Hotspots of black carbon and PM2.5 in an urban area and ...afrg.peas.dal.ca/publishing/Targino et al 2016_Hotspots of black carbon... · Mobile sampling Particulate matter abstract

lable at ScienceDirect

Environmental Pollution xxx (2016) 1e12

Contents lists avai

Environmental Pollution

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

Hotspots of black carbon and PM2.5 in an urban area and relationshipsto traffic characteristics*

Admir Cr�eso Targino a, *, Mark David Gibson b, Patricia Krecl a,Marcos Vinicius Costa Rodrigues c, Maurício Moreira dos Santos c,Marcelo de Paula Correa d

a Graduate Program in Environmental Engineering, Federal University of Technology, Av. Pioneiros 3131, 86036-370, Londrina, PR, Brazilb Department of Civil and Resource Engineering, Dalhousie University, 1360 Barrington Street, Halifax, Nova Scotia, B3H 4R2, Canadac Department of Environmental Engineering, Federal University of Technology, Av. Pioneiros 3131, 86036-370, Londrina, PR, Brazild Natural Resources Department, Federal University of Itajub�a, Av. BPS 1303, 37500-903, Itajub�a, MG, Brazil

a r t i c l e i n f o

Article history:Received 5 April 2016Received in revised form10 July 2016Accepted 11 July 2016Available online xxx

Keywords:Short-lived climate pollutantsUrban air pollutionMobile samplingParticulate matter

* This paper has been recommended for acceptanc* Corresponding author.

E-mail address: [email protected] (A.C. T

http://dx.doi.org/10.1016/j.envpol.2016.07.0270269-7491/© 2016 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Targinocharacteristics, Environmental Pollution (20

a b s t r a c t

Three instrumented bicycles were used to measure black carbon (BC) and PM2.5 concentrations in amidsized city in southern Brazil. The objective of this study was to map the spatial distribution of BC andPM2.5, to identify air pollution hotspots and to assess factors that may affect the concentrations of thesepollutants, e.g. traffic volume, number of heavy-duty diesel vehicles (HDDV), position of traffic signalsand street incline. The cyclists collected data in the city centre along streets of different traffic densityduring nine sampling sessions in the weekday morning and afternoon rush hours, between March 13 andApril 28, 2015. The sampling by bicycle covered an area of 2.70 km2, over variable elevation, and travelleda total distance of 215 km. BC and PM2.5 exhibited a large spatial variability on a scale of tens of metresand the concentrations were positively correlated with traffic counts, but exhibited a stronger rela-tionship with the number of HDDV. These results imply that older buses and diesel-powered trucks maybe the main driver behind the high pollution levels in the city's inner core. We observed a strongrelationship between BC concentrations at junctions managed by traffic signals and the quantity ofHDDV. The mean BC concentration was found to be 8.10 mg m�3 near traffic signals located on an inclinedstreet (HDDV > 100 vehicles h�1) compared to traffic signals on flat terrain (6.00 mg m�3), which can beattributed to the higher acceleration required at the start of motion. This pattern was less evident forPM2.5 concentrations.

© 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Emissions from motor vehicles are a major source of gaseousand airborne particulate matter (PM) air pollution in urban envi-ronments (Monks et al., 2009). The mass of particulate matter thatpasses through a size-selective inlet, with a 50% efficiency cut-off ata 2.5-mm aerodynamic diameter is referred to as PM2.5. PM2.5 is oneof the most widely monitored and reported metrics of urban airpollution due to its association with negative health outcomes(Stieb et al., 2008; Pope et al., 2002). In terms of number concen-tration, a large percentage of particles from gasoline and diesel

e by David Carpenter.

argino).

, A.C., et al., Hotspots of bla16), http://dx.doi.org/10.1016

combustion are found in the ultrafine particle size distribution(with median aerodynamic diameters below 100 nm) (Kumar et al.,2014; Gouriou et al., 2004). For example, Krecl et al. (2015) andKetzel et al. (2004) observed peaks of particle number concentra-tions between 28 and 56 nm and between 20 and 30 nm at kerbsidesites in Stockholm and Copenhagen, respectively. Nevertheless,collectively, the exhaust ultrafine particles are too small to accu-mulate aerosol mass, hence other contributions to PM2.5 in traf-ficked locations are non-exhaust particles from re-suspendedsurficial dusts, tyre and brake wear (Gibson et al., 2015; Mathissenet al., 2011; Wahlstr€om et al., 2010; Dahl et al., 2006) and growth ofultrafine particles by condensation of low volatility vapours (Ahlmet al., 2012; Woo et al., 2001). Despite the small mass of exhaustaerosols, the toxicity of specific components in PM2.5, such as blackcarbon (BC), makes BC a better indicator of the negative health

ck carbon and PM2.5 in an urban area and relationships to traffic/j.envpol.2016.07.027

mailto:[email protected]

www.sciencedirect.com/science/journal/02697491

http://www.elsevier.com/locate/envpol

http://dx.doi.org/10.1016/j.envpol.2016.07.027



A.C. Targino et al. / Environmental Pollution xxx (2016) 1e122

effects of airborne particles than unspecified PM (Janssen et al.,2012). BC is an air pollutant emitted by the incomplete combus-tion of fossil fuels, biomass or any carbon-containing matter. In theearly 1980s, theWorld Health Organization recognised the negativeeffects of BC particles on human health and formulated the firstguidelines on exposure thresholds for BC and, recently, new sci-entific evidence (e.g., Attfield et al., 2012) classified BC from dieselfumes as carcinogenic.

The concentrations of traffic-related PM in cities is highly vari-able in space and time due to the great heterogeneity of traffic rates,driving modes and emission factors. Krecl et al. (2014) found me-dian BC concentrations between 7.50 mg m�3 (in road tunnels) and3.20 mg m�3 (on highways) in Stockholm, whilst Hankey andMarshall (2015) found that concentrations of particle number andBC were correlated with street type and declined approximately20% by moving within small distances from a major road to adja-cent local roads in Minneapolis. Furthermore, the city's geograph-ical features impact local airflows and, ultimately, theconcentration of pollutants within the urban canopy layer (Kreclet al., 2015; Boddy et al., 2005).

Air pollution monitoring is mandated in developed and in(some) developing countries, however the infrastructure andtechnical capabilities available determine whether strategies toalleviate air quality problems and safeguard the health of thepopulation can be put into place. For example, in the EuropeanUnion and North America long-term air pollution data have beenused for air quality management and to streamline strategies (e.g.,installation of low emission zones, capping the age of the publicvehicle fleet) to reduce air pollution concentrations in target areasleading to successful control policies. Alternatively, air pollutiondata in Asian and Latin American countries are scattered and ofvariable quality, which hampers the implementation of regulatorystrategies to abate air pollution concentrations and protect publichealth (Hsu et al., 2013). These are particularly worrying issues inLatin America and the Caribbean where about 80% of the popula-tion lives in cities, making it the most urbanised region in thedeveloping world (United Nations, 2014). In Brazil, most urban airpollution data originate from monitoring networks in large citiessuch as S~ao Paulo, Rio de Janeiro and Curitiba, or from a few short-term focused studies (e.g., Krecl et al., 2016; Targino and Krecl,2016; Miranda et al., 2010). Therefore, there is a paucity ofcontinuous air pollution monitoring in midsized Brazilian citieswith populations ranging from 100,000 to 600,000, such asinvestigated in this study.

Assessing the within-city spatial variability of traffic air pollut-ants with mobile sampling platforms provides key information toestimate personal exposure and to identify heavily polluted areas.Mobile sampling techniques have been widely used over recentyears due to the availability of portable and scientifically robustinstrumentation that can be deployed inside (Westerdahl et al.,2005; Weijers et al., 2004) and outside (Krecl et al., 2014) motorvehicles, on trams (Hagemann et al., 2014), on bicycles (Hankey andMarshall, 2015; Pattinson et al., 2014) and carried by pedestrians(Dons et al., 2012). This is a vast improvement on exposure esti-mates that traditionally only usemeasurements from a few fixed airmonitoring stations, andwhich have been shown to not capture thefine spatio-temporal patterns required to pinpoint hotspots andperiods of the day that are detrimental to human health (Wheeleret al., 2008). Moreover, portable instruments can reach specificareas of the city and target microenvironments for a relatively fastscreening of air quality. This approach is very powerfulealbeit not asubstitute for regular monitoring stationse to gather intensive,short-term data in areas where air pollution measurements areuneven or non-existent. More specifically, they have proven veryattractive for studies of PM2.5, since this pollutant has a relatively

Please cite this article in press as: Targino, A.C., et al., Hotspots of blacharacteristics, Environmental Pollution (2016), http://dx.doi.org/10.1016

short lifespan, exhibits highly variable spatio-temporal patterns,and is subject to growth and/or removal processes in the atmo-sphere (e.g., Franklin et al., 2014; Zhu et al., 2002).

The objective of this study was to investigate the spatial distri-bution of PM2.5 and BC aerosols in the urban core of a midsizedBrazilian city using instrumental packs mounted on bicycles. Westreamlined our measurement campaign to identify air pollutionhotspots and to assess drivers that affect the within-city variabilityof the aerosol concentrations, such as traffic density, vehicle typeand location of traffic signals.

2. Experimental methods

2.1. Study area

The experiment was conducted in the centre of Londrina ea citywith 548,000 inhabitants located in southern Brazil (lat. 23�190S,long. 51�080W, alt. 630 m)e on selected days in March and April2015. The city's motorised fleet consists of approximately 362,000vehicles, with 52% vehicles that run on gasohol (gasoline blendedwith 27% (v/v) anhydrous ethanol), 33% flex engines that run ongasohol, alcohol or a mixture of these fuels, and 7% on diesel(mostly buses and trucks). In Brazil, diesel for on-road use contains6% biodiesel, and both low sulphur diesel S10 (10 mg kg�1 ofsulphur, only for vehicles manufactured after 2012) and S500(500 mg kg�1 of sulphur) are available. The city's fleet grew 85% inthe period 2005e2014 and consists mainly of light-duty passengervehicles (78%), whilst the number of motorcycles increased 68% inthe same period, as a convenient and affordable mode of transport(Targino and Krecl, 2016).

We simultaneously measured PM2.5 and BC concentrationsalong roads in the city centre using three bicycles as mobile sam-pling platforms.We sampled three designated routes with differentroad configurations and traffic densities, covering an area of2.70 km2 in a gridded pattern and over variable terrain. Route Aconsisted of a loop 5.4-km long crossing the central and northernparts of the city centre. This route forms the northern boundary ofthe surveyed perimeter along Arcebispo Dom Geraldo FernandesAve. (known as Leste-Oeste Ave.), a two-way, six-lane avenuealigned in the east-west direction with high traffic volume andintense commercial activity. Route B made a smooth loop of 7.0 kmon relatively flat terrain at the southern side of the city. The mainroad on this route is Juscelino Kubitschek (JK) Ave. that runs in theeast-west direction then turns to run in a north-south direction.This is a two-way, four-lane avenue that sets the western andsouthern borders of the city centre. Route C crossed the city's innercore on a 9.0-km loop and consists of streets with steep gradientsand canyons formed by tall buildings flanking either side of theroad. In an effort to avoid potential oversampling, the routes weredesigned to avoid overlapping and backtracking. Fig. 1 provides amap of the sampling area in Londrina.

We use the term run to denote a sampling session covering anentire route. The data collection consisted of nine runs onweekdaysbetween March 13 and April 28, 2015 totalling 215 km. The runsalways started simultaneously and at the same time each day (8a.m. and 5 p.m.) to avoid hourly variability in traffic patterns andalso to cover rush hour periods. The runs lasted between 50 and70 min depending on the route.

2.2. Instrument package

The bicycles were retrofitted with an L-shaped carbon steelplatform bolted to the stem of the handlebars, on which a heavy-duty tool bag was placed. The bag was secured to the platformwith brackets attached to the handlebars, and bungee cords for


Fig. 1. Map of the study area in Londrina city centre with the three sampling routes. Also shown: traffic counting points (yellow circles), bus stops (red triangles) and traffic lights(black circles). The arrows indicate the traffic direction. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

A.C. Targino et al. / Environmental Pollution xxx (2016) 1e12 3

extra stability (Fig. S1, SM).Each bicycle was equipped with a microaethalometer (AE51,

Aethlabs, USA) for BC measurements, a photometer (DustTrak8520, TSI, USA) for PM2.5 measurements and a high accuracy GPSreceiver and logger (DG-100, GlobalSat, Taiwan). The instrumentswere placed inside the heavy-duty bag and cushioned with foam toreduce vibration. Each aerosol instrument had its own samplinginlet made of conductive tubing placed approximately 1.15m aboveground level (Fig. S2, SM). Since our objective was to sample smallparticles, the position of the inlet ports did not need to fulfill therequirements for isokinetic sampling, which is only important forlarge particles. These instruments have been widely used by the


research community and a thorough description of their charac-teristics and principles of operation can be found elsewhere (e.g.,Hansen et al., 1984 and Chung et al., 2001, for the aethalometer andthe DustTrak 8520, respectively). In short, the microaethalometerdetermines the BC concentration by measuring the attenuation ofan 880-nm radiation beam transmitted through a filter strip madeof T60 teflon coated borosilicate glass fibre on which ambientaerosol particles are continuously deposited. The particles on thefilter spot attenuate the impinging radiation beam and the atten-uation over a sampling period is compared to an adjacent unex-posed (reference) portion of the filter. The BC concentration iscalculated by relating the change in attenuation with the



wavelength-dependent cross sectional absorption coefficient. Themethod assumes that carbonaceous particles are the only light-absorbing material on the sample, however interferences canoccur in environments dominated by mineral dust since theseparticles are the most relevant light-absorbing, non-carbonaceousparticles (Moosmüller et al., 2009). The AE51 instruments wereoperated at a flow rate of 100 mLmin�1, temporal resolution of 10 sand with no size-selective inlet. Hence, we collocated the AE51swith an Aethalometer AE42 (Magee Scientific, USA) fitted with a2.5-mm cyclone and compared their performance in ambient airduring three days. The one-hour average data were found to behighly correlated (R2¼ 0.896), which indicates that most BCmass isin the fine mode and the use of a 2.5-mm cyclone has little effect onthe comparison of the datasets reported in this study.

The DustTrak 8520 is a real-time, portable monitor formeasuring PM concentrations. The inlet flow (1.7 L min�1) is con-strained by a critical orifice followed by a greased PM2.5 impactor.The particles in the sample flow penetrate a chamber illuminatedby a 780-nm laser, and a photodetector detects the radiation scat-tered by the particles. This method determines the particle volumeby assuming the diameter of a spherical particle based on thescattered light's intensity, and subsequently derives the mass usingthe density of the respirable fraction of the Arizona Test Dustaerosol (r ¼ of 2.6 g cm�3). The DustTrak temporal resolution wasset to 10 s. The GPS logged latitude, longitude, altitude and speedwith a temporal resolution of 1 s.

With an instrument temporal resolution of 10 s and a cyclingspeed between 9.0 and 11.0 km h�1, each run yielded at least threedata points per city block. The approximate length of a typical cityblock in Londrina is 100 m.

2.3. Instrumentation preparation and data validation

Instrument preparation took place in our laboratory before eachrun and consisted of zero calibration checks, cleaning and greasingof the DustTrak impactor, cleaning of the DustTrak chamber,changing of the AE51 filter strip (the maximum attenuation was42%), battery and memory checks, and clock synchronisation usinga laptop connected to the internet to ensure clock accuracy. Flowmeasurements were performed with a factory calibrated flowmeter (model 4100, TSI, USA). Collocation periods occurred beforeand after each measurement session to check that the instrumentswere still performing as expected and also to assess precision. Inaddition, we also left the bicycles standing side by side forapproximately 20 min at the end of each run for further instrumentintercomparison.

2.4. Ancillary data

2.4.1. Traffic dataThe traffic characteristics during each run were analysed with

aid of voice records from the cyclists. We also conducted manualtraffic counting at 35 key points along the routes during the hourscoincident with the mobile data collection. For each hour, wecounted the number of vehicles during two 15-min intervals, withalternating 15-min pauses, divided into four categories: cars, mo-torcycles, trucks and buses. Thenwemultiplied the number in eachinterval and category by two to obtain the total volume for half anhour. On two-way roads, two observers counted the traffic flowseparately for each direction.

All traffic signals along the routes were identified and geore-ferenced to allow for analysis of their effects on particle massconcentrations, as suggested by Goel and Kumar (2014).

To analyse the relationship between traffic rate and BC andPM2.5 concentrations, we pooled pollutant concentrations in 20 m


buffers around each traffic counting point and further classified thevehicular hourly number density into light-duty vehicles (LDV) andheavy-duty diesel vehicles (HDDV).

2.4.2. Topographic dataThe topographic data of the study area was georeferenced in the

SIRGAS 2000 (Geocentric Reference System for the Americas)spatial reference system, and the cartographic projection presentedas UTM 22S (Universal Transverse Mercator, Zone 22 South). Theelevation data were obtained from the Advanced SpaceborneThermal Emission and Reflection Radiometer (ASTER) operating onNASA's Earth Observing Terra morning orbital platform. Moreprecisely, we used the Global Digital Elevation Model (GDEM) in1� � 1� tiles at 1 arc second (about 30 m) resolution (Abrams et al.,2015), and the geoprocessing software subsequently calculatedstreet slopes (%) from the elevation field.

2.5. Fire spots and backtrajectories

We combined the location of fire spots in South America (INPE,2015) with 120 h air mass backtrajectories calculated with theHYSPLIT (Hybrid Single-Particle Lagrangian Integrated Trajectory)model (Rolph, 2016) arriving in Londrina at 100 m above groundlevel during the campaign. The model was run using the reanalysis(global, 1948epresent) database.

2.6. Data analysis and processing protocols

The instruments' clocks were synchronised immediately beforeeach run, hence we assumed that the synchronisation was keptaccurate during the relatively short duration of the data acquisitionsessions. Since the GPS logged data at a rate of 1 s and the micro-aethalometer and DustTrak at a rate of 10 s, we ensured that all BCand PM2.5 measurements had coincident geo-location data andcould, thus, be geo-referenced. For each run, we used a computingprogram to scan the GPS, AE51 and DustTrak data files to findcoincident time stamps. The geolocation and air pollutant datawere subsequently pooled together into their respective times,yielding georeferenced data matrices.

We defined 65 m polygons along the three routes (which cor-responds to the length of a half block and a half intersection) and alldata points that fell into each individual polygon were used tocalculate an aggregated median pollutant concentration. Aggre-gating the data to a fixed length interval is recommended tominimise the effects of extreme concentrations caused by sporadicvehicle exhaust emissions (Brantley et al., 2015). We calculatedmedian concentrations because they are more representativemeasures of central tendency for non-normal distributed envi-ronmental data such as aerosol measurements.

We did not apply any post-processing algorithm to the BC datato reduce instrumental or electrical noises, since the 5th percentilevalues were higher than the detection limit of the AE51 monitors(between 0.20 and 0.25 mg m�3).

The mass concentrations reported by the DustTrak are largerthan the real mass concentrations of the urban aerosol (Tasic et al.,2012; Wallace et al., 2011 and references therein) because urbanaerosols have lower density than the Arizona Test Dust aerosols; forexample, r ¼ 1.36e1.55 g cm�3 in Shanghai (Yin et al., 2015) andr ¼ 1.60 g cm�3 in Erfurt (Pitz et al., 2003). Usually concurrentgravimetric measurements correlate the two sampling methodsand yield correction factors for the DustTrak, which have beenfound to range from 1.70 (McNamara et al., 2011) to 2.78 (Wallaceet al., 2011) in BC dominated environments. We did not havegravimetric measurements during the campaign and hence couldnot extract specific correction factors for our measurements,



however we applied a factor of 2 to the PM2.5 data. This value waschosen to ensure that the BC/PM2.5 ratio is kept within limitstypically found in literature, as will be discussed further in theResults section.

3. Results

3.1. General weather conditions

All sampling sessions were conducted on non-rainy days toavoid misrepresentation of typical urban air pollution conditionsdue to aerosol washout and to protect the instrument electronicsfrom damage. Table 1 summarises the average meteorologicalconditions in the hours around the sampling session, including data1 h immediately before and 1 h after the measurements.

The combined fire spot and backtrajectory plots show that thesampling days were not affected by long-range transport ofbiomass burning (Fig. S3, SM). Hence, air pollution in the surveyedarea is mostly due to local on-road traffic.

3.2. Traffic patterns

Fig. 2 shows the weekday traffic density and fleet compositionfor mornings and afternoons at the 35 counting points along thethree routes. The car is the dominant vehicle type (average of 81% inthe morning and 82% in the afternoon), followed by motorcycles,buses and trucks (Table 2). The traffic density and number of ve-hicles are comparable for each monitoring point and for morningand afternoon periods. Whilst this quantity remained similar indifferent sections of the same road, the total traffic rate along somebusy avenues varied due to diverging and merging traffic. The totalvalues for each traffic count point and vehicle category can be foundin Tables S1 and S2.

We observed the largest traffic density on the outer border ofroute A, along Leste-Oeste Ave., ranging from 2166 to 2686 vehiclesh�1 in the morning, and 2714 to 3074 vehicles h�1 in the afternoon(points 1A, 2A and 3A of Fig. 2). The traffic density on the southernand western borders of route B along JK Ave. ranged from 1960 to2934 vehicles h�1 in the morning and 2644 to 2870 vehicles h�1 inthe afternoon (points 2B, 3B, 4B and 5B). High traffic numbers fromsmall roads can be observed entering these avenues. The traffic ratein the city centre core was considerably lower when compared tothese roads, except at some points along Higienopolis Ave. (Points6C and 7C) that on average reached 1240 vehicles h�1 in themorning and 1670 vehicles h�1 in the afternoon. Higienopolis Ave.is a 2.8-km arterial road that runs in the north-south direction andcrosses commercial and residential areas.

3.3. Spatial distribution of BC and PM2.5 concentrations

The mobile platforms collected data in different microenviron-ments representative of roads with a variety of traffic densities,

Table 1Summary of the meteorological conditions during the sampling sessions.

Date [2015] Air temperature[�C] Air relative hu

March 13 morning 22.7 80.4March 20 morning 21.9 87.2March 23 morning 22.0 77.3March 24 afternoon 27.5 49.4March 31 afternoon 27.5 47.6April 7 afternoon 23.1 64.7April 10 morning 20.5 78.5April 22 morning 20.7 94.8April 28 afternoon 26.0 43.5


topographies and urban attributes and, as a result, BC and PM2.5concentrations varied greatly. The large spatial variability in BC andPM2.5 is summarised in Fig. 3, which shows the median pollutantconcentrations for all morning and afternoon data. Table 3 shows abreakdown of the main statistical indicators using 10 s values of BCand PM2.5 for the whole dataset. The mean and standard deviationsfor morning and afternoon BC concentrations are6.35 ± 20.00 mg m�3 and 5.10 ± 14.75 mg m�3, and for PM2.58.61 ± 19.39 mg m�3 and 8.43 ± 14.43 mg m�3 respectively (Table 3).The high standard deviation values for both pollutants are indica-tive of large variability in the BC and PM2.5 datasets. In general, thisvariability can be explained via three pathways: changes in trafficemissions, particle dynamics and wind conditions. The variabilityin traffic emissions might be mainly linked to changing drivingconditions (free flow, stop-and-go and saturated flow), trafficdensity and vehicle share. The traffic density was on average 26.5%higher in the afternoon than in themorning at most of the countingpoints (31 out of 35, Fig. 2). However, the share of HDDV decreasedon average from 4.3 to 2.9%, explaining the lower median BC valuein the afternoon (Fig. 3e). On the other hand, the larger number oftotal vehicles in the afternoon could have contributed to the in-crease in PM2.5 due to non-exhaust particles and secondary particleformation and growth (Ahlm et al., 2012; Woo et al., 2001). Themean wind speed was 2.2 m s�1 in the morning sampling sessionsand 3.3 m s�1 in the afternoon (these values can be even lower dueto friction with the urban canopy). This is a small difference, whichcannot account for discrepancies in the data variability.

The response of the DustTrak can be affected by the air relativityhumidity (RH) due to the increase of particle volume associatedwith water uptake. In our study, the RH of the morning measure-ment sessions was between 77 and 95%, while the afternoon ses-sions had RH values between 43 and 65%. Nevertheless, the medianPM2.5 concentration was higher in the afternoon than in themorning. Correcting the scattering enhancement effect requiresknowledge of the physico-chemical properties of the aerosols(Zieger et al., 2013).

The composite datasets show that the distributions of BC andPM2.5 measurements are skewed and better represented bylognormal distributions. The BC distributions, particularly, agreewith results from mobile campaigns conducted elsewhere (e.g.,Krecl et al., 2014).

The air pollution hotspots are easily discernible by inspectingthe red rectangles (BC) and dark blue circles (PM2.5) in Figs. 4 and 5.Overall, the concentrations are largest in the outer perimeter of thesurveyed area (along Leste-Oeste, JK and Duque de Caxias avenues),with values exceeding 8.00 mgm�3, and also along specific transectswithin the city centre. Another salient feature was the elevated BC(6.53 and 21.50 mg m�3, morning and afternoon, respectively) andPM2.5 (22.94 and 25.75 mg m�3) concentrations observed close tothe city's bus station. This station serves as a commuter terminaland on a typical weekday about 945 diesel buses serve approxi-mately 100,000 passengers. The discrepancies in morning and

midity [%] Wind speed [m s�1] Wind direction [�]

0.9 218.21.2 152.43.0 96.53.6 114.03.9 226.24.7 99.72.4 149.73.5 72.41.0 178.7


Fig. 2. Traffic volume (vehicles per hour) and fleet composition for morning and afternoon counts. The blue bar represents 1000 vehicles per hour. (For interpretation of thereferences to colour in this figure legend, the reader is referred to the web version of this article.)

Table 2Summary of the fleet composition in the 35 surveyed points of Londrina city centre.The values in brackets are the average of all observations.

Morning Afternoon

Cars 70e92% [81%] 72e92% [82%]Motorcycles 6e25% [15%] 4e24% [15%]Buses 0e8% [2%] 0-6% [2%]Trucks 0e9% [2%] 0-2% [1%]



afternoon BC concentrations may be explained by the larger hourlydensity of HDDV in the afternoon at traffic count points close to thestation (2A and 3A, SM). The difference in the PM2.5 concentrationsis not as large, which may be attributed to the presence of othernon-absorbing aerosols, such as suspended tyre and brake wear,and ultrafine particles from secondary production, which enhancesPM2.5 but not BC. The pollutant concentrations at the corner ofBenjamin Constant and Minas Gerais streets (indicated by a trafficlight icon in Figs. 4 and 5) stand out. The traffic at this junction ismanaged by a traffic signal, which diverges vehicles either forwardor to the right ontoMinas Gerais Street, which has an upslope grade


Fig. 3. Left column: statistical summary of BC and PM2.5 concentrations for morning and afternoon 10-s data. The box plots show the median (solid line), the interquartile range(box), and the 5th and 95th percentile ranges of the data (whiskers). Right column: frequency of occurrence of 10-s BC and PM2.5 measurements. The normal density function is alsodisplayed in the composite plot (black line).

Table 3Descriptive statistics of BC and PM2.5.

Morning Afternoon

PM2.5

[mg m�3]BC[mg m�3]

PM2.5

[mg m�3]BC[mg m�3]

Arithmetic mean 8.61 6.35 8.43 5.10Standard deviation 19.39 20.00 14.43 14.755th percentile 1.50 0.68 3.00 0.6025th percentile 4.00 1.48 4.50 1.30Median 5.50 2.42 6.00 2.1775th percentile 8.00 5.13 8.50 3.9795th percentile 20.50 20.40 19.50 16.30No. of samples 4610 5060 3107 3957


between 7 and 13%. The influence of the street slope on BC andPM2.5 concentrations is also observed on Pernambuco street(indicated by a car-on-street icon in Figs. 4 and 5). The maximumBC and PM2.5 concentrations in this transect are 7.05 and7.50 mg m�3, and 4.80 and 16.50 mg m�3, in the morning and af-ternoon, respectively. A general overview in 3-D of the spatialdistribution of BC and PM2.5 for afternoon samplings can be seen inVideo clips 1 and 2, respectively. Consistent with other studies (e.g.,Wu et al., 2015; Krecl et al., 2014), our data indicate that the spatialvariations of BC and PM2.5 are related to traffic volume and urbanattributes, such as traffic signals at junctions. These aspects are


discussed in detail further on in the next sections.Supplementary video related to this article can be found at

http://dx.doi.org/10.1016/j.envpol.2016.07.027.

3.4. Relationships between air pollutant concentrations and trafficcharacteristics

3.4.1. Traffic densityBC and PM2.5 concentrations on the southern, western and

eastern borders of the surveyed area stand out. The western andsouthern hotspots (JK Ave.) coincide with points of high trafficcounts (shown in Fig. 2), while the eastern border have relativelylower (but not depreciable) traffic densities (1360 vehicles h�1 atpoint 4A and 1100 vehicles h�1 at point 1B).

The vehicular fleet in Londrina and our vehicle counts aredominated by LDV whereas the HDDV share ranges from null atpoints 6B, 3C and 4C to 10.3% at point 8C (close to the bus station).Fig. 6 exhibits themedian BC and PM2.5 concentrations as a functionof HDDV and LDV rates, inwhich it is possible to observe a graduallyincreasing trend of BC (from 2.40 to 3.70 mg m�3) and PM2.5 (from4.30 to 7.00 mg m�3) as the LDV rate increases from <250 to >1500vehicles h�1. The positive trend in the pollutant concentrations ismore evident as HDDV rates increase, with BC concentrationsranging from 1.81 to 6.14 mg m�3, and PM2.5 ranging from 5.00 to9.25 mg m�3. Krecl et al. (2014) also found a similar relationship



Fig. 4. Spatial distribution of aggregated median BC and PM2.5 concentrations for morning sampling sessions.


between HDDV and BC concentrations in Stockholm, and theseresults are supported by studies on emission factors by Je�zek et al.(2015) who concluded that 25% of diesel-fuelled vehicles accountedfor 63% of BC emissions in a study in Slovenia.

The strong link between HDDV rates and BC reinforces the factthat despite contributing a very small proportion of the vehicle fleet(4.3%), HDDV contribute significantly to the ambient concentra-tions of PM2.5 and especially BC. Another important aspect is that27% of the buses and 56% of the trucks registered in Londrina weremanufactured before 1999, with PM emission factors of 1.071 and0.444 g km�1, respectively. New stringent exhaust emission


standards were implemented in Brazil after 1999 and PM emissionfactors for vehicles manufactured from 2012 onwards are 0.021(buses) and 0.009 g km�1 (trucks). Nevertheless, these rather newvehicles only account for 17% and 8% of the bus and truck fleets,respectively. Furthermore, there is no vehicle inspection andmaintenance program implemented in Paran�a state, which meansthat a large proportion of the trucks in the fleet are high PMemitters. Fig. 6 also shows that the BC/PM2.5 ratio lies between 0.36and 0.56, which is comparable to values found in other urbanstudies. For example, Begum et al. (2012) found BC/PM2.5 ratiosbetween 0.16 and 0.57 at an urban site in Dhaka (Bangladesh), and


Fig. 5. Same as Fig. 4, except for afternoon.


Santoso et al. (2013) found ratios between 0.17 and 0.45 in Indo-nesian cities. We found a rather large ratio (0.77) for HDDV trafficrates higher than 100 vehicles h�1, which may be attributed to: i)BC measurements close to vehicle exhaust, as conducted in thisstudy, yield incidental contributions to PM2.5 that exceed typicalvalues found in diluted air in urban environments; ii) the DustTraktends to underestimate measurements of fossil fuel and biomassburning because light scattering by BC particles is not as efficient.

3.5. Pollutant concentrations at traffic lights

We identified 95 traffic signals in the surveyed area, 68 of whichwere on flat and 27 on inclined streets. The BC and PM2.5 data wereclustered into 20 m buffer zones around each traffic signal,


thereafter aggregated median concentrations for each point werecalculated. This approach is based on Jayaratne et al. (2009) whoreported an increase in particle number concentrations at asignalized pedestrian crossing, and larger enhancement by over anorder of magnitude when diesel-powered buses are included in theapportionment of vehicle number. They found elevated particlenumber emissions not only at, but also after intersections equippedwith traffic signals, attributed to higher acceleration required at thestart of motion when the signal turns green.

As shown previously, the number of HDDV vehicles has a largeinfluence on particle concentrations, hence we further classifiedthe traffic signals according to the hourly HDDV rates. Overall, thereis an increasing trend of BC as the hourly HDDV rate increases(Fig. 7). However, up to a rate of 40 vehicle h�1, there is virtually no


Fig. 6. Median BC and PM2.5 concentrations as a function of hourly traffic rates. a) light-duty vehicles, b) heavy-duty diesel vehicles. The BC-to-PM2.5 ratios for each category are alsoshown.

Fig. 7. Median (a) BC and (b) PM2.5 concentrations calculated from data measured at traffic lights located on inclined and flat terrains.


difference in the BC concentrations for traffic signals located oninclined or on flat terrain. Thereafter, the concentrations increasemore rapidly for data on inclined junctions. More tellingly, the 20mbuffer around a traffic signal at an inclined street junction with anhourly HDDV rate of 80e100 vehicle h�1 has twice the BC con-centration as observed at a traffic signal located at a flat junction,indicating that BC concentrations respond to both HDDV rate andwhether the traffic signal is on flat or inclined terrain. When carsstop at a red light, engines idle and then accelerate when the lightchanges to green. This results in the emission of more particlesfound within the ultrafine size fraction. Goel and Kumar (2014)showed that particle number concentration may be approxi-mately 17-fold larger when a vehicle stops at a red light comparedto average typical roadside concentrations in European cities. Theincreasing trend in PM2.5 concentrations with increasing HDDV rateis also clear, however the concentrations exhibited a differentpattern when compared with the BC data. For a given HDDV rate,the difference in the median concentrations of the two PM2.5

datasets is small (the ratio between the data on inclined and flatjunctions ranges from 0.91 to 1.17), indicating a stronger depen-dence of PM2.5 concentrations on the HDDV rate than on thelocation of the traffic signal. Besides exhaust, PM2.5 sources intraffic-dominated urban environments include suspended tyre andbrake wear and ultrafine particles from secondary production andgrowth of existing aerosols, so even if a vehicle emits more BC on an


inclined junction as it accelerates, this will not necessarily reflectthe observed PM2.5 concentrations.

4. Conclusions

For the first time, the spatio-temporal concentration of BC andPM2.5 was mapped in a Brazilian city using bicycles as mobileplatforms. The study demonstrated the utility of the integration ofrobust air pollution instrumentation with inexpensive bicycles tohighlight the large spatial variability of BC and PM2.5 concentra-tions, which would not otherwise be possible with fixed moni-toring instrumentation. The main drivers of the pollutantconcentrations were traffic density (more specifically the rate ofHDDV), the location of traffic signals and street gradient. Weobserved a positive trend in BC and PM2.5 with increasing vehicledensity, however the trend was more evident when we correlatedthe pollutant concentrations and HDDV density, with BC concen-trations rising from 1.81 to 6.14 mg m�3, and PM2.5 from 5.00 to9.25 mg m�3, as HDDV rates increased from <20 to >100 vehiclesh�1. BC concentrations at high traffic density junctions, and alsoassociated with an inclined street, were up to two-fold larger thanat a junction on a flat area, indicating a relationship between HDDVdensity, traffic signals and terrain. PM2.5 concentrations showed aweaker dependence on the location of the traffic signals, but moreof a relationship with vehicle density. Since the main sources of



PM2.5 in traffic-dominated environments are non-exhaust surficialparticles and the growth of ultrafine particles, an enhancement ofBC will not necessarily be reflected in the PM2.5 concentrations.These results show that BC is a more suitable metric than PM2.5 toassess the impact of exhaust emissions from road traffic on airquality and health in a medium sized Brazilian city.

Acknowledgments

The research for this article was supported by grants 404146/2013-9 and 400273/2014-4 from the National Council for Scientificand Technological Development of Brazil (CNPq). The authorsthank LuizWiese and Jo~ao Paulo Ribeiro for their help with the datacollection, Bruno Lo Frano Machado for conducting manual trafficcounting, Dr. Jorge Alberto Martins for furnishing us with themeteorological data and Ellen Patrick for valuable comments on thetext. The Monitoring of Vegetation Fires Group at the NationalInstitute for Space Research of Brazil is acknowledged for supplyingfire spot data.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.envpol.2016.07.027.

References

Abrams, M., Tsu, H., Hulley, G., Iwao, K., Pieri, D., Cudahy, T., Karge, J., 2015. Theadvanced Spaceborne thermal emission and reflection radiometer (ASTER) afterfifteen years: review of global products. Int. J. Appl. Earth Obs. Geoinf. 38,292e301.

Ahlm, L., Liu, S., Day, D.A., Russell, L.M., Weber, R., Gentner, D.R., Goldstein, A.H.,DiGangi, J.P., Henry, S.B., Keutsch, F.N., VandenBoer, T.C., Markovic, M.Z.,Murphy, J.G., Ren, X., Scheller, S., 2012. Formation and growth of ultrafineparticles from secondary sources in Bakersfield. Calif. J. Geophys. Res. 117 http://dx.doi.org/10.1029/2011JD017144. D00V08.

Attfield, M.D., Schleiff, P.L., Lubin, J.H., Blair, A., Stewart, P.A., co-authors, 2012. Thediesel exhaust in miners study: a cohort mortality study with emphasis on lungcancer. J. Natl. Cancer. Inst. 104, 869e883.

Brantley, H.L., Hagler, G.S.W., Kimbrough, E.S., Williams, R.W., Mukerjee, S.,Neas, L.M., 2015. Mobile air monitoring data-processing strategies and effectson spatial air pollution trends. Atmos. Meas. Tech. 7, 2169e2183.

Boddy, J.W.D., Smalley, R.J., Dixon, N.S., Tate, J.E., Tomin, A.S., 2005. The spatialvariability in concentrations of a traffic-related pollutant in two street canyonsin York, UK e Part I: the influence of background winds. Atmos. Environ. 39,3147e3161.

Begum, B.A., Hossain, A., Nahar, N., Markwitz, A., Hopke, P.K., 2012. Organic andblack carbon in PM2.5 at an urban site at Dhaka. Bangladesh. Aerosol Air. Qual.Res. 12, 1062e1072.

Chung, A., Chang, D.P.Y., Kleeman, M.J., Perry, K.D., Cahill, T.A., Dutcher, D.,McDougall, E.M., Stroud, K., 2001. Comparison of real-time instruments used tomonitor airborne particulate matter. J. Air & Waste Manag. Assoc. 51, 109e120.

Dahl, A., Gharibi, A., Swietlicki, E., Gudmundsson, A., Bohgard, M., Ljungman, A.,Blomqvist, G., Gustafsson, M., 2006. Traffic-generated emissions of ultrafineparticles from pavementetire interface. Atmos. Environ. 40, 1314e1323.

Dons, E., Panis, L.I., van Poppel, M., Wets, G., 2012. Personal exposure to blackcarbon in transport microenvironments. Atmos. Environ. 55, 392e398.

Franklin, J.E., Drummond, J.R., Griffin, D., Pierce, J.R., Waugh, D.L., Palmer, P.I.,Parrington, M.P., Lee, J.D., Lewis, A.C., Rickard, A.R., Taylor, J.W., Allan, J.D.,Coe, H., Chisholm, L., Duck, T.J., Hopper, J.T., Gibson, M.D., Curry, K.R.,Sakamoto, K.M., Lesins, G., Walker, K.A., Dan, L., Kliever, J., Saha, A., 2014. A casestudy of aerosol depletion in a biomass burning plume over Eastern Canadaduring the 2011 BORTAS field experiment. Atmos. Chem. Phys. 14, 8449e8460.

Gibson, M.D., Haelssig, J., Pierce, J.R., Parrington, M.P., Franklin, J.E., Hopper, J.T.,Li, Z., Ward, T., 2015. A comparison of four receptor models used to quantify theboreal wildfire smoke contribution to surface PM2.5 in Halifax, Nova Scotiaduring the BORTAS-B experiment. Atmos. Chem. Phys. 15, 815e827.

Goel, A., Kumar, P., 2014. A review of fundamental drivers governing the emissions,dispersion and exposure to vehicle-emitted nanoparticles at signalised trafficintersections. Atmos. Environ. 97, 316e331.

Gouriou, F., Morin, J.-P., Weill, M.-E., 2004. On-road measurements of particlenumber concentrations and size distributions in urban and tunnel environ-ments. Atmos. Environ. 38, 2831e2840.

Hagemann, R., Corsmeier, U., Kottmeier, C., Rayk, R., Wieser, A., Vogel, B., 2014.Spatial variability of particle number concentrations and NOx in the Karlsruhe(Germany) area obtained with the mobile laboratory ‘AERO-TRAM’. Atmos.


Environ. 94, 341e352.Hankey, S., Marshall, J.D., 2015. On-bicycle exposure to particulate air pollution:

Particle number, black carbon, PM2.5 and particle size. Atmos. Environ. 122,65e73.

Hansen, A.D., Rosen, H., Novakov, T., 1984. The Aethalometer e an instrument forthe real time measurements of optical absorption by aerosol particles. Sci. TotalEnviron. 36, 191e196.

Hsu, A., Reuben, A., Shindell, D., de Sherbinin, A., Levy, M., 2013. Toward the nextgeneration of air quality monitoring indicators. Atmos. Environ. 80, 561e570.

INPE-Instituto Nacional de Pesquisas Espaciais, 2015. Portal do Monitoramento deQueimadas e Incendios. Online at. http://queimadas.cptec.inpe.br. Accessed 09June 2016.

Janssen, N.A.H., Gerlofs-Nijland, M.E., Lanki, T., Salonen, R.O., Cassee, F., Hoek, G.,Fischer, P., Brunekreef, B., Krzyzanows, M., 2012. In: Bohr, R. (Ed.), Health Effectsof Black Carbon. WHO Regional Office for Europe, Copenhagen, pp. 1e86. Onlineat. http://www.unece.org/fileadmin/DAM/env/lrtap/conv/Health_Effects_of_Black_Carbon_report.pdf. Accessed 05 July 2016.

Jayaratne, E.R., Wang, L., Heuff, D., Morawska, L., Ferreira, L., 2009. Increase inparticle number emissions from motor vehicles due to interruption of steadytraffic flow. Transp. Res. D 14, 521e526.

Je�zek, I., Katra�snik, T., Westerdahl, D., Mo�cnik, G., 2015. Black carbon, particlenumber concentration and nitrogen oxide emission factors of random in-usevehicles measured with the on-road chasing method. Atmos. Chem. Phys. 15,11011e11026.

Ketzel, M., Wåhlin, P., Kristensson, A., Swietlicki, E., Berkowicz, R., Nielsen, O.J.,Palmgren, F., 2004. Particle size distribution and particle mass measurements aturban, near-city and rural level in the Copenhagen area and Southern Sweden.Atmos. Chem. Phys. 4, 281e292.

Krecl, P., Johansson, C., Str€om, J., L€ovenheim, B., Gallet, J.-C., 2014. A feasibility studyof mapping light-absorbing carbon using a taxi fleet as a mobile platform. TellusB. http://dx.doi.org/10.3402/tellusb.v66.23533.

Krecl, P., Targino, A.C., Johansson, C., Str€om, J., 2015. Characterisation and sourceapportionment of submicron particle number size distributions in a busy streetcanyon. Aerosol Air. Qual. Res. 15, 220e233.

Krecl, P., Targino, A.C., Wiese, L., Ketzel, M., Correa, M.P., 2016. Screening of short-lived climate pollutants in a street canyon in a mid-sized city in Brazil.Atmos. Poll. Res. http://dx.doi.org/10.1016/j.apr.2016.06.004.

Kumar, P., Morawska, L., Birmili, W., Paasonen, P., Hu, M., Kulmala, M.,Harrison, R.M., Norford, L., Britter, R., 2014. Ultrafine particles in cities. Environ.Int. 66, 1e10.

Mathissen, M., Scheer, V., Vogt, R., Benter, T., 2011. Investigation on the potentialgeneration of ultrafine particles from the tire-road interface. Atmos. Environ.45, 6172e6179.

Monks, P.S., Granier, C., Fuzzi, S., Stohl, A., Williams, M.L., Akimoto, H., Amanni, M.,Baklanov, A., Baltensperger, U., co-authors, 2009. Atmospheric compositionchange e global and regional air quality. Atmos. Environ. 43, 5268e5350.

Moosmüller, H., Chakrabarty, R.K., Arnott, W.P., 2009. Aerosol light absorption andits measurement: a review. J. Quant. Spectrosc. Ra. 110, 844e878.

McNamara, M.L., Noonan, C.W., Ward, T.J., 2011. Correction factor for continuousmonitoring of wood smoke fine particulate matter. Aerosol Air. Qual. Res. 11,315e322.

Miranda, R.M., Andrade, M.F., Fornaro, A., Astolfo, R., Andre, P.F., Saldiva, P., 2010.Urban air pollution: a representative survey of PM2.5 mass concentrations insix Brazilian cities. Air Qual. Atmos. Health. http://dx.doi.org/10.1007/s11869-010-0124-1.

Pattinson, W., Longley, I., Kingham, S., 2014. Using mobile monitoring to visualizediurnal variations of traffic pollutants across two near-highway neighbour-hoods. Atmos. Environ. 94, 782e792.

Pitz, M., Cyrys, J., Karg, E., Wiedensohler, A., Wichmann, H.-E., Heinrich, J., 2003.Variability of apparent particle density of an urban aerosol. Environ. Sci.Technol. 37, 4336e4342.

Pope III, 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, 1132e1141.

Rolph, G.D., 2016. Real-time Environmental Applications and Display System(READY). NOAA Air Resources Laboratory, College Park, MD. Website. http://www.ready.noaa.gov.

Santoso, M., Lestiani, D.D., Hopke, P.K., 2013. Atmospheric black carbon in PM2.5 inIndonesian cities. J. Air Waste Manag. Assoc. 63, 1022e1025.

Stieb, D.M., Burnett, R.T., Smith-Doiron, M., Brion, O., Shin, H.H., Economou, V.,2008. A new multipollutant, no-threshold air quality health index based onshort-term associations observed in daily time-series analyses. J. Air WasteManag. 58, 435e450.

Targino, A.C., Krecl, P., 2016. Local and regional contributions to black carbonaerosols in a mid-sized city in southern Brazil. Aerosol Air. Qual. Res. 16,125e137.

Tasic, V., Jovasevic-Stojanovic, M., Vardoulakis, S., Milosevic, N., Kovacevic, R.,Petrovic, J., 2012. Comparative assessment of a real-time particle monitoragainst reference gravimetric method for PM10 and PM2.5 in indoor air. Atmos.Environ. 54, 358e364.

United Nations, 2014. World Urbanization Prospects: the 2014 Revision, Highlights(ST/ESA/SER.A/352). Department of Economic and Social Affairs, PopulationDivision.

Wahlstr€om, J., S€oderberg, A., Olander, L., Jansson, A., Olofsson, U., 2010. A pin-on-disc simulation of airborne wear particles from disc brakes. Wear 268, 763e769.




http://refhub.elsevier.com/S0269-7491(16)30597-8/sref1





http://dx.doi.org/10.1029/2011JD017144

http://dx.doi.org/10.1029/2011JD017144







































































http://queimadas.cptec.inpe.br

http://www.unece.org/fileadmin/DAM/env/lrtap/conv/Health_Effects_of_Black_Carbon_report.pdf

http://www.unece.org/fileadmin/DAM/env/lrtap/conv/Health_Effects_of_Black_Carbon_report.pdf


















http://dx.doi.org/10.3402/tellusb.v66.23533






http://dx.doi.org/10.1016/j.apr.2016.06.004





















http://dx.doi.org/10.1007/s11869-010-0124-1

http://dx.doi.org/10.1007/s11869-010-0124-1













http://www.ready.noaa.gov

http://www.ready.noaa.gov






























Wallace, L.A., Wheeler, A.J., Kearney, J., van Ryswyk, K., You, H., Kulka, R.H.,Rasmussen, P.E., Brook, J.R., Xu, X., 2011. Validation of continuous particlemonitors for personal, indoor, and outdoor exposures. J. Expo. Sci. Environ.Epidemiol. 21, 49e64.

Weijers, E.P., Khlystov, A.Y., Kos, G.P.A., Erisman, J.W., 2004. Variability of particulateconcentrations along road and motorways determined by a moving measure-ment unit. Atmos. Environ. 38, 2993e3002.

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

Wheeler, A.J., Smith-Doiron, M., Xu, X., Gilbert, N.L., Brook, J.R., 2008. Intra-urbanvariability of air pollution in Windsor, Ontario - measurement and modeling forhuman exposure assessment. Environ. Res. 106, 7e16.

Woo, K.S., Chen, D.R., Pui, D.Y.H., McMurry, P.H., 2001. Measurement of Atlanta


aerosol size distributions: observations of ultrafine particle events. Aerosol Sci.Technol. 34, 75e87.

Wu, H., Reis, S., Lin, C., Beverland, I.J., Heal, M.R., 2015. Identifying drivers for theintra-urban spatial variability of airborne particulate matter components andtheir interrelationships. Atmos. Environ. 112, 306e316.

Yin, Z., Ye, X., Jiang, S., Tao, Y., Shi, Y., Yang, X., Chen, J., 2015. Size-resolved effectivedensity of urban aerosols in Shanghai. Atmos. Environ. 100, 133e140.

Zhu, Y., Hinds, W.C., Kim, S., Sioutas, C., 2002. Concentration and size distribution ofultrafine particles near a major highway. J. Air Waste Manag. Assoc. 52,1032e1042.

Zieger, P., Fierz-Schmidhauser, R., Weingartner, E., Baltensperger, U., 2013. Effects ofrelative humidity on aerosol light scattering: results from different Europeansites. Atmos. Chem. Phys. 13, 10609e10631.






































Documents

Hotspots of black carbon and PM2.5 in an urban area and ...afrg.peas.dal.ca/publishing/Targino et al 2016_Hotspots of black carbon... · Mobile sampling Particulate matter abstract