVRC.AM: 0730e0930 h; NN: noon time from 1200 to 1400 h.

same period. The average and standard deviation of on-road blackcarbon concentrations from all streets are 39.2 ± 8.1 and20.7 ± 4.5 mg m�3 for Monday and Sunday, respectively, while thecorresponding UFP number concentrations are 45,503 ± 5352 and28,057 ± 2569 particles cm�3. The average roadside to on-roadpollutant ratios are 63 ± 23% and 107 ± 28% for BC and UFP,respectively. Many studies have documented the dispersion andconcentration profiles of traffic related pollutants near roadwayenvironments (Zhu et al., 2002; Pirjola et al., 2006), however, fewinvestigations have reported the observations of intermediateprocesses from on-road to roadside environments in complex ur-ban built environment and roadside configuration (Kozawa et al.,2012). The reduction of BC concentration as evidenced by theroadside to on-road ratios in the study indicates the dilution ofexhaust emissions from on-road to roadside environment domi-nated the dispersion of BC. On the other hand, during dispersion,the exhaust emissions may undergo physical and chemical trans-formation depending on the aerosol dynamical processes, such ascondensation, evaporation, nucleation etc (Zhang and Wexler,2004; Pirjola et al., 2006; Kumar et al., 2011). The dramatic differ-ence of the particle number concentration ratio (107 ± 28%)compared with black carbon (63 ± 23%) as observed in this study(p < 0.001) clearly suggests the role of nucleation in forming ul-trafine particles from on-road to roadside. The process may befurther enhanced by the street canyon effect in which higher con-centrations of traffic emissions of semi-volatile species may bepresent and contained inside the canyon that facilitate the

A. Rakowska et al. / Atmospheric Environment 98 (2014) 260e270268

nucleation formation of ultrafine particles (Kumar et al., 2008). Ourprevious investigation near a busy roadway showed rapidmixing ofblack carbon soot particles by semi-volatile species possibly viacondensation through mass size distribution measurements (Ninget al., 2013). The interesting relationship between BC and particlenumber is also evidenced in Fig. 8 which shows the correlationbetween roadside and on-road pollutant concentrations of BC andparticle number measured in different street environments. Muchbetter agreement between the two pollutants were found from on-road measurements (R2 ¼ 0.69) compared to roadside measure-ments (R2 ¼ 0.38). Correlations between BC and UFP numberconcentrations have been observed in different urban environ-ments with impact from vehicle exhaust emissions (Rodriguez andCuevas, 2007) due to their co-existence from exhaust emissions.The observed differences of correlations between BC and particlenumber further suggests different mechanisms that affect theirroadside concentrations. Whilst BC is a relatively inert and appro-priate tracer for primary traffic emissions, nucleation formation ofultrafine particles seems to have occurred in the short process fromon-road to roadside (Zhang andWexler, 2004). From the pedestrianexposure point of view, ultrafine particles exposure is a greaterhealth risk than PM2.5 due to their small size and ability to pene-trate deep in the lungs (Oberdorster et al., 2005). The high on-roadand roadside ratio of UFP number concentrations observed in the

Fig. 8. Correlation between BC and ultrafine particles number for: a) roadside and b)on-road.

study raises public health concern with pedestrian exposure insideurban street canyons where large numbers of pedestrians oftenpresent. Further investigations on the formation and nature of ul-trafine particles in near road environments are needed.

4. Conclusion and implication

The present study investigated the impact of traffic volume andfleet composition on the on-road and roadside air pollutant con-centrations with the settings of typical urban street canyon andopen roadways in one of the busiest business districts in HongKong. The mobile platforms instrumented with high resolution airmonitors proved to be a very useful tool to capture the spatial andtemporal variations of pollutant concentrations in complex urbanenvironments. Substantial concentration differences of trafficrelated pollutants, i.e., NOx, UFP, BC and PM2.5, were documented instreets within the small spatial distance of 2e3 blocks in the study.This reinforces concerns regarding reliance on regulatory-basedfixed air monitoring to capture concentration gradients or to esti-mate population exposure arising from local urban hot spots thatmight result from traffic emissions and/or the built urban envi-ronment. Compared with earlier studies in North America thatshowed more homogeneous spatial distribution of certain pollut-ants in large urban areas (Burton et al., 1996;Wilson and Suh,1997),the results from the present study suggested much higher hetero-geneity of the pollutants concentrations in more complex urbanterrain (Wilson et al., 2005; Zwack et al., 2011a, b). Similar findingsare reported in more recent literature in urban areas (Ito et al.,2004; Ducret-Stich et al., 2013), which poses an increasing chal-lenge to accurately quantify and assess human exposure to atmo-spheric pollutants in epidemiology studies. This also leads to arecent trend of shift from static ambient monitoring to spatio-temporally resolved personal exposure assessment (Steinle et al.,2013). To meet these challenges research activities haveexpanded to deploy appropriate air monitoring into these complexand highly polluted microenvironments where people live, transitand spend important parts of their days. Much higher temporalresolution and added pollution metrics have been included in thesemonitoring protocols.

This study also documented strong street canyon effects and itsimpact on the increased NOx and BC concentrations and elevatedultrafine particles possibly formed in street canyon conditions(Weber et al., 2006; Zwack et al., 2011a, b). Although there wasnearly ten times the traffic volume in a wide and open roadway(CRC in this study), its on-road and roadside pollutants showedmuch lower concentrations than the adjacent streets walled in busytall buildings (DVRC and QRC). The strong contrast observed in thestudy suggests the necessity of urban air quality managementthrough integrating urban planning strategy with their impact ontraffic related air pollution and pedestrian exposure to the pollut-ants. A variety of studies have attempted to link urban air qualitywith characteristic features and settings, such as street vegetationand roadside trees (Wania et al., 2012), roadside barriers (Bowkeret al., 2007; Ning et al., 2010), or distance from roadside struc-tures (Barros et al., 2013). Further efforts are needed to better un-derstand the interactions of urban built environment andpollutants concentrations as factors that impact human exposureand health risks in urban planning policy formulation.

Lastly, we have identified high association between the volumeof diesel powered vehicles and street level pollutants concentra-tions although they constitute only a fraction of traffic flow,showing that total traffic volume is not necessarily the only drivingfactor to the air quality. The finding implies traffic control strategytargeting total volume reduction may not be a cost effectiveapproach since small fraction of dirtiest vehicles actually contribute

A. Rakowska et al. / Atmospheric Environment 98 (2014) 260e270 269

a disproportionally high fraction of total emissions (Ning et al.,2012). Given the fact that vehicle emissions have become the ma-jor pollution sources in urban areas, many governments haveimplemented local transport policy to improve urban air quality ontop of tightening vehicle emission standard, such as low emissionzone in European cities (Wolff and Perry, 2010; Boogaard et al.,2012), congestion pricing in New York (Schaller, 2010), restrictionof odd/even license of private cars in Beijing (Zhou et al., 2010) etc.The benefits and cost-effectiveness of such transport policies inreducing urban air pollution concentrations need careful consid-eration and evaluation before they are implemented to assure thedesired air quality and public health improvements are realized,and the relation between pollutant concentrations and transportdesign in different urban built environments is the key factor(Nuvolone et al., 2009). This study provides a useful experimentalapproach to efficiently untangle the relationship between transportemissions, urban built environment and roadside air quality forfuture research investigation in support of effective policy making.

Acknowledgments

This study was supported by the Health and Medical ResearchFund, Food and Health Bureau, Hong Kong SAR Government(Ref. No. 10112061), and Environmental Conservation Fund, HongKong SAR Government (Ref. No. 01/2012). The work describedhereinwas also financed in part by the EUROSTARS grant E!4825 FCAeth and JR-KROP grant 3211-11-000519. The authors would like tothank WYNG Foundation and ADM Capital Foundation for theirkind support and the Hong Kong Environmental ProtectionDepartment (HKEPD) for their suggestions on policy implications.Lastly, the authors would like to thank the MVA and Civic Exchangefor their technical support in the traffic information collection andanalysis (in alphabetic order): Mr Keith Chan, Ms Olivia Chen, MsCaroline Cottet, Mr James Cunningham, Mr Martin Lai, Mr CurtisMak, Mr Joe Pang, Dr Hilings Yip, Ms Yan-yan Yip and CityU teammembers (in alphabetic order) for the assistance of field experi-ments: Dr. Nirmal Kumar Gali, Mr. Yang Hong, Miss Sabrina YananJiang, Ms. Flora Lau, Miss Li Sun, Miss Wai Ting Viki Tong, Mr.Maimaitireyimu Wubulihairen, Mr. Fenhuan Yang.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.atmosenv.2014.08.073.

References

Baik, J.J., Kwak, K.H., Park, S.B., Ryu, Y.H., 2012. Effects of building roof greening onair quality in street canyons. Atmos. Environ. 61, 48e55.

Barros, N., Fontes, T., Silva, M.P., Manso, M.C., 2013. How wide should be theadjacent area to an urban motorway to prevent potential health impacts fromtraffic emissions? Transp. Res. Part a Policy Pract. 50, 113e128.

Beelen, R., Hoek, G., van den Brandt, P.A., Goldbohm, R.A., Fischer, P., Schouten, L.J.,Armstrong, B., Brunekreef, B., 2008. Long-term exposure to traffic-related airpollution and lung cancer risk. Epidemiology 19 (5), 702e710.

Boogaard, H., Janssen, N.A.H., Fischer, P.H., Kos, G.P.A., Weijers, E.P., Cassee, F.R., vander Zee, S.C., de Hartog, J.J., Meliefste, K., Wang, M., Brunekreef, B., Hoek, G.,2012. Impact of low emission zones and local traffic policies on ambient airpollution concentrations. Sci. Total Environ. 435, 132e140.

Bowker, G.E., Baldauf, R., Isakov, V., Khlystov, A., Petersen, W., 2007. The effects ofroadside structures on the transport and dispersion of ultrafine particles fromhighways. Atmos. Environ. 41 (37), 8128e8139.

Bright, V.B., Bloss, W.J., Cai, X., 2013. Urban street canyons: coupling dynamics,chemistry and within-canyon chemical processing of emissions. Atmos. Envi-ron. 68 (0), 127e142.

Britter, R.E., Hanna, S.R., 2003. Flow and dispersion in urban areas. Annu. Rev. FluidMech. 35, 469e496.

Burton, R.M., Suh, H.H., Koutrakis, P., 1996. Spatial variation in particulate concen-trations within metropolitan Philadelphia. Environ. Sci. Technol. 30 (2),400e407.

Daniel, J.I., Bekka, K., 2000. The environmental impact of highway congestionpricing. J. Urban Econ. 47 (2), 180e215.

Drinovec, L., Zotter, P., Mo�cnik, G., Pr�evot, A.S.H., Ruckstuhl, C., Coz Diego, E.,Rupakheti, M., Mueller, T., Wiedensohler, A., Hansen, A.D.A., 2014. The “Dual-Spot” Aethalometer: improved measurement of aerosol black carbon with real-time loading compensation. Atmos. Meas. Tech. Discuss. (submitted forpublication).

Ducret-Stich, R.E., Tsai, M.Y., Ragettli, M.S., Ineichen, A., Kuenzli, N., Phuleria, H.C.,2013. Role of highway traffic on spatial and temporal distributions of air pol-lutants in a Swiss Alpine valley. Sci. Total Environ. 456, 50e60.

Invernizzi, G., Ruprecht, A., Mazza, R., De Marco, C., Mo�cnik, G., Sioutas, C.,Westerdahl, D., 2011. Measurement of black carbon concentration as an indi-cator of air quality benefits of traffic restriction policies within the ecopass zonein Milan, Italy. Atmos. Environ. 45 (21), 3522e3527.

Ito, K., Xue, N., Thurston, G., 2004. Spatial variation of PM2.5 chemical species andsource-apportioned mass concentrations in New York City. Atmos. Environ. 38(31), 5269e5282.

Kim, J.J., Baik, J.J., 2004. A numerical study of the effects of ambient wind directionon flow and dispersion in urban street canyons using the RNG k-ε turbulencemodel. Atmos. Environ. 38 (19), 3039e3048.

Kim, M.J., Park, R.J., Kim, J.-J., 2012. Urban air quality modeling with fullO3eNOxeVOC chemistry: implications for O3 and PM air quality in a streetcanyon. Atmos. Environ. 47 (0), 330e340.

Kozawa, K.H., Winer, A.M., Fruin, S.A., 2012. Ultrafine particle size distributions nearfreeways: effects of differing wind directions on exposure. Atmos. Environ. 63,250e260.

Kumar, P., Fennell, P., Britter, R., 2008. Measurements of particles in the 5-1000 nmrange close to road level in an urban street canyon. Sci. Total Environ. 390(2e3), 437e447.

Kumar, P., Ketzel, M., Vardoulakis, S., Pirjola, L., Britter, R., 2011. Dynamics anddispersion modelling of nanoparticles from road traffic in the urban atmo-spheric environment e a review. J. Aerosol Sci. 42 (9), 580e603.

Kunzli, N., Jerrett, M., Garcia-Esteban, R., Basagana, X., Beckermann, B., Gilliland, F.,Medina, M., Peters, J., Hodis, H.N., Mack, W.J., 2010. Ambient air pollution andthe progression of atherosclerosis in adults. PLoS One 5 (2).

Li, X.-X., Liu, C.-H., Leung, D.Y.C., 2009. Numerical investigation of pollutant trans-port characteristics inside deep urban street canyons. Atmos. Environ. 43 (15),2410e2418.

Liu, C.-H., Leung, D.Y.C., Barth, M.C., 2005. On the prediction of air and pollutantexchange rates in street canyons of different aspect ratios using large-eddysimulation. Atmos. Environ. 39 (9), 1567e1574.

Ning, Z., Chan, K.L., Wong, K.C., Westerdahl, D., Mocnik, G., Zhou, J.H., Cheung, C.S.,2013. Black carbon mass size distributions of diesel exhaust and urban aerosolsmeasured using differential mobility analyzer in tandem with Aethalometer.Atmos. Environ. 80, 31e40.

Ning, Z., Hudda, N., Daher, N., Kam, W., Herner, J., Kozawa, K., Mara, S., Sioutas, C.,2010. Impact of roadside noise barriers on particle size distributions and pol-lutants concentrations near freeways. Atmos. Environ. 44 (26), 3118e3127.

Ning, Z., Wubulihairen, M., Yang, F.H., 2012. PM, NOx and butane emissions from on-road vehicle fleets in Hong Kong and their implications on emission controlpolicy. Atmos. Environ. 61, 265e274.

Nuvolone, D., Barchielli, A., Forastiere, F., Epiair, G.C., 2009. Assessing the effec-tiveness of local transport policies for improvements in urban air quality andpublic health: a review of scientific literature. Epidemiol. Prev. 33 (3), 79e87.

Oberdorster, G., Oberdorster, E., Oberdorster, J., 2005. Nanotoxicology: an emergingdiscipline evolving from studies of ultrafine particles. Environ. Health Perspect.113 (7), 823e839.

Pant, P., Harrison, R.M., 2013. Estimation of the contribution of road traffic emis-sions to particulate matter concentrations from field measurements: a review.Atmos. Environ. 77 (0), 78e97.

Pirjola, L., Paasonen, P., Pfeiffer, D., Hussein, T., Hameri, K., Koskentalo, T.,Virtanen, A., Ronkko, T., Keskinen, J., Pakkanen, T.A., Hillamo, R.E., 2006.Dispersion of particles and trace gases nearby a city highway: mobile laboratorymeasurements in Finland. Atmos. Environ. 40 (5), 867e879.

Pope, C.A., Burnett, R.T., Thun, M.J., Calle, E.E., Krewski, D., Ito, K., Thurston, G.D.,2002. Lung cancer, cardiopulmonary mortality, and long-term exposure to fineparticulate air pollution. J. Am. Med. Assoc. 287 (9), 1132e1141.

Rodriguez, S., Cuevas, E., 2007. The contributions of “minimum primary emissions”and “new particle formation enhancements” to the particle number concen-tration in urban air. J. Aerosol Sci. 38 (12), 1207e1219.

Rosenlund, M., Forastiere, F., Porta, D., De Sario, M., Badaloni, C., Perucci, C.A., 2009.Traffic-related air pollution in relation to respiratory symptoms, allergicsensitisation and lung function in schoolchildren. Thorax 64 (7), 573e580.

Ruprecht, A.A., Invernizzi, G., 2009. The effects of the traffic restriction scheme(Ecopass) in Milan city center on particulate matter urban pollution: the resultsof a pilot study (L'effetto del provvedimento di restrizione del traffico nel centradi Milano (Ecopass) sull'inquinamento urbano da polveri: I risultati di unostudio pilota), 33 (1e2), 21e26.

Schaller, B., 2010. New York City's congestion pricing experience and implicationsfor road pricing acceptance in the United States. Transp. Policy 17 (4), 266e273.

Steinle, S., Reis, S., Sabel, C.E., 2013. Quantifying human exposure to air pollution-moving from static monitoring to spatio-temporally resolved personal expo-sure assessment. Sci. Total Environ. 443, 184e193.

Van Dingenen, R., Raes, F., Putaud, J.P., Baltensperger, U., Charron, A., Facchini, M.C.,Decesari, S., Fuzzi, S., Gehrig, R., Hansson, H.C., Harrison, R.M., Hüglin, C.,

A. Rakowska et al. / Atmospheric Environment 98 (2014) 260e270270

Jones, A.M., Laj, P., Lorbeer, G., Maenhaut, W., Palmgren, F., Querol, X.,Rodriguez, S., Schneider, J., Ten Brink, H., Tunved, P., Tørseth, K., Wehner, B.,Weingartner, E., Wiedensohler, A., Wåhlin, P., 2004. A European aerosol phe-nomenology e 1: physical characteristics of particulate matter at Kerbside,urban, rural and background sites in Europe. Atmos. Environ. 38 (16),2561e2577.

Wania, A., Bruse, M., Blond, N., Weber, C., 2012. Analysing the influence of differentstreet vegetation on traffic-induced particle dispersion using microscale sim-ulations. J. Environ. Manage. 94 (1), 91e101.

Weber, S., Kuttler, W., Weber, K., 2006. Flow characteristics and particle mass andnumber concentration variability within a busy urban street canyon. Atmos.Environ. 40 (39), 7565e7578.

Westerdahl, D., Fruin, S., Sax, T., Fine, P.M., Sioutas, C., 2005. Mobile platformmeasurements of ultrafine particles and associated pollutant concentrations onfreeways and residential streets in Los Angeles. Atmos. Environ. 39 (20),3597e3610.

WHO, 2012. IARC: Diesel Engines Exhaust Carcinogenic. International Agency forResearch on Cancer, World Health Organization press release N 213, 12 June2012.

Wilson, J.G., Kingham, S., Pearce, J., Sturman, A.P., 2005. A review of intraurbanvariations in particulate air pollution: implications for epidemiologicalresearch. Atmos. Environ. 39 (34), 6444e6462.

Wilson, W.E., Suh, H.H., 1997. Fine particles and coarse particles: concentrationrelationships relevant to epidemiologic studies. J. Air Waste Manage. Assoc. 47(12), 1238e1249.

Wolff, H., Perry, L., 2010. Trends in clean air legislation in Europe: particulate matterand low emission zones. Rev. Environ. Econ. Policy 4 (2), 293e308.

Yang, F., Kaul, D., Wong, K.C., Westerdahl, D., Sun, L., Ho, K.-f., Tian, L.,Brimblecombe, P., Ning, Z., 2014. Heterogeneity of passenger exposure to airpollutants in public transport microenvironments. Atmos. Environ. (submittedfor publication).

Yuan, Z., Yadav, V., Turner, J.R., Louie, P.K.K., Lau, A.K.H., 2013. Long-term trends ofambient particulate matter emission source contributions and the account-ability of control strategies in Hong Kong over 1998e2008. Atmos. Environ. 76(0), 21e31.

Zhang, K.M., Wexler, A.S., 2004. Evolution of particle number distribution nearroadways e part I: analysis of aerosol dynamics and its implications for engineemission measurement. Atmos. Environ. 38 (38), 6643e6653.

Zhou, Y., Wu, Y., Yang, L., Fu, L., He, K., Wang, S., Hao, J., Chen, J., Li, C., 2010. Theimpact of transportation control measures on emission reductions during the2008 Olympic Games in Beijing, China. Atmos. Environ. 44 (3), 285e293.

Zhu, Y.F., Hinds, W.C., Kim, S., Sioutas, C., 2002. Concentration and size distributionof ultrafine particles near a major highway. J. Air Waste Manage. Assoc. 52 (9),1032e1042.

Zwack, L.M., Paciorek, C.J., Spengler, J.D., Levy, J.I., 2011a. Characterizing local trafficcontributions to particulate air pollution in street canyons using mobilemonitoring techniques. Atmos. Environ. 45 (15), 2507e2514.

Zwack, L.M., Paciorek, C.J., Spengler, J.D., Levy, J.I., 2011b. Modeling spatial patternsof traffic-related air pollutants in complex urban terrain. Environ. Health Per-spect. 119 (6), 852e859.