lable at ScienceDirect

Atmospheric Environment 44 (2010) 1062e1070

Contents lists avai

Atmospheric Environment

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

Chemical characterization and source apportionment of fine and coarseparticulate matter in Lahore, Pakistan

Elizabeth Stone a, James Schauer a,*, Tauseef A. Quraishi b, Abid Mahmood b

aUniversity of Wisconsin-Madison, Madison, WI, USAbUniversity of Engineering and Technology, Lahore, Pakistan

a r t i c l e i n f o

Article history:Received 15 April 2009Received in revised form28 October 2009Accepted 12 December 2009

Keywords:AerosolSource apportionmentPakistan

* Corresponding author. Tel.: þ1 608 262 4495; faxE-mail address: jjschauer@wisc.edu (J. Schauer).

1352-2310/$ e see front matter � 2009 Elsevier Ltd.doi:10.1016/j.atmosenv.2009.12.015

a b s t r a c t

Lahore, Pakistan is an emergingmegacity that is heavily pollutedwith high levels of particle air pollution. Inthis study, respirable particulate matter (PM2.5 and PM10) were collected every sixth day in Lahore from 12January 2007 to 19 January 2008. Ambient aerosol was characterized using well-established chemicalmethods for mass, organic carbon (OC), elemental carbon (EC), ionic species (sulfate, nitrate, chloride,ammonium, sodium, calcium, and potassium), and organic species. The annual average concentration(�one standard deviation) of PM2.5 was 194 � 94 mg m�3 and PM10 was 336 � 135 mg m�3. Coarse aerosol(PM10�2.5) was dominated by crustal sources like dust (74� 16%, annual average� one standard deviation),whereas fine particles were dominated by carbonaceous aerosol (organic matter and elemental carbon,61 � 17%). Organic tracer species were used to identify sources of PM2.5 OC and chemical mass balance(CMB) modeling was used to estimate relative source contributions. On an annual basis, non-catalyzedmotor vehicles accounted for more than half of primary OC (53 � 19%). Lesser sources included biomassburning (10 � 5%) and the combined source of diesel engines and residual fuel oil combustion (6 � 2%).Secondary organic aerosol (SOA) was an important contributor to ambient OC, particularly during thewinter when secondary processing of aerosol species during fog episodes was expected. Coal combustionalone contributed a small percentage of organic aerosol (1.9 � 0.3%), but showed strong linear correlationwith unidentified sources of OC that contributed more significantly (27 � 16%). Brick kilns, where coal andother lowquality fuels are burned together, are suggested as themost probable origins of unapportionedOC.The chemical profiling of emissions from brick kilns and other sources unique to Lahorewould contribute toa better understanding of OC sources in this megacity.

� 2009 Elsevier Ltd. All rights reserved.

1. Introduction

One of the many problems afflicting the world's largest cities isparticulate matter in the atmosphere. High levels of ambient particu-late matter have been linked to increased mortality (Schwartz andMarcus, 1990; Saldiva et al., 1995) and the inhalation of respirableparticles provides a pathway of human exposure to air toxins (Smithet al., 1996). The World Health Organization (WHO) estimates thaturbanairpollutioncausesapproximately360,000prematuredeaths inAsia each year. Of particular interest to human health are the particlesthat canenter the lung,whichhave anaerodynamicdiameter less than2.5 or 10 mm (PM2.5 or PM10). These particles are considered to be airtoxins and were the subject of WHO guidelines in 2008 when 24-hmaximum thresholds for fine (PM2.5) and coarse (PM10) particulatematter were set at 25 and 50 mg m�3, respectively (WHO, 2008).

: þ1 608 262 0454.

All rights reserved.

Lahore, Pakistan is an emerging megacity afflicted with highlevels of particle air pollution well above WHO guidelines. The cityis highly-populated, heavily-industrialized, and located in an aridregion, which gives rise to numerous PM sources including motorvehicles, industry, power plants, and wind-blown dust. Air qualityissues in Lahore have been exacerbated by the high density ofprimary industrial sources, the absence of effective pollution controls(Biswas et al., 2008), and a lack of mass transportation (Aziz andBajwa, 2008). Severe wintertime pollution episodes, sometimeslasting two weeks, have caused marked reductions in visibility, dis-rupted transportation, and triggered both injury and death (Hameedet al., 2000). Previous studies of ambient air quality in Lahore havedocumented 24-h maximum wintertime PM2.5 concentrations of200 mg m�3 (Biswas et al., 2008), springtime PM10 concentrations of460 mg m�3 (Zhang et al., 2008a), and total suspended particleconcentrations above 900 mg m�3 (Ghauri et al., 2007). The PMin Lahore has been characterized as containing high levels of toxicand/or carcinogenic components such as lead (Pb) and polyaromatichydrocarbons (PAH) (Smith et al., 1996).

E. Stone et al. / Atmospheric Environment 44 (2010) 1062e1070 1063

Reductions in ambient PM and improvements in air qualityin Lahore requires an understanding of the composition of ambientaerosol and its major sources. This study provides a base line under-standing of PM composition and sources of fine and coarse particulatematter for theyear 2007 that canbe used to guide emissions reductionstrategies and comparatively to assess future changes in air quality.These data are crucial in air resources management and are largelyunavailable for rapidly growing megacities, particularly in under-studied regions of the world. The analysis of a year-long datasetprovides significant insight into the factors that drive seasonal varia-tions in particulate matter composition and sources. Results fromthis study are compared to earlier studies in Lahore to illustrate thiscity's ongoing air quality issues and are evaluated in the context ofmegacities worldwide.

2. Methodology

2.1. Sample collection

Particulatematter samples were collected in Lahore, Pakistan onthe campus of the University of Engineering and Technology (UET)on the roof of the Institute for Environmental Engineering andResearch at a height of 10 m. The medium-volume PM2.5 and PM10

sampling apparatus (URG-3000, Chapel Hill, NC, USA) was madeof Teflon-coated aluminum. Air flow through the sampler wasinitiated by a vacuum pump and controlled by critical orifices. Flowrates were approximately 16 liters per minute (lpm) through eachof two PM2.5 cyclones and 32 lpm through the PM10 inlet. Flow intothe sampler was split between eight filter holders to a rate of 8 lpmand was measured before and after sample collection with a cali-brated Rotameter. Particulate matter was collected on Teflon andpre-cleaned quartz-fiber substrates (47 mm, Pall Life Sciences, EastHill, NY, USA). All filters and samples were handled using cleantechniques. Samples were collected on sixty-three different days,at a rate of every sixth day from 12 January 2007 to 19 January2008. Collection of samples began at 11:00 local time andcontinued for 24 h. Field blanks were collected every fifth samplingday. Two samples of motor oil commonly used in Lahore for enginelubrication and as additives to gasoline for two-stroke engineswerecollected following methods described in the SupplementalInformation.

Lahore is located between 31�150and 31�450North latitude and74�010and 74�390 East longitude at 217 m above sea level. The RaviRiver flows on the northern side of Lahore and the internationalborder (with India) is about 15 miles (25 km) to the east of thesampling site. There is intense agriculture in areas around Lahore,including on the Indian side of the border, with the main crops beingwheat and rice. After crop harvesting the biomass residues aresometimes set on fire in the fields. The immediate surroundingsof the sampler location can be described as a typical urban area ofLahore. There are also small metallurgical, chemical and electro-plating units dispersed in the area. Nearby is Grand Trunk Road,a major arterial roadway in the city. Traffic includes animal-drivencarts, diesel-powered buses and trucks, 2-stroke motorcycles,and three-wheelers (rickshaws) running on a mixture of petroleumgasoline and motor oil, and motor vehicles that use petroleumgasoline or compressed natural gas. The sampling location wasapproximately 8 km northwest of the Lahore international airport,2 km north of a railway service station that services diesel locomo-tives, and 500m away from a natural gas burning power plant. Therewere a number of brick kilns approximately 6 km northeast, whichburned low quality coal, rubber tires, and other combustible mate-rials available at low cost. During the fall and winter season, heapsof solid waste comprising mostly of fallen leaves, garden cuttings,plastic bags, and paper are set alight along road sides.

2.2. Chemical analysis

Particulate mass was measured gravimetrically using a high-precision microbalance (Mettler Toledo, USA) and taking the differ-ence of the pre- and post-sampling filter weights. Elemental andorganic carbon (EC and OC) were measured on-filter using a thermal-optical analyzer (Sunset Laboratories, USA) following the ACE-Asiabase case protocol (Schauer et al., 2003). Water-soluble ions (sulfate,nitrate, chloride, ammonium, sodium, calcium, and potassium)were measured in water-extracts of individual Teflon filters usingion-exchange chromatography (Dionex Corporation, USA). Organicspecies were measured in monthly composite samples that consistedof equal quartz-fiber filter portions from 4 to 6 samples per month.Filterswere pre-spikedwith isotopically-labeled standard solutions toaid in quantification then extracted by dichloromethane and meth-anol using Soxhlets, followed by rotary evaporation and reduction involume under purified nitrogen. Extracts were analyzed twice by GC(6890)-MS(5973) (Agilent Technologies, with DB-5 capillary column)after derivitization of carboxylic acids with diazomethane and aftersilylation of hydroxyl groups (Nolte et al., 2002). Additional detailsabout the GCMS analysis are described in detail elsewhere (Stoneet al., 2008). The following molecular markers were observed: levo-glucosan, 17-a(H)-21-b(H)-hopane, 17-b(H)-21-a(H)-norhopane, 17-a(H)-22,29,30-trisnorhopane, abb-20(S&R)-C29-sitostane, benzo(b)flu-oranthene, benzo(k)fluoranthene, benzo(e)pyrene, indeno(1,2,3-cd)pyrene, benzo(ghi)perylene, picene, and C31eC33 n-alkanes. Motoroil samples were diluted in dichloromethane to concentrations of0.2e0.4% by weight and were analyzed analogously to particulatematter samples for quantification of organic species. All measure-ments were field blank subtracted and uncertaintieswere propagatedfrom the standard deviation of field blanks and a percentage of themeasurement.

2.3. Source apportionment

Source contributions to OC were estimated using the UnitedStates Environmental Protection Agency's Chemical Mass Balancemodel (EPA CMB v8.2) which solved the effective-variance least-squares solution to a linear combination of PM sources and theirrelative contributions to aerosol species (Watson et al., 1984). Themodel assumed that the input source profiles were representativeof sources in Lahore. Source profiles were selected from the litera-ture based on a previous CMB study of PM10 in Lahore; theyincluded non-catalyzed gasoline vehicles (Schauer et al., 2002),diesel engines (Lough et al., 2007), coal soot from low temperaturecombustion representative of kilns or small industries (Zhang et al.,2008b), vegetative detritus (Rogge et al., 1993a), and open burningof biomass (Lee et al., 2005). The source profile for diesel was foundto be co-linear with that of fuel oil (Rogge et al., 1997), which wasexpected to be an important source based on the presence of Ni andV (Pakkanen et al., 2001). Consequently diesel and residual oilcombustionwere lumped together in source apportionment as onesource. Contributions to OC from natural gas (Rogge et al., 1993b)were not statistically significant.

Molecular marker compounds were selected as fitting speciesand assumed to be atmospherically stable during transportfrom source to receptor (Schauer et al., 1996). Model results wereconsidered acceptable if R2 > 0.80, c2 < 7, and if calculated speciesconcentrations agreed within 25% of the measured value. Thecriteria used to evaluate CMB results were based upon the softwaremanual (EPA, 2004) and prior publications (Sheesley et al., 2007;Zhang et al., 2008a). For July and August samples, calculated OCexceeded measured OC by 2% and 10%, respectively. For these twomonths other sources were not statistically significant and percentcontributions to OC were normalized to the calculated value.

E. Stone et al. / Atmospheric Environment 44 (2010) 1062e10701064

Vegetative detritus contributions were not apportioned during themonth of November because n-alkanes did not demonstrate anodd-carbon preference indicative of modern plant material.

3. Results and discussion

3.1. Particulate mass and composition

Ambient concentrations of PM2.5 and PM10 in Lahorewere amongthe highest ever documented in the world. Monthly-average fine(PM2.5) and coarse (PM10�2.5) particlemass and bulk composition areshown in Fig. 1 and are summarized in Table 1. Monthly PM2.5concentrations averaged 200� 80 mgm�3 (�one standard deviation)and PM10 concentrations averaged 340 � 100 mg m�3. Daily PM2.5concentrations were in excess of the WHO guideline value (of25mgm�3) every samplingday (n¼ 63); in fact, theywere in excess of100 mg m�3 on 84% of days. Likewise, PM10 concentrations were inexcess of theWHOguideline value (of 50 mgm�3) every sampling dayand were greater than 200 mg m�3 on 84% of days. The highest PM2.5concentration observed was 410 mg m�3 and occurred on 14November 2007, while the maximum PM10 concentration of650 mgm�3 occurred on 9 October 2007. PM levels in Lahorewere onthe same order of magnitude as previous studies in this megacity(Husain et al., 2007; Biswas et al., 2008; Zhang et al., 2008a) and otherurban locations inPakistan (Ghauri et al., 2007), butwere significantlyhigher than in less-populated Asian locations (Hopke et al., 2008).

PM2.5

late-Jan Feb Mar Apr May Jun

Co

ncen

tratio

n (μg

m

-3

)

0

100

200

300

400

PM10-2.5

late-Jan Feb Mar Apr May Jun

Co

ncen

tratio

n (μg

m

-3

)

0

100

200

300

400

Fig. 1. Chemical composition of ambient fine (PM2.5) and coarse (PM10e2.5) particulate mattewith organic carbon (hydrogen, oxygen, nitrogen, etc.), toxic metals, and species that were

Carbonaceous aerosol, defined as OC and EC, was the dominantcomponent of PM2.5, accounting for 37 � 8% of fine particle massbut was not a major component of coarse PM, contributing only5 � 2% of mass. Monthly-average PM2.5 OC concentrations rangedfrom a lowof 26 mgm�3 in July to a high of 152 mgm�3 in November,while EC followed an analogous trend and ranged from 4.5 to21 mg m�3. EC accounted for an average of 1.4% of coarse PM, whichprobably derived from tire wear, emissions from inefficientcombustion including brick kilns, and atmospheric coagulation ofsoot particles with coarse PM (Chuang et al., 2003).

A previous study in Lahore documented average EC concentra-tions of 22 mg m�3 during November to January 2006 (Husain et al.,2007). The ratio of OC to EC increased during 2007 from 4.1 in late-January to 9.1 in December, a trend similar to previous observationsin Hong Kong (Zheng et al., 2006). The water-soluble fraction of OCaveraged 19 � 9% and ranged from a low of 10% during the summermonsoon to 41% in October as shown in Fig. 2.

The amount of dust in fine and coarse PM was estimated as thesum of crustal elements in their oxide form: Al2O3, SiO2, TiO2, K2O,CaO, MnO2, and Fe2O3, where Si concentrations were estimatedfrom Al:Si ratios (Seinfeld and Pandis, 1998) and crustal-potassiumrepresented the non-water-soluble fraction. Ambient concentra-tions of metal oxides are shown in Supplemental Table S1 andadditional details can be found in von Schneidemesser et al. (inpress). By this calculation, dust accounted for an average of74 � 16% of coarse PM and 14 � 6% of PM2.5. For the month of May,

Jul Aug Sep Oct Nov Dec early-Jan

Jul Aug Sep Oct Nov Dec early-Jan

Organic CarbonElemental Carbon Dust Chloride Sulfate Nitrate Ammonium Other Mass

r in Lahore on a monthly basis in 2007e2008. Other mass includes elements associatednot measured.

Table 1Measured components of fine (PM2.5) and coarse (PM10e2.5) particulate matter in Lahore, Pakistan. Dust was estimated based on concentrations crustal metals in oxide formand inorganic ions represent water-soluble fraction.

Date Mass mg m�3 OrganicCarbon mg C m�3

Elementalcarbon mg C m�3

Dust mg m�3 Sulfate mg m�3 Nitrate mg m�3 Chloride mg m�3 Ammoniummg m�3

PM2.5

Jan-07 294.9 76.5 18.5 13.2 16.6 16.2 14.6 9.3Feb-07 204.5 63.9 14.5 15.3 8.9 6.7 5.1 6.1Mar-07 177.1 44.6 11.1 16.3 7.8 4.7 4.4 3.8Apr-07 189.1 46.2 12.6 20.5 6.4 3.3 5.5 2.1May-07 138.0 40.7 6.6 29.0 6.8 3.0 2.5 1.5Jun-07 117.5 26.1 4.5 31.5 6.9 2.6 3.4 1.3Jul-07 96.1 31.5 5.2 19.8 10.3 1.3 0.3 3.6Aug-07 113.5 31.7 5.3 16.5 8.0 1.7 2.1 2.7Sep-07 135.2 41.9 5.6 16.0 14.0 3.1 1.5 4.9Oct-07 268.1 70.4 13.1 36.1 11.8 7.6 7.4 1.7Nov-07 343.3 152.0 21.0 47.0 22.4 15.9 8.1 5.5Dec-07 258.6 125.6 13.8 24.1 8.8 11.3 10.5 2.1Jan-08 215.0 85.7 14.0 27.1 7.3 7.8 7.4 2.8

PM10e2.5

Jan-07 148.3 9.1 1.5 85.3 0.6 1.4 0.6 bdFeb-07 144.7 8.2 3.4 91.4 1.3 1.0 0.9 bdMar-07 135.4 6.7 1.5 78.6 0.8 1.3 0.5 bdApr-07 182.0 7.0 4.8 82.4 1.3 1.5 0.7 bdMay-07 173.6 8.3 0.2 179.3 1.9 2.3 0.7 bdJun-07 158.6 3.3 0.0 141.1 1.4 2.1 0.8 bdJul-07 99.8 1.4 0.1 68.9 1.3 2.9 0.4 bdAug-07 99.9 4.4 0.8 73.4 1.2 2.2 0.7 bdSep-07 115.8 4.7 1.6 100.6 1.4 3.8 0.7 bdOct-07 172.8 6.4 2.2 145.6 2.2 3.5 1.5 bdNov-07 180.1 1.9 6.1 135.1 4.3 5.5 3.4 bdDec-07 150.7 3.1 6.0 108.5 3.0 2.7 3.8 bdJan-08 81.2 4.7 0.0 71.0 0.6 1.1 0.7 bd

bd ¼ Below detection.

E. Stone et al. / Atmospheric Environment 44 (2010) 1062e1070 1065

the amount of dust estimated exceeded coarse PM mass by 3%,which indicated that this estimation method was not optimized forLahore. The contribution of dust to PM followed an annual trend inwhichmaximum concentrations occurred in the spring dust seasonand the dry winter season.

Inorganic ions that were significant components of PM2.5included sulfate (6 � 2%), nitrate (3 � 1%), chloride (3 � 1%), andammonium (2 � 1%). Sodium and potassium contributed to <1% offine particle mass. These ions were not significant contributors tocoarse PM, constituting<4% ofmass. Sulfate, nitrate, and ammoniumwere considered to be secondary inorganic species that were formedin the atmosphere from gas-phase precursors. Very high chloridelevels of 5e15 mg m�3 were observed, which were associated withmetal industry sources in a preceding study (Quraishi et al., 2009).

The difference between measured components and total masscorresponded to 35 � 9% of PM2.5 and 17 � 14% of coarse PM.Unmeasured components include the auxiliary elements associatedwith OC that make up organic matter (OM) including oxygen,nitrogen, hydrogen, and others, which were estimated to equal 80%of OC mass in a earlier study in Lahore (Zhang et al., 2008a). Byapplying this same ratio, auxiliary elements accounted for 81%and 34% of unidentified PM2.5 and coarse mass, respectively. Theremaining mass was expected to represent particle-phase water,additional components of dust and/or soil, and other metal species.

3.2. Organic species and source identification

The relative importance of OC to PM2.5 warranted further inves-tigation of this aerosol component. To this end, organic molecularmarkers were measured in monthly composite samples of PM2.5.While molecular markers make up a small fraction of total aerosolmass in Lahore, they contain valuable information about the presenceof aerosol sources and can be used quantitatively to assess relative

source contributions (Zhang et al., 2008a). Ambient concentrations ofseveralmolecularmarkers observed in Lahore are shown in Fig. 2 andare summarized in Table S2 in the Supplemental Information.

Polyaromatic hydrocarbons (PAH) were indicative of combustionsources and five PAH isomers with molecular weights (MW) of252 amu are shown in Fig. 2. These compounds were present atan average concentration of 370� 50 ngmg�1 OC�1 (�one standarddeviation) and followed the same temporal trend as PM2.5 and OC.Many PAH are known or believed to be carcinogenic and highconcentrations are considered to be a threat to human health. InLahore, monthly-average concentrations of benzo(a)pyrene rangedfrom 2 to 13 ng m�3 and averaged 5.3 ng m�3, which were compa-rable to previous levels observed in Lahore (Smith et al., 1995). PAHlevels were elevated in comparison to other Asian locations likeMumbai (2.0 ngm�3) (Kulkarni and Venkataraman, 2000) andHongKong (0.75e1.0 ng m�3) (Zheng et al., 2006) and were an order ofmagnitude greater than levels observed in downtown Los Angeles(0.18e0.44 ng m�3) (Schauer et al., 1996).

Picene is a PAH with a MW of 278 amu that is producedspecifically by burning of coal (Oros and Simoneit, 2000) and wasfound at a concentration of 0.57 � 0.11 mg mg�1 OC�1 in coal sootgenerated by a residential stove (Zhang et al., 2008b). In Lahore,picene was observed at monthly-average concentrations rangingfrom 0.6 to 2.4 ng m�3. These levels were comparable to springand wintertime concentrations of 1.3e1.5 ng m�3 observed inneighboring Delhi, India (Chowdhury et al., 2007). The seasonaltrend of picene in Lahore suggested that coal combustion contrib-uted to ambient PM2.5 throughout the year, while and maximumcontributions occurred during November and December.

Levoglucosan is awell-establishedmarker for biomass combustionas it is the major product of cellulose pyrolysis (Simoneit et al., 1999);it occurs at concentration ranges of 95 � 40 mg mg�1 OC�1 in open-type biomass burning (Lee et al., 2005). Concentrations of this marker

n-Alkanesla

te-J

an Feb

Mar Ap

rM

ay Jun

Jul

Aug

Sep

Oct

Nov

Dec

early

-Jan

Con

cent

ratio

n (n

g m

-3)

0

100

200

300

400

500

Heptacosane Octacosane Nonacosane Triacontane Hentriacontane Dotriacontane Tritriacontane

Picene

Con

cent

ratio

n (n

g m

-3)

0.0

0.5

1.0

1.5

2.0

2.5

3.0Levoglucosan

Con

cent

ratio

n (μ

g m

-3)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Hopanes

late

-Jan Feb

Mar Apr

May Jun

Jul

Aug

Sep

Oct

Nov

Dec

early

-Jan

Con

cent

ratio

n (n

g m

-3)

0

50

100

150

200

17α(H)-22,29,30-Trisnorhopane 17β(H)-21α(H)-30-Norhopane 17α(H)-21β(H)-Hopane

PAH (MW 252)

Con

cent

ratio

n (n

g m

-3)

0

10

20

30

40

50

60

Benzo(b)fluorantheneBenzo(k)fluoranthene Benzo(j)fluorantheneBenzo(e)pyrene Benzo(a)pyrene

Carbonaceous Aerosol

Con

cent

ratio

n (μ

g m

-3)

0

50

100

150

200

ECnon-WSOC WSOC

*see

text

for d

etai

ls

Fig. 2. Concentrations of carbonaceous aerosol components and organic species in Lahore by month.

E. Stone et al. / Atmospheric Environment 44 (2010) 1062e10701066

were the highest of any observed organic species, and ranged from0.16 to 2.6 mg m�3. The lowest levoglucosan concentrations occurredduring the summer monsoon season and were five times greater inJanuary, November, and December compared to other months. Whilecowdung is used as a cooking fuel in the region, its primarymolecularmarkersdcoprostanol and cholestanol (Sheesley et al., 2003)dwerenot observed in Lahore aerosol samples, indicating that this sourcewas not a major contributor to aerosol in the urban area during thisstudy.

Hopanes are biomarkers that are formed geologically overmillions of years and are present in emissions from fossil fuelcombustion and related sources (Simoneit, 1986): coal combustion(Oros and Simoneit, 2000), motor vehicles (Lough et al., 2007), andfuel oil combustion (Rogge et al., 1997). A homologous series ofhopanes, including 17a(H)-22,29,30-trisnorhopane, 17b(H)-21a(H)-30-Norhopane, and 17a(H)-21b(H)-hopane, is shown in Fig. 2. Thesumof these three compounds ranged froma lowof 44 ngm�3 in Julyto a high of 110 ng m�3 in November with consistent relative ratiosthroughout the year. The levels of hopanes in Lahorewere among thehighest in the world and were much greater than those observed inmajor cities in the United States and in Beijing, China, as previouslynoted by Zhang et al. (2008a).

Concentrations of C27e33 n-alkanes are shown in Fig. 2. The rangeof individual species concentrations inmost months ranged from 6.4to 68 ngm�3 and demonstrated anodd-numbered carbonpreferencewith amaximumat C29, indicative ofmodernplantwaxcontributionsto ambient OC. Drastic differences were observed in the month ofNovember when concentrations were elevated to 400e1100 ng m�3

for these compounds and no odd or even preference was observed,suggesting fossil fuel sources were their dominant source.

3.3. Source apportionment of primary organic carbon

Primary source contributions to ambient PM2.5 OC calculated bythe CMB model are shown in Fig. 3 and are summarized in Table 2.Input sources were based upon observed molecular markers dis-cussed in the previous section and included non-catalyzed gasolinevehicles, diesel and residual oil combustion, coal soot, biomassburning, and vegetative detritus. The amount of OC apportioned tothese sources ranged from 62 to 100% of the measured value, andthe remainder was attributed to “other” sources that representedunknown primary sources and secondary organic aerosol (SOA) thatwere not included in the model (discussed in the following section).

late-Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec early-Jan

So

urc

e C

on

trib

utio

n (μg

C m

-3

)

0

20

40

60

80

100

120

140

160

non-Catalyzed Gasoline Vehicles Diesel and Residual Oil CombustionCoal SootBiomass Burning Vegetative Detritus Other Sources

late-Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec early-Jan

So

urc

e C

on

trib

utio

n (%

OC

)

0

20

40

60

80

100

a

b

Fig. 3. Source contributions to ambient PM2.5 organic carbon (OC) estimated using chemical mass balance (CMB) modeling on (a) absolute and (b) percent scales. Other sources arethe sum of secondary sources and unidentified sources.

E. Stone et al. / Atmospheric Environment 44 (2010) 1062e1070 1067

The largest single contributor to ambient PM2.5 OC was non-catalyzed gasoline vehicles whose monthly-average contributionranged from 20 � 3 to 51 � 8 mg m�3 (�standard error) andcontributed 29e86% of the observed OC (average � one standarddeviation of 53 � 19%). The maximum contributions from non-catalyzed gasoline vehicles occurred in winter months whereas therelative contributions from this source peaked during the summer.The magnitude of contributions from this source is similar to thoseobserved previously in Lahore (Zhang et al., 2008a). However,concentrations were significantly higher than observed in otherlarge metropolitan areas like Los Angeles, CA where the annualaverage contribution of catalyzed and non-catalyzed gasoline vehi-cles to fine OC was 1.6 � 0.2 mg m�3 (Schauer et al., 1996) and HongKong where seasonal average contributions from gasoline exhaustranged from 0.8� 0.4 mgm�3 to 1.4� 0.4 mgm�3 at a roadside site in2000e01 (Zheng et al., 2006). The impact of diesel and residualoil combustion was small in comparison to non-catalyzed gasolinevehicles, but contributed an average of 3.8 � 1.9 mg m�3 or 6.3% ofPM2.5 OC. Diesel contributions in Lahore were similar to Los Angeleswhere annual average contributions were 2.7� 0.3 mgm�3 (Schaueret al., 1996). In Lahore, coal soot contributed an average1.2� 0.8 mgm�3 or 2% of OC on amonthly basis. Likemotor vehicles,diesel and residual oil and coal soot were relatively consistentin Lahore over the course of the annum, indicating the year-roundimportance of these combustion sources.

Due to the motor vehicle influence on organic aerosol in Lahore,the accuracy of themotor vehicle contribution estimatewas furtherinvestigated by the analysis of Pakistani motor oil. Since two-strokeengines burn a combination of motor oil and gasoline, therewas expected to be a large potential for motor oil to contribute toambient OC in Lahore. The chemical signatures observed in Pak-istani motor oil samples were similar to those previously docu-mented in a study in the United States (Zielinska et al., 2008) asfurther described in the Supplemental Information. This indicatedthat motor vehicle emissions in Lahore could be adequatelymodeled using American source profiles and the motor vehicleemissions from two-stroke engines in Lahore were expected to begrouped with the source category of non-catalyzed vehicles.

Biomass burning was the second-largest known contributorto PM2.5 OC after non-catalyzed gasoline motor vehicles. Biomasscontributions to OC were highly variable with season and rangedfrom a low of 1.5 � 0.7 mg m�3 (5% of OC) in July to a maximum of22.6 � 10.7 mg m�3 (15% of OC) in November. The greatest relativecontribution of biomass burning occurred in late-January 2007 at22%. The estimates of biomass burningwere based upon a profile foropen-type burning (Lee et al., 2005), which has been demonstratedto be a high-end estimate of biomass contributions compared toother profiles for fireplace or wood stove combustion (Sheesleyet al., 2007). Seasonal trends are consistent with crop residueburning in agricultural areas and yard waste burning in urban areas.

Table 2Source contributions to organic carbon in fine particulate matter and uncertainties (unc). Statistically significant source contributions are shown in bold.

Date Non-catalyzedgasoline vehicles

Diesel and residualoil combustion

Coal soot Vegetative detritus Biomass burning Other sources R2 c2

mg C m�3 unc mg C m�3 unc mg C m�3 unc mg C m�3 unc mg C m�3 unc mg C m�3 unc

Jan-07 22.4 3.5 6.5 0.7 1.8 0.3 1.3 0.3 16.8 7.7 27.7 8.7 0.88 3.9Feb-07 23.8 3.7 5.2 0.5 1.3 0.2 0.9 0.2 6.1 3.0 26.5 5.1 0.86 4.3Mar-07 19.9 3.1 4.1 0.4 0.9 0.2 0.8 0.2 2.9 1.4 16.0 3.6 0.85 4.8Apr-07 24.8 3.7 4.6 0.4 1.0 0.2 1.4 0.3 3.0 1.4 11.3 4.2 0.86 4.8May-07 21.3 3.2 2.3 0.2 0.7 0.2 0.9 0.2 3.0 1.5 12.5 3.7 0.86 4.7Jun-07 19.6 3.0 1.5 0.2 0.6 0.1 0.7 0.1 2.1 1.0 1.5 3.2 0.85 5.1Jul-07 27.5 4.0 1.8 0.2 0.6 0.2 0.7 0.1 1.5 0.7 0.0 4.1 0.83 5.9Aug-07 28.9 4.1 1.8 0.2 0.6 0.2 1.1 0.2 2.4 1.1 0.0 4.4 0.84 5.3Sep-07 26.2 3.9 1.9 0.2 0.5 0.2 0.8 0.2 4.7 2.3 7.7 4.7 0.81 6.3Oct-07 32.6 4.9 4.6 0.5 1.1 0.2 1.2 0.3 8.1 3.9 22.7 6.5 0.85 5.0Nov-07 51.3 7.8 7.0 0.8 2.7 0.5 nq 22.6 10.7 68.4 13.7 0.89 3.8Dec-07 47.2 7.2 4.4 0.6 2.7 0.5 1.6 0.3 15.1 7.3 54.6 10.7 0.86 4.2Jan-08 35.6 5.5 4.8 0.5 1.7 0.3 1.2 0.3 12.9 6.2 29.5 8.8 0.85 4.7

nq ¼ Not quantified, see text for details.

E. Stone et al. / Atmospheric Environment 44 (2010) 1062e10701068

The seasonal variation in biomass burning is similar to that observedin Godavari, Nepal (Stone et al., in press), suggesting this sourcefollows region-wide trends. Vegetative detritus was a very smallcontributor to ambient OC at an average of 2.0 � 0.6%.

3.4. Secondary organic carbon and unidentified sources

The CMBmodel apportionedOC towell-defined primary sourcesand unapportioned OC is related to SOA and uncharacterized

PM2.5 Secondary Organic Aerosol

late-Jan Feb Mar Apr May Jun

Con

tribu

tion

to O

C (μ

gC m

-3)

0

10

20

30

40

50

60

PM2.5 Unidentified Sources

late-Jan Feb Mar Apr May Jun

Con

tribu

tion

to O

C (μ

gC m

-3)

0

10

20

30

40

50

60

70

a

b

Fig. 4. (a) Estimates of SOA contributions to PM2.5 organic carbon (OC) in Lahore. (b) PM2.5 Oby SOA.

primary sources. Secondary contributions to OC were estimated asthe water-soluble OC that was not associated with biomass burningemissions. Open-type biomass burning was found to be 71% water-soluble (Sannigrahi et al., 2006) and was considered to be thepredominant source of primarywater-soluble OC and all of SOAwasassumed to be water-soluble. This type of calculation was deter-mined to give a reasonable estimate of SOA at a suburban site inMexico City and was validated by co-located measurements (Stoneet al., 2008). The uncertainty in this estimationwas propagated from

Jul Aug Sep Oct Nov Dec early-Jan

Jul Aug Sep Oct Nov Dec early-Jan

C that was not apportioned to primary sources by the CMB model and not accounted for

E. Stone et al. / Atmospheric Environment 44 (2010) 1062e1070 1069

theuncertaintyof thewater-solubleOCmeasurementand71%of theuncertainty of the biomass burning contribution.

SOA contributions to PM2.5 OC in Lahore and their associateduncertainties are shown in Fig. 4a. This estimate suggested that onan annual average basis, secondary sources contributed to 11 � 8%of organic aerosol in Lahore, such that the vast majority was freshaerosol from primary sources. The estimated concentrations of SOAranged from summertime lows of 1e4 mg m�3 (5e11% of OC) towintertime highs reaching 23e27 mgm�3 in October andNovember(32% and 18%, respectively). The annual trend in SOA suggested thatpeak concentrations occurred inwintertime, specifically October toearly-January. This trend is somewhat contrary to SOA observed intheMidwestern United States, which is greatest in the summertime(Lewandowski et al., 2008), but is consistent with trends insecondary sulfate formation in Lahore that has been observed to beenhanced by wintertime fog episodes (Hameed et al., 2000; Ratti-gan et al., 2002; Biswas et al., 2008).

Unidentified source contributions are shown in Fig. 4b and wereestimated as the difference between “other” sources from the CMBmodel and estimated SOA. The uncertainty of unidentified sourceswas propagated from the uncertainty of “other” sources and SOA.These unidentified sources were not found to be statistically signif-icant during JuneeOctober but were important contributors to PM2.5

OC from November to May, accounting for 15e35% of the measuredconcentration. The annual trend in unidentified sources follows thatof total PM2.5 and consequently most other aerosol components.The seasonal trend revealed that unidentified sources peaked in thewintertime and suggested wintertime combustion-related sources.

A major contributor to the “unidentified source” category wasexpected to be brick kilns in Lahore. Previous studies in Lahore havesuggested the brick kilns are an important source of PAH (Smithet al., 1996) and sulfur (Biswas et al., 2008) to the atmosphere, alsomaking them likely contributors to organic aerosol. Kilns are knownto burn low quality coal, biomass, and other materials like rubbertires. Source profiles for Lahore brick kilns were not available at thetime of this study, norwas there detailed information on the relativeamounts of different fuels burned. The co-combustion of coal withthese other fuels, however, provides a method of assessing howthese kilnsmay have contributed to OC. To this end, we compare thecontributions of coal combustion to OC with estimates of uniden-tified sources. A relatively strong linear correlation (R2 ¼ 0.90)between the two is shown in Supplemental Figure S1. This findingsupports the hypothesis that much of OC from unidentified sourcescame from sources burning coal, such as brick kilns.

4. Conclusions

Ambient concentrations of PM2.5 and PM10 were very high inLahore at levels comparable to other Asian megacities. The magni-tude of particulate air pollution in this megacity likely has a signif-icant impact on the surrounding region. Carbonaceous componentswere amajor contributor to ambient PM2.5: OC accounted for 31% ofPM2.5, while organic matter (OC plus oxygen, hydrogen, and otherassociated elements) was estimated to account for a total of 56% ofPM2.5. The sources contributing to ambient OC were estimatedusing CMB modeling, which revealed that the most dominantcontributor to ambient OC was non-catalyzed motor vehicles, withan annual average contribution of 29 � 10 mg m�3. Other importantcontributors included diesel engines, residual fuel oil combustion,and biomass burning. While motor vehicle contributions wererelatively consistent over the course of the year-long study, biomassand coal sources demonstrated seasonal variability and peaked inthe wintertime. SOA contributions also peaked in the wintertimeand were potentially enhanced by fog processing. Coal contribu-tions were strongly-correlated with unidentified sources of OC; this

suggested that uncharacterized sources included the co-combus-tion of coal along with other materials and pointed to brick kilns asa probable source. The further characterization of emissions frombrick kilns and other industrial point sources in Lahore presents anopportunity to improve upon source apportionment. Coarse-modeparticulate matter (PM10e2.5) was largely attributed to crustalsources, like dust, which presents difficulty from an air qualitymanagement perspective. Sources of PM2.5, particularly theanthropogenic sources implicated in this study, present a betteropportunity for reducing emissions.

Acknowledgements

We thank the Government of Pakistan, the Pakistani HigherEducation Commission, and the United States Agency for Interna-tional Development (US-AID) for funding this research. We alsothank Jeff DeMinter, Brandon Shelton, Christopher Worley, MaryaOrf, Todd Jasienski, Erika von Schneidemesser, and Dr. MartinShafer for their assistance with chemical and/or data analysis.

Appendix. Supplementary material

Supplementary data associated with this article can be found inthe online version at doi:10.1016/j.atmosenv.2009.12.015.

References

Aziz, A., Bajwa, I.U., 2008. Erroneous mass transit system and its tended relationshipwith motor vehicular air pollution (an integrated approach for reduction ofurban air pollution in Lahore). Environmental Monitoring and Assessment 137(1e3), 25e33.

Biswas, K.F., Ghauri, B.M., Husain, L., 2008. Gaseous and aerosol pollutants duringfog and clear episodes in South Asian urban atmosphere. Atmospheric Envi-ronment 42 (33), 7775e7785.

Chowdhury, Z., Zheng, M., Schauer, J.J., Sheesley, R.J., Salmon, L.G., Cass, G.R.,Russell, A.G., 2007. Speciation of ambient fine organic carbon particles andsource apportionment of PM2.5 in Indian cities. Journal of GeophysicalResearch-Atmospheres 112 (D15).

Chuang, P.Y., Duvall, R.M., Bae, M.S., et al., 2003. Observations of elemental carbon andabsorption during ACE-Asia and implications for aerosol radiative properties andclimate forcing. Journal of Geophysical Research-Atmospheres 108 (D23).

Epa, U.S., 2004. EPA-CMB8.2 Users Manual.Ghauri, B., Lodhi, A., Mansha, M., 2007. Development of baseline (air quality) data in

Pakistan. Environmental Monitoring and Assessment 127 (1e3), 237e252.Hameed, S., Mirza,M.I., Ghauri, B.M., Siddiqui, Z.R., Javed, R., Khan, A.R., Rattigan, O.V.,

Qureshi, S., Husain, L., 2000. On the widespread winter fog in NortheasternPakistan and India. Geophysical Research Letters 27 (13), 1891e1894.

Hopke, P.K., Cohen, D.D., Begum, B.A., Biswas, S.K., Ni, B.F., Pandit, G.G., Santoso, M.,Chung, Y.S., Davy, P., Markwitz, A., Waheed, S., Siddique, N., Santos, F.L.,Pabroa, P.C.B., Seneviratne, M.C.S., Wimolwattanapun, W., Bunprapob, S.,Vuong, T.B., Hien, P.D., Markowicz, A., 2008. Urban air quality in the Asianregion. Science of the Total Environment 404 (1), 103e112.

Husain, L., Dutkiewicz, V.A., Khan, A., Ghauri, B.M., 2007. Characterization of carbo-naceous aerosols in urban air. Atmospheric Environment 41 (32), 6872e6883.

Kulkarni, P., Venkataraman, C., 2000. Atmospheric polycyclic aromatic hydrocar-bons in Mumbai, India. Atmospheric Environment 34 (17), 2785e2790.

Lee, S., Baumann, K., Schauer, J.J., Sheesley, R.J., Naeher, L.P., Meinardi, S., Blake, D.R.,Edgerton, E.S., Russell, A.G., Clements, M., 2005. Gaseous and particulateemissions from prescribed burning in Georgia. Environmental Science &Technology 39 (23), 9049e9056.

Lewandowski, M., Jaoui, M., Offenberg, J.H., Edney, E.O., Sheesley, R.J., Schauer, J.J.,2008. Primary and secondary contributions to ambient PM in the MidwesternUnited States. Environmental Science & Technology 42 (9), 3303e3309.

Lough,G.C., Christenson, C.C., Schauer, J.J., Tortorelli, J., Bean, E., Lawson,D., Clark,N.N.,Gabele, P.A., 2007. Development of molecular marker source profiles for emis-sions from on-road gasoline and diesel vehicle fleets. Journal of the Air & WasteManagement Association 57 (10), 1190e1199.

Nolte, C.G., Schauer, J.J., Cass, G.R., Simoneit, B.R.T., 2002. Trimethylsilyl derivativesof organic compounds in source samples and in atmospheric fine particulatematter. Environmental Science & Technology 36 (20), 4273e4281.

Oros, D.R., Simoneit, B.R.T., 2000. Identification and emission rates of moleculartracers in coal smoke particulate matter. Fuel 79 (5), 515e536.

Pakkanen, T.A., Loukkola, K., Korhonen, C.H., Aurela, M., Makela, T., Hillamo, R.E.,Aarnio, P., Koskentalo, T., Kousa, A., Maenhaut, W., 2001. Sources and chemicalcomposition of atmospheric fine and coarse particles in the Helsinki area.Atmospheric Environment 35 (32), 5381e5391.

E. Stone et al. / Atmospheric Environment 44 (2010) 1062e10701070

Quraishi, T., Schauer, J.J., Zhang, Y., 2009. Understanding sources of airborne water-soluble metals in Lahore, Pakistan. Kuwait Journal of Science & Engineering 36(1A), 43e62.

Rattigan, O.V., Mirza, M.I., Ghauri, B.M., Khan, A.R., Swami, K., Yang, K., Husain, L., 2002.Aerosol sulfate and trace elements in urban fog. Energy & Fuels 16 (3), 640e646.

Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1993a.Sources of fine organic aerosol .4. Particulate abrasion products from leaf surfacesof urban plants. Environmental Science & Technology 27 (13), 2700e2711.

Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1993b.Sources of fine organic aerosol .5. Natural-Gas Home Appliances. EnvironmentalScience & Technology 27 (13), 2736e2744.

Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1997.Sources of fine organic aerosol .8. Boilers burning No. 2 distillate fuel oil.Environmental Science & Technology 31 (10), 2731e2737.

Saldiva, P.H.N., Pope, C.A., Schwartz, J., Dockery, D.W., Lichtenfels, A.J., Salge, J.M.,Barone, I., Bohm, G.M.,1995. Air-pollution andmortality in elderly peoplee a time-series study in Sao-Paulo, Brazil. Archives of Environmental Health 50 (2),159e163.

Sannigrahi, P., Sullivan, A.P., Weber, R.J., Ingall, E.D., 2006. Characterization of water-soluble organic carbon in urban atmospheric aerosols using solid-state C-13NMR spectroscopy. Environmental Science & Technology 40 (3), 666e672.

Schauer, J.J., Kleeman, M.J., Cass, G.R., Simoneit, B.R.T., 2002. Measurement of emis-sions from air pollution sources. 5. C-1eC-32 organic compounds from gasoline-powered motor vehicles. Environmental Science & Technology 36 (6), 1169e1180.

Schauer, J.J., Mader, B.T., Deminter, J.T., Heidemann, G., Bae, M.S., Seinfeld, J.H.,Flagan, R.C., Cary, R.A., Smith, D., Huebert, B.J., Bertram, T., Howell, S., Kline, J.T.,Quinn, P., Bates, T., Turpin, B., Lim, H.J., Yu, J.Z., Yang, H., Keywood, M.D., 2003.ACE-Asia intercomparison of a thermal-optical method for the determination ofparticle-phase organic and elemental carbon. Environmental Science & Tech-nology 37 (5), 993e1001.

Schauer, J.J., Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., 1996. Sourceapportionment of airborne particulate matter using organic compounds astracers. Atmospheric Environment 30 (22), 3837e3855.

Schwartz, J., Marcus, A., 1990. Mortality and air-pollution in London e a time-seriesanalysis. American Journal of Epidemiology 131 (1), 185e194.

Seinfeld, J.H., Pandis, S.N., 1998. Atmospheric Chemistry and Physics: From AirPollution to Climate Change. John Wiley and Sons, New York.

Sheesley, R.J., Schauer, J.J., Chowdhury, Z., Cass, G.R., Simoneit, B.R.T., 2003. Charac-terization of organic aerosols emitted from the combustion of biomass indige-nous to South Asia. Journal of Geophysical Research-Atmospheres 108 (D9).

Sheesley, R.J., Schauer, J.J., Zheng, M., Wang, B., 2007. Sensitivity of molecularmarker-based CMB models to biomass burning source profiles. AtmosphericEnvironment. doi:10.1016/j.atmosenv.2007.08.011.

Simoneit, B.R.T., 1986. Characterization of organic-constituents in aerosols inrelation to their origin and transport e a review. International Journal ofEnvironmental Analytical Chemistry 23 (3), 207e237.

Simoneit, B.R.T., Schauer, J.J., Nolte, C.G., Oros, D.R., Elias, V.O., Fraser, M.P., Rogge, W.F., Cass, G.R., 1999. Levoglucosan, a tracer for cellulose in biomass burning andatmospheric particles. Atmospheric Environment 33 (2), 173e182.

Smith, D.J.T., Edelhauser, E.C., Harrison, R.M., 1995. Polynuclear aromatic hydro-carbon concentrations in road dust and soil samples collected in the United-Kingdom and Pakistan. Environmental Technology 16 (1), 45e53.

Smith, D.J.T., Harrison, R.M., Luhana, L., Pio, C.A., Castro, L.M., Tariq, M.N., Hayat, S.,Quraishi, T., 1996. Concentrations of particulate airborne polycyclic aromatichydrocarbons and metals collected in Lahore, Pakistan. Atmospheric Environ-ment 30 (23), 4031e4040.

Stone, E.A., Schauer, J.J., Pradhan, B.B., Dangol, P.B., Habib, G., Venkataraman, C.,Ramanathan, V. Characterization of emissions from South Asian biofuels andapplication to source apportionment of carbonaceous aerosol in the Himalayas.Journal of Geophysical Research-Atmospheres, in press., doi:10.1029/2009JD011881.

Stone, E.A., Snyder, D.C., Sheesley, R.J., Sullivan, A.P., Weber, R.J., Schauer, J.J., 2008.Source apportionment of fine organic aerosol in Mexico City during theMILAGRO experiment 2006. Atmospheric Chemistry and Physics 8, 1249e1259.

von Schneidemesser, E., Stone, E.A., Quraishi, T., Shafer M.M., Schauer, J.J. ToxicMetals in the Atmosphere in Lahore, Pakistan. Science of the Total Environment,in press. doi:10.1016/j.scitotenv.2009.12.022.

Watson, J.G., Cooper, J.A., Huntzicker, J.J., 1984. The effective variance weighting forleast-squares calculations applied to the mass balance receptor model. Atmo-spheric Environment 18 (7), 1347e1355.

WHO, 2008. Health Topics: Air. World Health Organization, Regional Office for theWestern Pacific. wpro.who.int/health_topics/air (accessed 02.04.08.).

Zhang, Y.X., Quraishi, T., Schauer, J.J., 2008a. Daily variations in sources of carbo-naceous aerosol in Lahore, Pakistan during a high pollution spring episode.Aerosol and Air Quality Research 8 (2), 130e146.

Zhang, Y.X., Schauer, J.J., Zhang, Y.H., Zeng, L.M., Wei, Y.J., Liu, Y., Shao, M., 2008b.Characteristics of particulate carbon emissions from real-world Chinese coalcombustion. Environmental Science & Technology 42 (14), 5068e5073.

Zheng, M., Hagler, G.S.W., Ke, L., Bergin, M.H., Wang, F., Louie, P.K.K., Salmon, L.,Sin, D.W.M., Yu, J.Z., Schauer, J.J., 2006. Composition and sources of carbona-ceous aerosols at three contrasting sites in Hong Kong. Journal of GeophysicalResearch-Atmospheres 111, D20.

Zielinska, B., Campbell, D., Lawson, D.R., Ireson, R.G., Weaver, C.S., Hesterberg, T.W.,Larson, T., Davey, M., Liu, L.J.S., 2008. Detailed characterization and profiles ofcrankcase and diesel particulate matter exhaust emissions using speciatedorganics. Environmental Science & Technology 42 (15), 5661e5666.