ARTICLE IN PRESS

1352-2310/$ - se

doi:10.1016/j.at

�Correspondfax: +662524 5

E-mail addr

Atmospheric Environment 40 (2006) 3367–3380

www.elsevier.com/locate/atmosenv

Particulate air pollution in six Asian cities: Spatial andtemporal distributions, and associated sources

N.T. Kim Oanha,�, N. Upadhyaya, Y.-H. Zhuangb, Z.-P. Haob, D.V.S. Murthyc,P. Lestarid, J.T. Villarine, K. Chengchuae, H.X. Cof, N.T. Dungg, E.S. Lindgrenh

aEnvironmental Engineering and Management, Asian Institute of Technology, ThailandbResearch Center for Eco-Environmental Sciences Beijing, China

cDepartment of Chemical Engineering, Indian Institute of Technology, MadrasdDepartment of Environmental Engineering, Institute of Technology of Bandung, Indonesia

eManila Observatory, Quezon City, PhilippinesfFaculty of Environmental Sciences, Hanoi University of Science, Vietnam

gInstitute for Environmental Science and Technology, Hanoi University of Technology, VietnamhHogskolan I Boras, Sweden

Received 8 October 2005; received in revised form 21 December 2005; accepted 18 January 2006

Abstract

A monitoring program for particulate matter pollution was designed and implemented in six Asian cities/metropolitan

regions including Bandung, Bangkok, Beijing, Chennai, Manila, and Hanoi, within the framework of the Asian regional

air pollution research network (AIRPET), coordinated by the Asian Institute of Technology. As uniform the

methodologies as possible were intended with an established QA/QC procedure in order to produce reliable and

comparable data by the network. The monsoon effects and seasonal changes in the sources/activities require long-term

monitoring to understand the nature of air pollution in the cities. During phase 1 (2001–2004) of the AIRPET around 3000

fine and coarse particulate matter samples were collected from characteristic urban sites, which provide insight into

temporal and spatial variations of PM in the cities. In all six cities, the levels of PM10 and PM2.5 were found high, especially

during the dry season, which frequently exceeded the corresponding 24 h US EPA standards at a number of sites. The

average concentrations of PM2.5 and PM10 in the cities ranged, respectively, 44–168 and 54–262 mgm�3 in the dry season,

and 18–104 and 33–180 mgm�3 in the wet season. Spatial and temporal distribution of PM in each city, the ratios of PM2.5

to PM10, and the reconstructed mass were presented which provide useful information on possible PM sources in the cities.

The findings help to understand the nature of particulate matter air pollution problems in the selected cities/metropolitan

regions.

r 2006 Elsevier Ltd. All rights reserved.

Keywords: Monitoring; Particulate matter; Contributing sources; Asian cities; AIRPET

e front matter r 2006 Elsevier Ltd. All rights reserved

mosenv.2006.01.050

ing author. Tel.: +662524 5641;

625.

ess: kimoanh@ait.ac.th (Nguyen Thi Kim Oanh).

1. Introduction

Atmospheric particles can cause multiple effectson human health and the environment. Particleswith the size less than 10 mm (PM10) have long been

.

ARTICLE IN PRESSNguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–33803368

implicated in causing adverse health effects andincreased mortality (Dockery and Pope, 1994)whereas fine (PM2.5) and ultrafine particles imposeeven higher risk (Donaldson et al., 1998; Schwartzand Neas, 2000; Ostro et al., 2006). Atmosphericparticles also interact directly and/or indirectly withthe Earth’s radiation energy balance and cansubsequently affect the global climate (IPCC, 2001;Liu and Daum, 2002). The recently discoveredatmospheric/Asian brown cloud, which containssuspended particles as the major component, isbelieved to cause multiple effects on regional airtemperature, precipitation, agriculture, air quality,and health (UNEP and C4, 2002). Due to their abilityto absorb and scatter solar radiation the atmosphericparticles, especially fine ones, effectively reducevisibility (Watson, 2002; Kim et al., 2001).

There has been a growing concern on monitoringand characterization of size-segregated ambientparticulate matter (PM) in the recent years. Effortsand resources are being spent to understand itsnature and to develop mechanisms that would helpcontrol this harmful pollutant. In many developingcountries, however, information on levels of fineparticles in the ambient air is still scarce. Untilrecently, total suspended particulate matter (TSP)was the most monitored and is still a regulatedpollutant in most developing countries. However,with the increased awareness of their harmful effectson human health and the environment, there is atendency toward monitoring fine particles in theambient air. Fragmented data available in someAsian developing countries suggest high ambientlevels of fine particles. A study in Beijing shows thatthe weekly PM2.5 in 1999–2000, at two sites, rangedfrom 37 to 357 mgm�3 (He et al., 2001). In Vietnam,the averaged level during 1996–1998 for PM2 andPM10–2, at a site in Ho Chi Minh city, was 16 and32 mgm�3, respectively (Hien et al., 2001), while fora site in Hanoi the corresponding levels, averagedfor 1998–1999, were 35 and 50 mgm�3, respectively(Hien et al., 2002). In Bangkok, Thailand, recordsshow that in the years of 2000, 2001 and 2002, thePM10 levels at curbside stations exceeded the 24 hThai National Ambient Air Quality Standard(120 mgm�3) with a frequency of 12.8%, 10.5%,and 3.8%, respectively, while at ambient sites thecorresponding exceedance frequency was 2.1%,0.3%, and 0.39% (PCD, 2001, 2002, and 2003).Limited published PM2.5 data for Bangkok showthat 24 h PM2.5 in busy parts of the city may be ashigh as 100 mgm�3 (Ostro et al., 1999).

There are no systematic and long-term publishedrecords of PM mass and composition, which areregionally compatible, in the Asian developingcountries. Such records, especially with focus onfine particles, will help to create and enhancescientific knowledge necessary to address theworsening particulate air quality in these countries.In particular, the data will help to understand thenature of the particulate pollution in a city inrelation to local sources, long-range transport andatmospheric transformation processes, which isessential for the formulation of effective air qualitymanagement strategies.

To partly fill up the gap, the Asian regional airpollution research network, or AIRPET, (http://www.serd.ait.ac.th/airpet), has designed a monitor-ing program to collect such data. AIRPET, now inphase 2, has one of the main research objectives inproviding a comprehensive assessment of PMpollution. The focus is on PM2.5 and PM10/PM10–2.5 levels and composition with the spatialand temporal distribution in six involved cities/metropolitan regions in Asia, namely, the BangkokMetropolitan Region (BMR, Thailand), Bandung(Indonesia), Beijing (China), Chennai (India), Me-tro Manila Region (Philippines), and Hanoi Me-tropolitan Region (Vietnam). The network is mainlyfunded by the Swedish International DevelopmentCooperation Agency (Sida) through the AsianRegional Research Program in EnvironmentalTechnology (ARRPET) and is coordinated by theAsian Institute of Technology (AIT). This paperpresents some findings on levels, composition andpossible source contributions of the PM in the sixcities during AIRPET phase 1, 2001–2004.

2. Methodology

2.1. Monitoring program design

Six research institutions (NRIs) are involved inthis network, including AIT; Institute of Technol-ogy of Bandung; Research Centre for Eco-Environ-mental Sciences/Beijing Normal University, Beijing;Manila Observatory; Hanoi University of Science;and Indian Institute of Technology, Madras.Each institution conducts monitoring for one city/metropolitan region in a country.

The selected cities/regions are different in termsof topography, meteorology, population, energyuse, industry, and vehicular mix and density. Theclimate of the region is dominated by monsoon with

ARTICLE IN PRESSNguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–3380 3369

two distinct seasons, dry and wet. Though the dryand wet seasons may cover different months of theyear in different countries, it has been decided tohave only these two broad season categories for theconsistency in intercity comparison of the monitor-ing results. In all cities, the wet season is within thesummer months of the Northern Hemisphere(May–October) and the dry season is the rest ofthe year, except for Bandung (Indonesia), which isjust opposite due to its location in the SouthernHemisphere. Note that even during the wet season,the samples were collected when there was nointensive continuous rain.

The PM monitoring program was designed toprovide as harmonious data in the six cities/metropolitan regions as possible. Sampling sitesare classified into six categories characterizing theurban airshed including the upwind, traffic, resi-dential, commercial, industrial, and urban mixed(downwind) sites. Each city/metropolitan region has3–5 out of these six site categories. The classificationof sites is based on the vicinity of major influencingemission sources/activities, and was done in order tohave as much consistency for all six cities aspossible. Accordingly, the upwind site refers to theupwind direction of the city in the dry season, thetraffic site is located in the city central districts andclose to traffic lanes (within 3–10m), the residentialsite is in the city crowded area, the commercial siteis in the commercial areas in the city center, theindustrial site is located in areas with moreconcentrated industrial activities within the citymetropolitan region, and the urban mixed site islocated in the downwind direction during the dryseason and is affected by most of the importantsources in the city/region. It was intended to havethe sampling intake located at the height of 3–5mfor all cities. However, due to siting difficulty a fewsites were located higher than 5m, especially thetraffic site and the residential site in Beijing, whichwere at around 40m above the ground. This,consequently, may cause inconsistency in thecomparison of results site by site, particularly forlarge particles.

In all cities monitoring for PM2.5 and PM10 orPM10–2.5 was conducted using dichotomous (di-chot), Minivol, or other equivalent samplers de-pending upon the availability. Each sampler used inthe network was compared with a dichot, as will bediscussed later. For consistency, here we presentand discuss the 24 h PM2.5 and PM10 massconcentrations. Fine (PM2.5) and coarse (PM10–2.5)

fractions by dichots were summed up to get thePM10 mass concentration. Filter media wereselected to suit the subsequent analytical methods.It was intended that all the filters/samples used inthe network to be conditioned for about 24 h(temperature 2075 1C, relative humidity 40710%) for the gravimetric mass determination,though small deviations were observed dependingon the actual conditions at each research institution.The quartz filters were fired at 550 1C for about6 h prior to conditioning for pre-weighing. Thetreatment of quartz filters at this temperatureis necessary to remove any carbonaceous/organicpollutants and is recommended for samples tobe analyzed for organic components. Detail onQA/QC employed is presented in a separatesection later.

2.2. Sample analysis

The filter samples were segregated and subjectedto multiple compositional analyses. The watersoluble ions were analyzed by IC using theprocedure given in Chow (1995). Filter samplesin Chennai (India NRI), however, were analyzedfor NH4

+, sulfate and nitrate using UV-Visiblespectrophotometer (Lodge, 1989). Elements wereanalyzed by ICP-AES (China), ICP-MS (India,Vietnam), PIXE (AIT, Philippines), XRF (Vietnam,Philippines). In Indonesia, the elements wereanalyzed by both INAA and AAS to detect moreelements.

Black carbon (BC) was measured using M43Ddigital smoke stain reflectometers for samplescollected by AIT, India, Vietnam, Indonesia, andPhilippine NRIs. A number of samples were alsoanalyzed for EC/OC, including samples collected inthe dry season at the upwind site of BMR by thethermal/optical transmittance (TOT) method (Sun-set analyzer) and samples from Beijing by a C/H/Nelemental analyzer following the method describedin Dan et al. (2004). Quartz filters were segregatedfor analysis of ions and EC/OC, while samples onmixed cellulose or Teflon filters were analyzed forelements. The numbers of analyzed parameters areslightly different for different cities due to theavailability of the equipment. The network aimsat getting as complete data sets as possible onPM composition for receptor modeling purposewhich primarily include mass, water soluble ions,elements, and BC and/or EC/OC.

ARTICLE IN PRESSNguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–33803370

2.3. Quality assurance and quality control (QA/QC)

Data quality is of the foremost concern, andimplementation of a harmonized QA/QC procedureis especially important when many research institu-tions are involved. Common methodology forsampling and analyses, as well as the comparisonbetween different sampling and analytical techni-ques used by NRIs in the network are important forobtaining a harmonized regional database. The QA/QC of the monitoring scheme were followed at allpossible stages from monitoring (siting, samplingmethod selection, sampler calibration, co-locatedsampling, selection of filter media, filter prepara-tion, sampling execution, filter weighing, filtersegregation) to the chemical analysis. The analyticalmethods for elements used by different institutionswithin the network were checked using a standardreference material (SRM), which was the coal flyash (GBW 08401), provided by the China NRI. AllNRIs analyzed this SRM except for the IndonesiaNRI which analyzed the NIST SRM #1648 (urbanPM). In phase 1 of AIRPET, the PM compositionanalytical results are considered acceptable if thedeviation is within 720–30% of the certified/reference values for major elements (concentrations41 mg g�1) and larger deviations (o2 fold) forminor elements depending on the analytical meth-ods used. Taking into account also the standarddeviations of both the certified values and theanalytical results, it is considered that the SRManalytical results produced by most of the NRIsagree reasonably with the SRM certified values,except for the method used for Chennai samples.The SRM analytical results by this NRI (ICP-MS)are off the range for most listed elements, hence, thePM composition data for Chennai will not befurther discussed in this paper.

The M43D Digital Smoke Stain Reflectometerused at AIT was calibrated against the opticalspectrometer at Boras, Sweden, which is describedin Moloi et al. (2002). Samples from Vietnamwere also measured for BC by the AIT equipment.India used the conversion formula obtained basedon the AIT reflectometer calibration. The Philippineand Indonesia NRIs measured BC using reflect-ometers available in the countries through theregional project on air pollution and trends of theInternational Atomic Energy Agency (IAEA, 2005),which were calibrated following the method pre-pared by the Max Planck Institute of Chemistry,Germany.

Regular sampler calibration was planned for thewhole network. The coarse fraction of dichotsamples was corrected for the carried-over finefraction. Most samples were collected over 24 hperiod, except for some samples in Beijing, whichwere taken for 4–8 h due to high PM loadings onfilters. The latter were subsequently normalized to24 h and the data sets for all cities are 24 h averaged.

One filter for about every 20–30 filters in a lot wasused as a blank. All the analyzed data were blankcorrected. For illustration, the quartz filters at AIT,by IC, have blank levels for Ca2+ and SO4

2� of0.4–0.5mg cm�2 filter area, for Cl�, NO3

�, and Na+ of0.14–0.15 mg cm�2, and the lowest for NH4

+, Mg2+,and K+, of around 0.03 mg cm�2. Mixed cellulosefilters for elements, by PIXE, at AIT have theaverage blank levels for Cl, S, Si, Ca, and Al (listedin the increasing order) of 0.08–0.2 mg cm�2 filterarea, whereas other elements are well below0.01 mg cm�2 and basically zero for As, Sb, andAg. Other NRIs reported the filter blank levelscomparable with those of AIT for the same filtertypes. In addition, the Whatman G/F used inBeijing have high blank levels for Cl�, NH4

+,NO3�, and SO4

2� (0.5–1 mg cm�2).Each NRI was provided with one Sierra Ander-

son dichot (series 241), which was used as areference sampler of the network and was collocatedwith other samplers used by the NRIs. AIT usedtwo collocated dichots for routine sampling whichwere compared with the US EPA FRM samplers(RASS and Partisols-Plus). The dichot results(x-axis) agree well with the FRM and with mostof other samplers (y-axis) used by different NRIsyielding a slope of 0.93–1.06 for PM2.5, and a slopeof 0.97–1.05 for PM10 and PM10–2.5 (R2 of 0.86–0.99). In addition, good agreements were alsoobtained, by AIT, between the ionic compositionof PM2.5 samples by dichots and those by FRM(Opal, 2002). The results of the collocated samplingby the Indonesia NRI (dichot and LowVol),however, have a slope of 0.84 for PM2.5 (R2 of0.57). The medium volume sampler (APM550) usedby the India NRI produced higher PM2.5 andPM10–2.5 than the collocated dichot with a slope of1.13 (R2 ¼ 0:97) for PM2.5 and 1.25 (R2 ¼ 0:99) forPM10–2.5. The Philippine NRI used Minivols forroutine sampling, which agree reasonably with thedichot with a slope of 0.93 for PM2.5 (R2 ¼ 0:86).It is noted that the comparison has been madebased on over 30 collocated samples collected byeach NRI, except for China (5–6 samples) and India

ARTICLE IN PRESSNguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–3380 3371

(13 samples). For the NRIs producing water solubleion data, the ion charge balance was also checked.The ratios of the sum of the cation equivalents(NH4

+, Na+, K+, Mg2+, Ca2+) to that of anionequivalents (Cl�, NO3

�, SO42�) are closer to 1 for

PM2.5 samples and larger (41–3) for the coarsefraction. Detail on ion balance will be presented inour forthcoming papers.

3. Results and discussion

3.1. PM levels

Totally, in phase 1 the AIRPET research teamcollected nearly 3000 PM samples in the six cities, ofwhich around 1650 are PM2.5 samples and the restare PM10/PM10–2.5. More samples were collected inthe polluted dry season than the wet season. Largernumbers of samples were collected at two desig-nated intensive sites in each city, the upwind and theurban mixed site, than other sites. All samples wereanalyzed for mass and most of the samples wereanalyzed for the chemical composition. Arithmeticaverage PM mass concentrations, the standarddeviations, and the sample numbers used for

PM2.5-d

050

100150200250300350400

Bangkok Beijing Chennai

Co

nce

ntr

atio

n, u

g/m

3

Upwind Traffic Mixed R

Upwind Traffic Mixed R

96 14 46 10No. samples 43 44 15 15 35 34 14

PM10-d

0

100

200

300

400

500

Bangkok Beijing Chennai

Co

nce

ntr

atio

n, u

g/m

3

96 14 46 10No. samples 65 47 10 10 35 34 14

(a)

(b)

Fig. 1. Average concentrations of PM2.5 (a) and PM10 (b) in the dry sea

of the mass concentration data series at each site. Note that SD mainly

emission and meteorology conditions throughout a season).

average mass calculation for each site in the sixcities during the dry and wet season are presented inFigs. 1 and 2, respectively.

PM concentrations in all the cities in the dryseason are remarkably higher than those in the wetseason. Absence of intensive wet removal, increasein local emission, e.g. coal burning for space heatingin China (He et al., 2001) and open biomassburning, more road dust/soil suspension, andstagnant meteorological conditions (Kim Oanhet al., 2005; Hien et al., 2002; He et al., 2001) aresome major factors leading to the high air pollutionconcentrations in the dry season. Long-rangetransport of pollutants from upwind regions follow-ing the Northeast monsoon may also contribute toPM levels in some cities.

Based on the data collected during 2001–2004 byAIRPET, PM2.5 in the dry season, on average, wasthe highest in Beijing (Fig. 1a). During the wetseason, the difference in PM2.5 between the cities isnot so large though Beijing still recorded consider-ably higher levels (Fig. 2a). PM10 results showalmost a similar trend as PM2.5 with the highestlevels in Beijing, followed by Hanoi (traffic and resi-dential site) and Chennai (traffic and industrial site).

ry season

Bandung Manila Hanoi

esidential Industrial Commercial

esidential Industrial Commercial

42 6 46 6 6 55 55 189 47 61 58 10 7

ry season

Bandung Manila Hanoi

42 6 46 6 6 10 29 32 19 58 20 58 20 14

son in six cities. (Whiskers represent one standard deviation, SD,

indicates the variations in daily concentrations due to changing

ARTICLE IN PRESS

PM2.5-wet season

0

50

100

150

200

250

0

50

100

150

200

250

Bangkok Beijing Chennai Bandung Manila Hanoi

Co

nce

ntr

atio

n, u

g/m

3

Upwind Traffic Mixed Residential Industrial Commercial

Upwind Traffic Mixed Residential Industrial Commercial

32 27 10 27No. samples 960 30 30 3 3 4 11 6 9 6 6 44 63 165 46 58 11 5 5

PM10-wet season

300

Bangkok Beijing Chennai Bandung Manila Hanoi

Co

nce

ntr

atio

n, u

g/m

3

32 27 10 27No. samples 40 21 21 7 3 3 4 11 6 9 6 6 6 26 74 30 11 10 22 10 10

(a)

(b)

Fig. 2. Average concentration of PM2.5 (a) and PM10 (b) in the wet season in six cities (Whiskers represent one standard deviation, SD, of

the mass concentration data series at each site. Note that SD mainly indicates the variations in daily concentrations due to changing

emission and meteorology conditions throughout a season).

Nguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–33803372

The high PM levels in Beijing, remarkably higherthan BMR and Manila which have the comparablepopulation, may be attributed to both natural andman-made factors. Extensive coal burning in Beijingduring the cold winter, when there is notably limiteddispersion, may cause high PM levels during thisperiod of the year, and the dust storms maycontribute significantly to PM in the spring (Heet al, 2001; Sun et al., 2004). The high PM levelsobserved at the upwind site of Beijing in the dryseason also reflect the impacts of PM transport fromoutside into the city. In addition, the low annualaverage precipitation in Beijing as compared toother cities would limit the wet removal of airpollutants. Though relatively small in size andpopulation, Hanoi recorded high PM levels duringthe dry season. Cold winter and more stableatmosphere under the influence of a high-pressureridge with the limited vertical dispersion due toradiative and subsidence inversion (Hien et al.,2002) enhance the high PM build up in Hanoi inthe dry season. The highest PM2.5 was observed atthe residential site, 200 mgm�3 (averaged for 10samples), which implies a high exposure for people

and also suggests a local contribution, mostprobably from domestic cooking in the nearbyliving compounds. More samples, however, arerequired to produce a reliable average PM atthis site. Note that during the dry season someweak influence of the high-pressure ridge may be feltin BMR which may reduce the pollutant dispersion.However, the coastal location would enhance thedispersion in the four cities (Bangkok, Bandung,Chennai, and Manila), Coupled with the warmclimate (no space heating), this may substantiallylower ambient air pollution even though traffic andother stationary sources in these cities may besignificant.

Remarkably higher concentrations of PM2.5 andPM10 at the traffic sites, during both seasons, inBangkok and Manila, and PM10 in Hanoi (no PM2.5

data) reflect a relatively high contribution from thetraffic. In the wet season, the difference in PM levelsbetween the traffic site and other sites in these threecities is even more pronounced, meaning the higherrelative influence of traffic, which is most probablycaused by a lesser intensity of other local andlong-range transport contributions. In Bandung, the

ARTICLE IN PRESSNguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–3380 3373

PM2.5 and PM10 levels in both seasons wererelatively low and are comparable between the sites.Average levels at different sites in Beijing are alsocomparable but are high at all sites. It is noted thatthe sampling intake at both the traffic and theresidential sites in Beijing was high (40m), whichmay not produce comparable results with othersites. In Chennai, higher PM levels, especially PM10,were observed at the industrial site, suggesting animportant contribution from industry. However,small numbers of samples collected at the sites inChennai during both seasons may not yet producerepresentative results.

PM levels varied significantly from day to day ateach site during a season, which is seen in the largestandard deviations in Figs. 1 and 2. Especially inBeijing during the dry season, when PM2.5 fromtime to time reached above 400 mgm�3 and PM10

reached above 1400–1600 mgm�3 (for example, atthe upwind site and the traffic site in December2001). Highly fluctuating PM levels in the cities arethe reflection of the changes in the meteorology andemission throughout a season with some sourcesthat are discontinuous and episodic in nature, suchas dust storms or biomass burning.

Average PM2.5 and PM10 concentrations acrossthe cities (averaged for all sites in a city) in eachseason are presented in Table 1 together with themajor PM compositions. To reduce the bias relatedto the difference in sample numbers at differentsites across a city the average was calculated fromthe seasonal averaged PM2.5 and PM10 at each site(not from the 24 h data series). PM2.5 averaged forthe dry season was the highest in Beijing(168 mgm�3) followed by Hanoi (124 mgm�3), whichwell exceeded even the short-term ambient airquality standard of the US EPA, 24 h PM2.5 of65 mgm�3. For PM10 in the dry season, the highestwas in Beijing (262 mgm�3) followed by Hanoi(186 mgm�3) and Chennai (169 mgm�3), which areeven higher than the US EPA 24 h standard of150 mgm�3. In the wet season, the PM2.5 level inBeijing was also the highest, 104 mgm�3, while inother cities the levels did not differ much with theaverage of less than 45 mgm�3. For PM10, thehighest level was in Beijing (180 mgm�3), which isclosely followed by Chennai (145 mgm�3). It isnoted that the numbers of samples differ from asite to another, and between the cities with thesmallest numbers of samples collected in Chennai.The intensive sites generally have more than 40samples during the dry season and above 25–30

samples during the wet season, but the mixed site inManila has above 160 PM2.5 samples in eachseason. The seasonal average levels calculated basedon small numbers of 24 h PM samples are not likelyto be highly representative. Long-term monitoringis desired to produce large numbers of samples atthe selected sites and hence reliable seasonalaverages, which is one of the objectives of AIRPETphase 2.

TSP was also monitored in BMR, Bandung,Beijing, and Chennai. Beijing recorded an averageTSP of 305 mgm�3 (58 samples) in the wet seasonand 440 mgm�3 (34 samples) in the dry season.Samples collected at Chennai during December–February in 2001–2002 showed an average TSP levelof 367m gm�3 (77–863 mgm�3). In Bandung, theTSP level during the dry season (July–October 2002)was 121mgm�3 (85–191 mgm�3) based on 26 sam-ples. TSP in BMR was in the range 60–140mgm�3

during the wet season and 100–300 mgm�3 duringthe dry season.

3.2. PM2.5 to PM10 ratio

The average ratio of PM2.5 to PM10 (PM2.5/PM10)was calculated for each site based on simultaneouspairs of PM2.5 and PM10 samples, and the resultsare presented in Table 2. PM2.5 constitutes a largerfraction of PM10 in most of the cities with the citywide average ratios generally X0.6 except forChennai where the ratio was only 0.3–0.32. Thecitywise as well as the sitewise average ratios for thewet season are lower than those for the dry seasonin BMR, Beijing, Bandung, and Hanoi. In Manila,lower average ratios were obtained for the dryseason at every site but with larger standarddeviations than the wet season, while in Chennaithe ratios obtained for both season are more or lessthe same.

The fine (PM2.5) and coarse particles (PM10–2.5)are relatively different in their physical and chemicalcompositions. They generally originate from differ-ent sources or by different activities within the samesource. Coarse particles are normally generated bymechanical activities such as grinding or windblowing, and are dominated by materials ofgeological origin (Chow, 1995). Fine particlesmainly consist of secondary particles, which areformed in the atmosphere by chemical reactions(e.g., sulfuric acid, ammonium bisulfate, ammo-nium sulfate, ammonium nitrate), and primaryorganic carbon and elemental carbon. Due to the

ARTICLE IN PRESS

Table 1

City wise average (for all sites) of mass and major components of PM2.5 and PM10 in mgm�3

n Massa Crustal OM Soot Sea salt NH4+ NO3

� SO42� Trace elements % Mass explained Smokeg Pbg

PM2.5 Dry

Bangkok 181 50 1.1 21.4 8.2 1.7 1.6 1.2 5.6 0.4 80 (41)e 0.8 0.04

Beijing 142 168 9.9d 64.3 18.7 1.6 12.5 14.2 20.8 1.5 40 [50]f 1.9 0.18

Chennai 83 46

Bandung 106 53 2.8 — 9.8 0.7b 3.4 5.5 8.2 0.8 59 0.5 0.19

Manila 407 44 1.4 — 21.6 0.9b — — (1.5)c 0.7 56 0.2

Hanoi 75 124 7.5 — — 1.1b — — (6.0)c 7.1 18 1.0 0.19

PM10 Dry

Bangkok 181 76 10.0 26.6 9.4 3.6 1.9 2.9 7.0 0.5 78 (45)e 0.8 0.07

Beijing 132 262 43.1d — — 2.4 13.4 19.0 29.3 3.5 42 2.4 0.32

Chennai 83 169

Bandung 106 83 7.8 — 9.8 4.6b 4.1 8.9 11.4 1.2 58 0.6 0.31

Manila 122 54 12.9

Hanoi 220 186 23.4 — 8.9 2.0 6.6 12.2 28.3 1.5 45 1.3 0.5

PM2.5 Wet

Bangkok 106 18 0.9 — 5.3 1.5 0.5 0.4 2.4 0.3 71 0.4 0.05

Beijing 115 104 4.5d 19.2 5.3 0.5 10.4 12.0 17.9 1.0 57 [30]f 1.1 0.14

Chennai 10 42

Bandung 38 38 3.1 — 7.5 0.8b 3.9 3.5 6.3 1.5 71 0.3 0.30

Manila 376 43 1.4 — 22.7 0.9b — — (0.8)c 1.6 63 0.1

Hanoi 21 33 4.0 — 4.3 — — — (3.0)c 3.8 47 0.5 0.41

PM10 Wet

Bangkok 106 33 6.1 — 6.3 3.5 0.6 1.1 3.1 0.5 74 0.4 0.07

Beijing 89 180 24.8d — — 1.1 12.1 19.1 25.1 2.6 48 1.1 0.23

Chennai 10 145

Bandung 38 62 8.6 — 7.5 5.8b 4.6 4.7 8.8 2.4 68 0.4 0.5

Manila 136 55

Hanoi 63 79 19.8 — 6.6 2.7 1.2 2.5 7.8 2.1 54 0.5 1.2

n is total number of samples taken for the average; (—) data not available; blank cells: data are not presented due to high uncertainties.aAverage of all sites in the city.bNa analyzed by method other than IC.cSulfur element content only.dSi data is not available hence is not included.eValue is for AIT site with OC data, average of other sites without OC data is in parentheses.fNo EC and OC data was included for this average calculation, in parentheses is % contribution of EC and OC alone to mass.gThese values are given as additional information. They are partly included in the previous columns on crustal (K) and trace elements,

respectively.

Nguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–33803374

time period required for the formation and atmo-spheric mixing processes, the secondary particles areexpected to have a rather uniform spatial massdistribution in the cities at a given time. Thefluctuation in PM2.5/PM10 ratios from one site toanother is largely due to the primary sources,especially those contributing to the coarse particles.Lower PM2.5/PM10 ratios at all sites in Chennaiindicate significant contributions from primarysources such as resuspended soil/road dust(unpaved roads), and other mechanical activities.Higher PM2.5/PM10 obtained in other cities suggestsrelatively high contributions from secondary

particles and the combustion sources. For examples,high ratios of PM2.5/PM10 at the traffic sites inBMR and Manila in both seasons may be attributedto the emission from the vehicular fuel combustion.Likewise, the high PM2.5/PM10 ratio at the residen-tial site in Hanoi, 0.91, suggests again a directcontribution from the surrounding residential fuelcombustion. Higher PM2.5/PM10 ratios in somecities during the dry season as compared to the wetseason may be attributed to higher fossil fuelcombustion emission; more secondary particlesformation due to more intensive photochemicalreactions, e.g. higher ozone levels in BMR in dry

ARTICLE IN PRESS

Table 2

Average and variation (1 standard deviation) of PM2.5-to-PM10 ratios at different sites in six cities

Upwind Traffic (mixed) Urban mixed Residential center Industrial Commercial City average

Dry season

Bangkok 0.6270.1 0.7170.02 0.5770.10 0.6670.08 — — 0.64

Beijing 0.5570.27 0.6570.22 — 0.6570.09 0.5470.14 — 0.60

Chennai 0.3970.07 0.2570.05 — — 0.2570.04 — 0.30

Bandung 0.7370.07 0.5270.07 0.6070.08 0.6970.10 0.6470.05 — 0.63

Manila 0.5470.19 0.7070.18 0.6170.16 — — 0.6170.23 0.61

Hanoi 0.6970.10 0.6070.04 0.6770.17 0.9170.05 — 0.8370.12 0.74

Wet season

Bangkok 0.3270.1 0.6570.08 0.3970.05 0.5270.13 — — 0.47

Beijing — 0.5170.20 — 0.4970.18 0.4670.19 0.8670.04 0.58

Chennai 0.4370.16 0.2670.08 — — 0.2770.07 — 0.32

Bandung 0.6870.16 0.6170.11 0.5670.12 0.5970.04 0.6170.04 — 0.61

Manila 0.6070.14 0.7670.10 0.6970.10 — — 0.6770.10 0.68

Hanoi — — 0.7570.21 0.5870.04 — 0.5270.05 0.62

(—) either or both PM2.5 and PM10 data not available for the ratio calculation. The ratios were calculated based on daily PM2.5/PM10 at

each site.

Nguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–3380 3375

months (Zhang and Kim Oanh, 2002); less wetremoval; and possible long-range transport of fineparticles during the NE monsoon.

3.3. Major chemical components and reconstructed

PM mass

By chemical composition the major componentsof PM10 in most urban areas can be presented inseveral groups: (i) geological material (oxides of Al,Si, Ca, Ti, and Fe), (ii) organic carbon (OC), (iii)elemental carbon, (iv) sulfate, (v) nitrate, and (vi)ammonium (Chow, 1995). In this study, to accountfor the PM nature and origin, hence to gain certaininsights into the contributing sources, the finalgroups used include the crustal, organic matter(OM), soot, sea-salt, NH4

+, SO42�, NO3

�, and ‘traceelements’ (the rest of the elements, depending on theanalytical methods used by NRIs).

It is noted that in some cities the compositiondata are not available for all ‘n’ samples presented inTable 1, which were used for the average massconcentration calculation. For example, the averageelement composition presented in Table 1 forManila was calculated based only on around 25%of the PM2.5 samples at each site in the dry season,and 15% in the wet season, for which the data areavailable. Only a few PM10 samples have thecomposition data and a limited number of PM2.5

samples in Manila have the ionic composition data,hence the related parameters are not displayed in

the table. For Hanoi, the element data are availablefor all samples but ionic data are available for onlyaround 25–30% samples at each site. Datasets ofBandung, Beijing, and BMR are more complete forall ‘n’ samples. However, Beijing has EC/OC dataonly for about 30% of samples and no BC data. ForBMR, the composition data is available for allreported samples, except for EC/OC, which isavailable only for samples at the upwind (AIT) sitein the dry season. Consequently, in some cases theaverage computed from a subset of samples at a fewsites in a city was used to represent the citywise PMcomposition, which may be biased. The use ofdifferent methods for PM2.5 and PM10 samplecollection, non-simultaneous sampling of these twoPM parameters in some cases, and differentnumbers of PM2.5 and PM10 samples at a givensite, may together cause inconsistency in thecitywise average values of the reconstructed terms.

The crustal group includes oxides of Al, Ca, Si,Ti, Fe, Na, Mg, and K. Following Chan et al.(1997), the crustal matter in this study is estimatedas 1.16 (1.9Al+2.15Si+1.41Ca+1.67Ti+2.09Fe),where the factor of 1.16 is used to compensate forthose crustal oxides (including K2O) that wereexcluded. As expected, the crustal group contributesmuch less to PM2.5 than to PM10, which indicatesthat the crustal elements are present mainly in thecoarse fraction. The dry season has higher crustalmatter levels than the wet season. The highest valuesin mgm�3 were obtained for Beijing but the relative

ARTICLE IN PRESSNguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–33803376

contributions of the crustal group in this city are inthe same ranges as in other cities: around 5% of thePM2.5 and 15% of PM10 mass. Note that, Si isreportedly high in the Asian dust (Ma et al., 2001)hence it may be high in Beijing. However, the Sidata was not available for Beijing samples to beincluded in the ‘crustal’ mass, which may substan-tially reduce the estimated ‘crustal’ contribution toPM in this city. Examining the crustal group at eachsite in the cities reveals that often the traffic siteshave the highest contribution of the crustal matterto PM10, followed by industrial and mixed urbansites, especially those near construction sites, whichindicate an important contribution from road dustresuspension and construction activities to PM.

The OM group is obtained by multiplying the OCcontent by a factor, which is in the range 1.4–2.1(Turpin and Lim, 2001). The factors may varywidely between the cities and from one season toanother depending on the nature of PM emissionsources and chemistry. However, in this study, forsimplicity, a factor of 1.7 (Hopke, 2005, personalcommunication) was used for both cities reportingOC data, BMR (the upwind site in the dry season)and Beijing (a subset of PM2.5 samples at every site).OM has a large relative contribution to PM mass inthe cities, from around 20% to above 42%.Availability of OC data improves the mass enclo-sure substantially. Thus, during the dry season, withOC data the average mass explained for samples atthe AIT site is 78–80% as compared to 41–45%obtained for other sites in BMR with no OC data.For Beijing, the EC and OC alone contribute onaverage 50% of PM2.5 mass in the dry season and30% in the wet season. Without EC and OCcomposition, the average mass explained for PM2.5

and PM10 in Beijing was around 40–57% in bothseasons.

‘Soot’ is the BC or EC (if the EC data areavailable). The majority of soot is found in PM2.5

with higher levels (in mgm�3) obtained for the dryseason. However, the difference in ‘soot’ levelsbetween the two seasons is not so pronounced ineach city except for PM2.5 in Beijing. This suggeststhat the major contributors to soot, such as traffic,remain relatively constant throughout the year inthe other cities while in Beijing the soot in the dryseason may also come from other seasonal sourcessuch as combustion for space heating. Soot is thehighest for Manila, where it contributes 49% ofPM2.5 mass in the dry season and 52% of PM2.5 inthe wet season, which is mainly due to high-average

BC at the traffic site (above 40 mgm�3 in bothseasons). For other cities, the relative contributionsof soot are 11–19% of PM2.5 during the dry seasonand 5–33% during the wet season, with the lowervalues obtained for Beijing.

The sea-salt contribution is commonly estimatedby applying a factor of 2.54 to the water solublesodium (Na+) concentrations (Ho et al., 2003;Chan et al., 1997) which are measured by the ionchromatography (IC). In this study the Na+ datawere available for PM2.5 and PM10 samples of BMRand Beijing, and for PM10 in Hanoi, during bothseasons. Therefore, for other cities the elemental Naconcentrations analyzed by other methods (than IC)were used to estimate the sea-salt contribution. Thisin turn can introduce some uncertainty as theelemental Na may also be originated from soil,which is location specific though is generallybelieved to be small. Some sea-salt contributionwas found in PM2.5, but the large difference betweensea-salt content in PM10 and in PM2.5 indicates thata larger portion of sea-salt in most cities, asexpected, was in the coarse fraction, PM10–2.5. Thesea-salt levels in PM10 were highest in Bandung, thecity surrounded by the sea, and the second highestin BMR. The reported sea-salt levels of BMR areclose to those reported for Brisbane, 0.7 mgm�3 inthe fine and 2.5 mgm�3 in the coarse fraction (GUsite, Chan et al., 1997), which may be a reflection ofthe coastal location of both cities. Besides, the sea-salt levels (in mgm�3) in PM for BMR are more orless the same in both seasons, which may be due tothe general weak large-scale wind in BMR (NE andSW monsoons) and rather strong effects of sea-landbreeze.

The highest NO3� and SO4

2� in PM2.5 and PM10

were reported for Beijing in both seasons, whichindicate a significant contribution of secondaryparticles. Higher NO3

� and SO42� content in the

dry season as compared to wet season may be partlylinked to more coal burning during winter in thiscity. The levels found in this study are comparablewith those presented by He et al. (2001). During thedry season, Hanoi also reported high nitrate andsulfate levels in PM10 (and high S in PM2.5). Theregional transport of particles to the city followingthe NE monsoon as well as the precursors emissionfrom local sources, such as SO2 from residential coalbriquettes burning, in the Hanoi MetropolitanRegion are some of the causes for this highsecondary particles mass. Relatively large amountsof NO3

� and SO42� ions are also found in the coarse

ARTICLE IN PRESSNguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–3380 3377

fraction (PM10–2.5) in all cities. For comparison,coarse particles in Brisbane have been reported toshare above 20% of total sulfate and 60% of totalnitrate in PM10 (Chan et al. 1997). Higher values ofNO3� and SO4

2� in PM2.5 were observed in the dryseason than the wet season in all the cities, whichmay partly be due to less wet removal and higheremission of the gaseous precursors. In addition, theincreased atmospheric photochemistry in the dryseason (seen in higher ozone levels, Zhang and KimOanh, 2002) would enhance conversion of thegaseous precursors to sulfate and nitrate particles.Note that for cities, which do not have sulfate data,elemental sulfur (S) is presented and also included inthe reconstructed mass.

NH4+ is present mainly in the fine fraction, as

ammonium salts, hence the difference betweenthe concentrations in PM2.5 and PM10 is small.NH4

+ was the highest in Beijing for both seasons.Note that, apart from differences in emission,the lower ambient temperature in Beijing mayreduce the volatilization loss of ammonium (nitrate)during the sampling, which results in the high NH4

+

(and NO3�) in PM. Use of appropriate sampling

methods to reduce the ammonium (nitrate) lossduring sampling should be considered for betterresults.

The trace elements group includes all elementsexcept for Na, S and those listed in the crustal groupmentioned above. The elements that belong to thisgroup vary depending on analytical methods by theNRIs but generally it includes Mg, V, Mn, Ni, Zr,Sn, Sb, Cu, Zn, As, Se, Br, Rb, and Pb. Thepresence of this group in PM2.5 suggests that somecontribution may come from combustion sources.Hanoi shows the highest values of this group, whichmay be partly due to the high number of analyzedelements by both ICP-MS and XRF.

Due to the importance of the biomass burningsource in the study region the ‘smoke’ term was alsocalculated following Chan et al. (1997). The methodassumes that the crustal matter would have theaverage ratio K/Fe of 0.6 and the excess of K, i.e.[Total K–0.6Fe], can be assigned to the biomassburning contribution (of K). It is noted that thereference ratio of K/Fe for the crustal matter maybe location/city specific, which needs further in-vestigation, but for simplicity this formula is appliedfor all cities. It is understandable that the ‘smoke’,calculated based only on the K content, is just anindicator or surrogate of the biomass burning butnot its absolute contribution to ambient PM levels

and it was not accounted for in the reconstructedmass. (The potassium content of the PM wasalready accounted for in the ‘crustal’ group.) The‘smoke’, as expected, was found predominantly inthe fine fraction (PM2.5). In the case of BMR, whereall PM2.5 and PM10–2.5 were collected simulta-neously by dichots, the, smoke, contribution toeach coarse fraction sample was close to zero. ForBeijing, due to the lack of element K data, the watersoluble K+ was used instead which may reduce the‘smoke’ term in this city. Nevertheless, the levels inBeijing are still the highest during both seasons,around 2 mgm�3 in the dry season and 1 mgm�3 inthe wet season, which are about two fold thecorresponding levels in other cities. The percentageof ‘smoke’ to the PM mass is less than 1–2% in allcities. Lower percentage of ‘smoke’ was reported forBrisbane, Australia (Chan et al. 1997), namely,0.34% of PM2.5 mass and 0% of the coarse fraction.Note that the PM concentration mass reported forBrisbane was only around 10–15% of thoserecorded in the cities in this study.

Due to its toxicity, the seasonal average of Pb inthe cities is also presented separately in Table 1,though Pb was already included in the ‘traceelements’ group for the reconstructed mass. It isnoted that different methods used for the Pbanalysis may not render the full compatibility ofthe concentration data. For PM2.5, the Pb levelsfound for Bandung, Beijing, and Hanoi are compar-able while lower levels are found in BMR. Since theanalytical method used by the Philippine NRI didnot detect Pb in SRM, the Pb results of Manilasamples are not presented in Table 1. For PM10, thehighest seasonal average Pb levels were obtained atthe upwind site in Hanoi where the maximum 24 haverage was 4.7 mgm�3 in the wet season and1.0 mgm�3 in the dry season, which are higher thanthe traffic site (maximum 24 h average of0.6–0.7 mgm�3 in both seasons). This suggeststhat the Pb found in Hanoi PM10 samples maycome also from other sources than only the traffic.It is noted that in all six cities, only Bandungstill uses leaded gasoline. Leaded gasoline hasalready been phased out in others cities for sometime, e.g. Thailand phased out leaded gasoline in1996 (Supat, 1999), Vietnam in 2001 (ADB, 2002),Metro Manila in April 2000 and the Philippinesnationwide in December 2000 (EMB, 2003), Beijingsince 1999 and China nationwide since January2001 (ABSC, 1998), India nationwide since 2000(Parivesh, 2001).

ARTICLE IN PRESSNguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–33803378

Thus, in general for all cities, the major compo-nents of PM2.5 and PM10 are OM, soot (EC/BC),SO4

2�, NO3�, and crustal. Mass enclosure is much

improved for the site/cities where all these compo-nents are available. OM and EC may come fromvarious sources such as traffic, open biomassburning, and domestic fuel combustion. In parti-cular, substantial OMmay be formed and present asthe secondary organic aerosols in the atmosphere.The ‘smoke’, estimated based on K content of PM,may provide additional indicator for the biomassburning contribution.

The PM spatial and temporal distribution,and the reconstructed mass give qualitativeinformation on major contributors to PM levels ina city. The high PM2.5/PM10 ratios in the citiesindicate the contribution of secondary particles andfuel combustion sources. The quantitative sourceapportionment results by receptor modeling forthe cities will be published in our forthcomingpapers. Note that the major contributors obtainedby receptor modeling agree with those identifiedin this paper. The preliminary results show thatthe traffic, biomass burning and secondary particles(ammonium sulfate and ammonium nitrate) arethe largest contributors to PM2.5 pollution inmost of the cities in both the dry and wet seasonswith a collective share of around 60–90% of thePM2.5 mass.

4. Summary and conclusions

The designed monitoring program has capturedthe urban PM pollution characteristics in the sixcities, which show high levels of PM2.5 and PM10,especially in the dry season. PM2.5 and PM10

concentrations fluctuated from one site to anotherwithin a city reflecting the influence of primary PMfrom sources in the vicinity of the sites. Inter citycomparison shows the highest levels of PM2.5 andPM10 in Beijing in both seasons followed by Hanoiand Chennai. Both natural and man-made factorsplay important roles in the PM build up in the citiesleading to the large differences in pollution levelsbetween them. The PM2.5 constitutes a largerfraction of PM10 in most of the cities. The majorchemical components of PM2.5 across six cities areOM, soot, NH4

+, NO3�, and SO4

2�. The crustalgroup is also significant in PM10. Analysis of thespatial and temporal variations of the PM levels, thePM2.5/PM10 ratios, and the reconstructed massprovide insight into the contributors/sources. The

major contributors to PM levels in the cities includethe traffic, secondary sulfate and nitrate particles,biomass burning, and soil dust. Contribution ofother sources such as stationary fuel combustionfor space heating, sea-salt, and industry may besignificant in some cities. Continuous update andrefinement of the PM ambient data with emphasison QA/QC are still required to improve the AIR-PET monitoring results. The outcome of this study,however, provides useful information to understandthe nature of PM pollution in the region.

Acknowledgement

We would like to thank the Swedish InternationalDevelopment Agency for the generous financialsupport without that the study would not bepossible. The California Air Resource Board(CARB) is acknowledged for donating sevendichotomous samplers to AIT for the use in thenetwork. Tom Parsons, John Higuchi, MichaelKenny, Tom Cackette, Dick Baldwin, Bill Loscut-off, Douglas Fox, and many other colleagues whohelped in the process of acquisition, preparation,and shipment of the samplers from CARB to AITare cordially thanked. Mike Kleeman and Jorn D.Herner at the UC Davis are duly acknowledged forthe generous support in EC/OC analysis for the AITsamples. The collaborating institutions, whichprovided the PM composition analysis services forrespective NRIs, are acknowledged for offering themost affordable price.

Further, we acknowledge the contribution of thefollowing researchers and students involved inthe project activities in each country, includingPrapat Pongkiatkul, Paisarn Kittiruangpol, OpalPrakobboon, Vorramaz Thammapinan, SujitraYootong and Do Thanh Canh at AIT; HaryonoHuboyo, Yulia Hendra, Yenny Ruslinda, EndahYunary, Rahmat Boedy santosa, Lilik Mulyani,Indira Rahmawati, Annissa Mayangsari, Arumwulandari, Oktri sunarni and Yandhinur at ITB;Guoshun Zhuang, Yele Sun, Wenjie Zhang, YingWang, Aohan Tang, Hui Yuan and Mo Dan atthe Beijing Normal University; Jinghua Guo, WeiZhang at RCEES; T. Swaminanthan, L. Philip,K. Ramakrishna, K. Jagannathan, Eldon Rene,K. NarayanaRao at IIT Madras; Dam Duy An,Hoang Anh Le, Le Thuy Linh, Pham Van Quan atHanoi University of Science; Pham Manh Thao atLe Qui Don Technical University; Nguyen Ba Hien,Nguyen Thanh Duy, and Vu Kiem Thuy at Hanoi

ARTICLE IN PRESSNguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–3380 3379

University of Technology; Emmanuel G. Anglo,James Bernard B. Simpas, Rikki Pamela L. PinedaGenevieve H. Lorenzo at the Manila Observatory.

Appendix A. Supplementary data

Supplementary data associated with this article,including the general information on the 6 cities,detail on PM sampling and SRM analysis results,can be found in the online version at doi:10.1016/j.atmosenv.2006.01.050.

References

ABSC (Administrative Bureau of the State Council, China),

1998. Regulation on the prohibition of the production

and sale of leaded gasoline, (GBF129-1998). Published

by the Administrative Bureau of the State Council,

Beijing, China.

Asian Development Bank, 2002. Reduction of vehicle emission

program: integrated action plan to reduce vehicle emission in

Vietnam. RETA 5937 Reducing Vehicle Emissions in Asia.

Chan, Y.C., Simpson, R.W., McTainsh, G.H., Vowles, P.D.,

Cohen, D.D., Bailey, G.M., 1997. Characterisation of

chemical species in PM2.5 and PM10 aerosols in Brisbane,

Australia. Atmospheric Environment 31, 3773–3785.

Chow, J.C., 1995. Measurements methods to determine com-

pliance with ambient air quality standards for suspended

particles. Journal of Air and Waste Management Association

45, 320–382.

Dan, M., Zhuang, G.-S., Li, X.-X., Tao, H.-R., Zhuang, Y.-H.,

2004. The characteristics of carbonaceous species and their

sources in PM2.5 in Beijing. Atmospheric Environment 38,

3443–3452.

Dockery, D.W., Pope III, C.A., 1994. Acute respiratory effects of

particulate air pollution. Annual Review of Public Health 15,

107–132.

Donaldson, K., Li, X.Y., MacNee, W., 1998. Ultrafine (nano-

metre) particle mediated lung injury. Journal of Aerosol

Science 29, 553–560.

EMB, Philippines, 2003. 2002 National Air Quality Status

Report. Environmental Management Bureau Department of

Environment and Natural Resources (DENR), Philippines

2003.

He, K., Yang, F., Ma, Y., Zhang, Q., Yao, X., Chan, C.K.,

Cadle, S., Chan, T., Mulawa, P., 2001. The characteristics of

PM2.5 in Beijing, China. Atmospheric Environment 35,

4959–4970.

Hien, P.D., Binh, N.T., Truong, Y., Ngo, N.T., Sieu, L.N., 2001.

Comparative receptor modelling study of TSP, PM2 and

PM2–10 in Ho Chi Minh City. Atmospheric Environment 35

(15), 2669–2678.

Hien, P.D., Bac, V.T., Tham, H.C., Nhan, D.D., Vinh, L.D.,

2002. Influence of meteorological conditions on PM2.5 and

PM2.5–10 concentrations during the monsoon season in

Hanoi, Vietnam. Atmospheric Environment 36 (21),

3473–3484.

Ho, K.F., Lee, S.C., Chan, C.K., Yu, J.C., Chow, J.C., Yao,

X.H., 2003. Characterization of chemical species in PM2.5

and PM10 aerosols in Hong Kong. Atmospheric Environ-

ment 37, 31–39.

IAEA, 2005. Air pollution and trends, UNDP/RCA/

IAEA Project (RAS/97/030/A/01/18). (http://www-tc.iaea.org/

tcweb/abouttc/strategy/thematic/pdf/summary/Summary_

Report_AirPollution.pdf).

IPCC, 2001. Climate change 2001: The scientific basis. Aerosols,

their direct and indirect effects. Contribution of Working

Group I to the Third Assessment Report of the Intergovern-

mental Panel on Climate Change (IPCC). Cambridge Uni-

versity Press, Cambridge, England. (Chapter 5).

Kim, Y.J., Kim, K.W., Oh, S.J., 2001. Seasonal characteristics of

haze observed by continuous visibility monitoring in the

urban atmosphere of Kwangju, Korea. Environmental

Monitoring and Assessment 70, 35–46.

Kim Oanh, N.T., Chutimon, P., Ekbordin, W., Supat, W.,

2005. Meteorological pattern classification and applica-

tion for forecasting air pollution episode potential in

a mountain-valley area. Atmospheric Environment 39,

1211–1225.

Liu, Y., Daum, P.H., 2002. Anthropogenic aerosols: indirect

warming effect from dispersion forcing. Nature 419, 580–581.

Lodge, J.P., Jr. (Ed.) 1989. Methods of Air Sampling and

Analysis, third ed. Lewis Publishers, Chelsea, MI.

Ma, C.-J., Kasahara, M., Holler, R., Kamiya, T., 2001.

Characteristics of single particles sampled in Japan during

the Asian dust-storm period. Atmospheric Environment 35,

2707–2714.

Moloi, K., Chimidza, S., Selin Lindgren, E., Viksana, A.,

Standzenieks, P., 2002. Black carbon, mass and elemental

measurements of airborne particles in the village of Serowe,

Botswana. Atmospheric Environment 36, 2447–2457.

Opal, P., 2002. Comparative study of existing monitoring

methods of PM2.5 for Bangkok conditions. AIT Master

Thesis, No. EV-02-13.

Ostro, B., Chestnut, L., Vichit-vadakan, N., Laixthai, A., 1999.

The impact of particulate matters on daily mortality in

Bangkok, Thailand. Journal of Air and Waste Management

Association 49 (special issue on PM, September 1999),

100–107.

Ostro, B., Broadwin, R., Green, S., Feng, W.-Y., Lipsett, M.,

2006. Fine particulate air pollution and mortality in nine

California counties: results from CALFINE. Environmental

Health Perspectives 114, 29–33.

Parivesh, 2001. Air Pollution and Human Health. Newsletter,

Central Pollution Control Board (India). (Archived: hhttp://

www.cpcb.nic.in/sept2001air6.htmi.

PCD, 2001, 2002, and 2003. Air Pollution and Noise Situation

and Management Annual Reports 2001, 2002, and 2003.

Air Quality and Noise Management Division, Pollution

Control Department, Ministry of Science, Technology and

Environment.

Schwartz, J., Neas, L.M., 2000. Fine particles are more strongly

associated than coarse particles with acute respiratory health

effects in schoolchildren. Epidemiology 11, 6–10.

Sun, Y., Zhuang, G., Wang, Y., Han, L., Guo, J., Dan, M.,

Zhang, W., Wang, Z., Hao, Z., 2004. The air-borne

particulate pollution in Beijing—concentration, composition,

distribution and sources. Atmospheric Environment 38,

5991–6004.

ARTICLE IN PRESSNguyen Thi Kim Oanh et al. / Atmospheric Environment 40 (2006) 3367–33803380

Supat, W., 1999. Thailand’s automotive air pollution control

strategies. Paper presented at the Federation of Engineering

Institutions of Southeast Asia and the Pacific (FEISEAP ‘99),

Kanchanab11–i. Thailand, May 18–20, 1999.

Turpin, B.J., Lim, H.-J., 2001. Species contributions to PM2.5

mass concentrations: revisiting common assumptions for

estimating organic mass. Aerosol Science and Technology

35, 602–610.

UNEP and C4, 2002. The Asian Brown Cloud: Climate and

Other Environmental Impacts. UNEP, Nairobi.

Watson, J.G., 2002. Visibility: science and regulation. Journal of

Air and Waste Management Association 52, 628–713.

Zhang, B.-N., Kim Oanh, N.T., 2002. Photochemical smog

pollution in the Bangkok Metropolitan Region of Thailand in

relation to O3 precursor concentrations and meteorological

conditions. Atmospheric Environment 36 (26), 4211–4222.