TM Spheric Ollution Esearch

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Atmospheric Pollution Research 4 (2013) 181‐190

Atm spheric Pollution Research


www.atmospolres.com

Chemical characterization and source apportionment of PM10 and


PM2.5 in the metropolitan area of Costa Rica, Central America
Jorge Herrera Murillo 1, Susana Rodriguez Roman 2, Jose Felix Rojas Marin 1, Arturo Campos Ramos 3,
Salvador Blanco Jimenez 3, Beatriz Cardenas Gonzalez 3, Darrel Gibson Baumgardner 4
1
Laboratorio de Análisis Ambiental, Escuela Ciencias Ambientales, Universidad Nacional, Heredia, Costa Rica
2
Escuela de Química, Universidad de Costa Rica, Ciudad Universitaria Rodrigo Facio, Costa Rica
3
Dirección General del Centro Nacional de Investigación y Capacitación Ambiental, Instituto Nacional de Ecología, Distrito Federal, México
4
Centro de Ciencias de la Atmósfera, Universidad Nacional Autónoma de México, Ciudad Universitaria, Distrito Federal, México

ABSTRACT
PM10 and PM2.5 levels, concentrations of major ionic components, trace elements, and organic and elemental carbon Corresponding Author:
were evaluated from samples collected in 4 sites (industrial, commercial and residential zones) located in the Jorge Herrera Murillo
metropolitan area of Costa Rica. The annual mean PM levels were higher in high traffic–commercial (HE–01) and  : +506‐22‐77‐3292
industrial (BE–02) sites, 55 g m and 52 g m for PM10 and 37 g m and 36 g m for PM2.5, respectively. The
–3 –3 –3 –3
 : +506‐22‐77‐3696
major components of PM2.5 were organic matter (OM) and elemental carbon (EC) (44.5–69.9%), and secondary ions  : jorge.herrera.murillo@una.cr
(16.1–27.2%), whereas the major components of PM10 were OM+EC (32.7–59.4%), crustal material (23.5–35.6%) and
secondary ions (11.4–26.9%). For the most of the sampling sites, PM10 and PM2.5 concentrations were lower during
the dry season and increased gradually in the rainy season due to wind patterns. PMF model identified 8 principle
sources for PM10 and PM2.5 in the industrial site (crustal, secondary sulfate, secondary nitrate, secondary organic,
traffic, sea–salt aerosols, industrial and oil combustion), 6 and 5 sources in commercial and residential sites,
respectively. The source contributions showed a clear seasonal pattern for all the sites.

Keywords: PM10, PM2.5, chemical composition, Costa Rica, source apportionment

Article History:
Received: 31 October 2012
Revised: 28 February 2013
Accepted: 28 February 2013

doi: 10.5094/APR.2013.018

1. Introduction
The growing interest aroused by atmospheric aerosols is due to
Atmospheric particulate matter (PM) is made up of solid their impact on human health, air quality and global climate change
and/or liquid particles (except pure water) of various sizes and (WHO, 2006; Solomon et al., 2007; Colette et al., 2008). A large
compositions, including mineral dust, metals, metalloids, sea number of epidemiological studies show that aerosol particles, and in
salts, ammonium nitrate and sulfate, organic compounds, particular their fine fraction, pose a risk to human health because of
elemental carbon, etc. The relative abundances of these their adverse effects both on the respiratory and cardiovascular
atmospheric components in particulate matter are temporally systems (Pope and Dockery, 2006). Aerosol particles affect air quality
and spatially highly variable. Some of them are directly emitted with potentially negative effects on visibility, agricultural and natural
into the atmosphere by both natural and anthropogenic ecosystems, material and cultural heritages (Gao et al., 2002;
sources (primary particles), while some others are the result of Lazaridis et al., 2002; Yadav and Rajamani, 2006; Yuan et al., 2006;
homogeneous or heterogeneous nucleation and condensation Yatkin and Bayram, 2008). Atmospheric aerosols also influence the
of gaseous precursors (secondary particles) (Chan and Kwok, Earth's radiation balance both directly, by scattering and absorbing
2001; Morawska and Zhang, 2002). Secondary inorganic the in–coming and out–going radiations, and indirectly, by acting as
aerosols (SIA), i.e. ammonium sulfates and nitrates, are a result cloud condensation nuclei (CCN) or ice nuclei (IN), thus changing the
of these atmospheric reactions, being commonly observed in microphysical structure, the optical properties and the precipitation
polluted air masses (Lammel and Leip, 2005). The PM can have efficiency of clouds (Kaufman and Koren, 2006; Yu et al., 2006;
different sources according to their size, for fine particles Solomon et al., 2007). However, these sole mass level determinations
(particles with aerodynamic diameter ≤2.5 μm) they can be make our understanding of the nature of several sources of
emitted primarily (such as diesel soot), or formed secondarily contamination and the processes which affect the air quality rather
from gaseous precursors by nucleation and/or condensation on incomplete. The careful analysis of the chemical composition is now
existing particles (Finlayson–Pitts and Pitts, 2000). The coarse considered of paramount importance in the chemistry of the urban
fraction is also primarily emitted by mechanical processes atmosphere.
(such as mineral dust and sea spray). Coarse secondary
particles may also be found (Raes et al., 2000), due to chemical The metropolitan area of Costa Rica is located in a central
interaction of gases with primary particles of crustal or marine plateau of around 3 000 km2 surface including in a mountain system
origin. that cross the country from NW to SE, with mountains having a

© Author(s) 2013. This work is distributed under the Creative Commons Attribution 3.0 License.
Herrera Murillo et al. – Atmospheric Pollution Research (APR) 182

maximum elevation of around 4 000 m. The metropolitan area For PM2.5 sampling, four low volume air samplers (Air Metrics)
is the highest–ranking center in the urban system in Costa Rica, were used with a flow rate of 5 liters per minute. The separation of
accounting for 75% of the vehicle fleet (approximately 734 200 the PM2.5 fraction was done at the entrance of the sample by a head
units), 65% of the domestic industry and 60% of the country's with two impactors, one (located on the top) that separates the total
population (2 580 000), according to data from the last PM10 and a second impactor that separates the fraction of PM2.5
population census conducted in 2010 (INEC, 2011). Lack of particles of the PM10.
urban planning has implicated a considerable ruin of air
quality, as a result of growth experienced by the cities of the Samples were taken during 24 h on pre–fired (at least 5 h at
metropolitan area of Costa Rica during the past 20 years. 900 °C) pre–conditioned (24 h in a dessicator under the following
conditions: temperature of 15–30 °C and humidity less than 40%) and
This paper focuses on the much–needed information on pre–weighted quartz fiber filters (Whatman CAT No. 1851–865 and
particulate matter chemical composition and their sources of Pallflex TYPE: Tissuquartz 2500 QAT–UP for PM10 and PM2.5,
formation, as well as their spatial and seasonal characteristics respectively). Before and after collection, the samples were stored in
in urban and semi–urban areas of the metropolitan area of the freezer and they were also kept frozen during transport. After
Costa Rica. This comprehensive dataset will be a key tool to particle collection, the filters were reconditioned for another 24 h in
policy makers in providing vital information for designing an air conditioned room and subsequently analyzed for total mass.
effective particulate matter control strategies. After re–weighing, the exposed filters were stored in a freezer at
–5 °C to limit losses of volatile components.
2. Experimental
All procedures during handling of filters were strictly quality
2.1. Sampling controlled to avoid any possible contamination.

For the PM10 and PM2.5 sampling, four monitoring sites 2.2. Chemical analysis
were selected (Figure 1). The sites were representative of
commercial, industrial and residential areas, all located in the Samples collected in quartz filters were used for gravimetric
Costa Rica metropolitan area (Table 1). Sampling campaign was analysis in order to determine the PM10/PM2.5 concentrations. The
conducted between June 2010 and July 2011. Simultaneous weighing of the low volume filters was performed using a semi–
samples were collected once every six days. To collect PM10 microanalytic balance (Mettler). The readability of the balance is
samples, four high volume air samplers, Thermo Andersen with 10 g with a precision of 40 g corresponding to mass concen‐
MFC (Mass Flow Controller) were used with a flow rate of tration uncertainty of 0.86 g m–3 for PM2.5 samples. High volume
(1.13±10%) m3 min–1. Flow calibration of each sampler was filters used in PM10 collection were weighted in an analytical balance.
performed comparing the readings of pressure drop generated The mass concentration uncertainty of PM10 samples was
by the flow passing through a calibrated critical orifice (TISCH 1.22 g m–3.
VARIFLO Model 454) with the sampler pressure drop reading.

Table 1. Description of sampling sites


Site Sampling Site Type Municipality Location
SJ‐03 Commercial San Jose National Electrical Company (CNFL) Office
HE‐01 High Traffic‐Commercial Heredia National University Central Administration Building
BE‐02 Industrial Belen Intermodal Company Office
MO‐01 Residential Moravia EATON Company Office

Figure 1. Sampling sites for PM10/PM2.5 in the metropolitan area of Costa Rica.
Herrera Murillo et al. – Atmospheric Pollution Research (APR) 183

A small portion of quartz filters were analyzed for organic filters. An overall bias between –8 and 13% was obtained for metal
carbon (OC) and elemental carbon (EC) using DRI Model 2001 concentrations measured in samples.
Thermal/Optical Carbon Analyzer (Atmoslytic Inc., Calabasas, CA,
USA). A 0.68 cm2 punch from each filter was analyzed for eight Finally, contents of Si and CO32– were determined indirectly
carbon fractions following the IMPROVE TOR protocol. This from the contents of Al, Ca and Mg, on the basis of prior
produced four OC fractions (OC1, OC2, OC3, and OC4 at 120, 250, experimental equations (1.89 Al = Al2O3, 3 Al2O3 = SiO2; 1.5 Ca + 2.5
2–
450, and 550 °C, respectively, in a helium atmosphere), a pyrolyzed Mg = CO3 ) (Marcazzan et al., 2001).
carbon fraction (OP, determined when reflected laser light attained
its original intensity after oxygen was added to the combustion 2.3. Positive matrix factorization (PMF)
atmosphere), and three EC fractions (EC1, EC2, and EC3 at 550,
700, and 800 °C, respectively, in a 2% oxygen and 98% helium The PMF model was used here to identify PM10 and PM2.5
atmosphere). IMPROVE OC is operationally defined as OC1+OC2+ sources and estimate source contributions. It is a multivariate
OC3+OC4+OP and EC is defined as EC1+EC2+EC3–OP. For the OC receptor model that has been described in detail by Paatero (1997)
and EC determination, the analyzer was calibrated using different and implemented in the PMF2 program, which doesn’t require
aliquots (0, 3, 5, 7, 10, 12, 15, 20 and 25 l) of a standard sucrose source profile knowledge unlike traditional source receptor models
–1
solution (4 260 mg L ) over a filter blank (pre–heated quartz filter like Chemical Mass Balance (CMB). This program is now being
punch). The limits of detection (LODs) for OC and EC were widely used to analyze airborne particulate matter sources (Kim et
746 ng m–3 and 180 ng m–3, respectively. Analytical uncertainties al., 2003; Hien et al., 2004; Mazzei et al., 2008).
for OC and EC were estimated to be 16% and 9%, respectively. NIST
8785 reference material was used in order to evaluate the PMF is a receptor model based on the principle that a
analytical method accuracy for the determination of organic and relationship between sources and receptor exists when mass
elemental carbon in PM10/PM2.5. This reference material consists of conservation can be assumed (Paatero and Hopke, 2003). In this
a thin fraction of SRM 1649 (Urban Dust) deposited on a quartz case, and when chemical speciation of ambient PM is available, a
fiber filter. Seven replicates of NIST 8785 were analyzed. An overall mass balance equation of the following form can be written:
bias between –3.7 and 5% was obtained. There was no significant
difference between the means of the obtained values and the p
reference values, according to a t–test with a 95% confidence level. x ij   g ih fhj  eij (1)
h 1
Another filter portion was extracted in 50 mL deionized water
during 35 minutes in an ultrasonic bath. The analysis of ionic th th
species was performed by dual microbore suppressed Ion where xij is the j species concentration measured in i sample, gih
Chromatography using a DIONEX ICS–3000 equipment with a is the particulate mass concentration from the hth source
quaternary pump. A fresh calibration curve was prepared for every contributing to the ith sample, fhj is the jth species concentration
th
20 samples, together with a dissolution of quality control of measured in the i sample, and p is the total number of
–1
5 mg L prepared from a certified DIONEX synthetic sample. independent sources. The corresponding matrix equation is:
Detection limits for each ion are shown in Table 2.
X=GF+E (2)
Table 2. Limit of detection (LOD) for different chemical species analyzed in
PM2.5 and PM10 where X is a n×m matrix with n measurements and m elements. E
is an n×m matrix of residuals. G is n×p source contribution matrix
Chemical Specie LOD Chemical Specie LOD with p sources, and F is a p×m source profile matrix.
‐3 ‐3
Metals (ng m ) Inorganic Ions (g m )
‐ The signal–to–noise ratio (SNR) was used to select the species
Na 125 F 0.01

for further analysis. This ratio is defined as:
K 57 Cl 0.05

Ca 34 Br 0.02 n
 ( xij  sij )
‐ 2
Mg 17 NO2 0.02
Cu 0.78 NO

0.01  S   i 1 (3)
N j
3
n
2
 sij
3‐
Fe 0.54 PO 4 0.06
Mn 0.43 SO4
2‐
0.03 i 1
+
Al 0.88 NH 4 0.01
Species with S/N below 0.2 were classified as bad values and
Cr 0.97
were thus excluded from further analysis. The application of PMF
Ni 0.51 depends on the estimated uncertainties for each of the data
V 0.14 values. The uncertainty estimation provides a useful tool to
Pb 0.35 decrease the weight of missing and below detection limit data in
the solution. The procedure of Polissar et al. (2001) was used to
assign measured data and the associated uncertainties as the input
Other filter portion was extracted by adding 5 mL of ultra– data to the PMF. The concentration values were used for the
pure concentrated nitric acid and 25 mL of deionized water and measured data, and the sum of the analytical uncertainty and 1/3
heated on a hot plate until almost dryness. The remaining solution of the detection limit value was used as the overall uncertainty
was poured into a 25 mL volumetric flask. A second extraction was assigned to each measured value. Values below the detection limit
done with 1 mL of concentrated HClO4. Analyses of metals were were replaced by half of the detection limit values and their overall
made using atomic absorption spectrometry (PERKIN ELMER uncertainties were set at 5/6 of the detection limit values. Missing
AANALYST 700) with a graphite furnace. Detection limits in ng m–3, values were replaced by the geometric mean of the measured
using the IUPAC method, were obtained. The results are shown in values and their accompanying uncertainties were set at four times
Table 2. Blank filters were analyzed for metals and inorganic ions, of this geometric mean value. In addition, the estimated
obtaining lower concentrations of 5% of those found in samples. uncertainties of species that have scaled residuals larger than 72
The accuracy of the metal chemical analysis was periodically need to be increased to reduce their weight in the solution
checked using a certified standard (SRM 1648) spiked on blank (Paatero, 2000; Hopke and Paatero, 2002).
Herrera Murillo et al. – Atmospheric Pollution Research (APR) 184

3. Results and Discussion to the contribution of high levels of coarse particles generated by
local industrial activities. The PM2.5/PM10 ratio values were 0.70,
3.1. PM levels and chemical composition 0.72, 0.67 and 0.65 for SJ–03, MO–01, HE–01 and BE–02,
respectively. More than 90% of sulfate particles are fine.
Table 3 shows a statistical analysis of the PM component Additionally, trace elements like V and Ni, have a predominantly
levels. The annual mean PM levels were higher in high traffic fine granulometry especially in commercial and industrial sampling
commercial (HE–01) and industrial (BE–02) sites, 55 g m–3 and sites as these metals are emitted mostly by anthropogenic
52 g m–3 for PM10 and 37 g m–3 and 36 g m–3 for PM2.5, activities such as traffic, industrial combustion, etc. Values lower
respectively. These two sites showed values that exceeded the than 0.50 were obtained in the Al, Ca and other major components
Costa Rican Air Quality Annual Standard of 50 g m–3 for PM10. associated with crustal material.
Residential site showed the lower values, 25 g m–3 and 18 g m–3,
for PM10 and PM2.5 respectively. It was found that the PM2.5 An aerosol mass reconstruction (AMR) analysis was conducted
–3
concentration in all the sampling sites was higher than 15 μg m , to explore the relative contributions of the measured inorganic and
which is the annual goal of the USA National Ambient Air Quality organic chemical species and their relationship to the total
Standard. Daily concentrations exhibited 34% and 26% exceedan‐ measured gravimetric mass. For the purpose of chemical mass
ces of the 24 h limit of the USA PM2.5 National Ambient Air Quality reconstruction, chemical components were grouped into six
Standard (35 μg m–3) at the high traffic commercial (HE–01) and categories: crustal materials (CM), trace elements (TE), organic
the industrial site (BE–02), respectively; indicating the necessity for matter (OM), elemental carbon (EC), sea salt (SS), and secondary
local authorities to implement appropriate measures for PM ions (SI). CM represents the sum of typical crustal materials,
emissions reduction. The comparison was done with USA including Al, K, Fe, Ca, Mg, Ti and Si. Each of these species were
standards as Costa Rica doesn’t have an air quality standard for multiplied by the appropriate factors to account for their common
PM2.5. oxides based on the following equation (Marcazzan et al., 2001;
Hueglin et al., 2005):
The two fractions of particulate matter are well correlated
over the whole campaign, especially in the residential site CM=1.89 Al+1.21 K+1.40 Ca+1.66 Mg+2.14 Si+1.43 Fe (4)
(R2=0.85). Lower correlation coefficients were obtained in
commercial (R2=0.76) and industrial (R2=0.70) sites, probably, due

Table 3. Annual average PM10 and PM2.5 levels and their chemical compositions at the metropolitan area of Costa Rica

SJ‐03 MO‐01 HE‐01 BE‐02


PM10 PM2.5 PM10 PM2.5 PM10 PM2.5 PM10 PM2.5
N 67 64 55 61 64 59 58 62
Inorganic Ions (g m )
‐3

PM 37±9 26±7 25±7 18±5 55±15 37±12 52±18 36±15



F 0.08±0.07 0.04±0.06 0.08±0.04 0.05±0.03 0.08±0.02 0.06±0.03 0.07±0.03 0.08±0.02

Cl 1.17±0.86 0.52±0.51 1.08±0.65 0.51±0.47 0.98±0.60 0.46±0.42 1.01±0.62 0.57±0.77

NO 2 0.20±0.06 0.15±0.04 0.20±0.03 0.14±0.05 0.27±0.08 0.18±0.06 0.21±0.05 0.14±0.04

NO3 0.86±0.34 0.63±0.15 0.92±0.32 0.58±0.27 1.13±0.41 0.74±0.39 1.02±0.36 0.60±0.29
3‐
PO 4 0.67±0.55 0.31±0.28 0.71±0.57 0.24±0.34 0.54±0.17 0.40±0.15 0.41±0.17 0.53±0.12
2‐
SO4 3.23±1.68 3.34±1.23 3.45±2.17 3.15±1.66 3.87±1.65 3.52±1.38 3.97±1.80 3.81±1.62
+
NH4 1.51±0.84 1.59±0.57 1.34±0.81 1.46±0.49 1.62±0.74 1.71±0.61 1.48±0.73 1.55±0.47
OC 6.67±2.85 6.73±2.17 4.27±2.56 4.45±2.37 14.67±8.44 13.86±6.57 13.34±4.92 12.55±5.12
EC 4.23±2.38 4.5±2.24 1.36±0.70 1.37±0.85 4.82±1.42 3.74±1.25 3.81±1.69 3.55±1.83
‐3
Metals (ng m )
V 3.7±1.6 2.9±1.1 1.2±0.6 0.9±0.7 2.8±1.2 2.1±0.9 5.4±3.6 3.6±2.4
Pb 11.8±5.3 7.6±3.7 8.6±3.0 7.29±2.17 11.5±3.9 8.75±1.77 11.2±6.8 9.9±4.7
Cr 11.5±4.5 7.0±2.6 5.7±1.6 5.1±1.9 4.9±1.3 6.7±1.1 8.7±4.4 7.9±2.5
Cu 47±28 43±20 86±38 57±41 84±54 62±43 146±65 96±51
Ni 4.6±1.4 3.86±0.95 3.4±2.3 2.94±1.81 2.1±0.8 1.74±0.55 7.0±3.6 5.9±2.4
Mn 58±32 61±29 84±39 44±17 139±71 67±36 136±24 58±17
Al 670±106 259±174 575±117 278±71 757±87 274±61 742±277 353±191
Fe 361±150 257±162 189±58 141±25 431±203 191±85 553±247 212±108
Ca 283±76 208±83 351±74 184±68 279±107 152±90 367±223 232±175
Mg 139±48 47±28 112±42 41±27 102±52 37±29 154±68 55±44
Na 1 440±425 615±317 1 291±805 554±392 1 306±567 677±357 1 394±376 665±228
K 236±104 125±91 123±39 81±45 124±45 65±32 231±102 117±81
Si 3 798±402 1 467±375 3 260±298 1 577±387 4 295±411 1 554±543 4 208±627 2 002±492
Herrera Murillo et al. – Atmospheric Pollution Research (APR) 185

The Ca and Mg oxides were calculated using the non–sea salt OM+EC (44.5–69.9%), and secondary ions (16.1–27.2%). The
(nss) portion of Ca and Mg. OM was obtained by multiplying the crustal contribution to PM2.5 mass increased from 23–25% in
measured concentration of organic carbon (OC) by a factor of 1.6, industrial and commercial areas to 36% in residential areas which
which is based on an average of the recommended ratios of can be due to the presence of nearby non–urban lands, since these
1.6±0.2 for urban aerosols and 2.1±0.2 for aged or non–urban regions are subjected to wind resuspension (or entrainment)
aerosols (Turpin and Lim, 2001). This factor is commonly used to processes. The resuspension of dust (crustal materials of diameter
estimate the unmeasured hydrogen and oxygen in organic <20 μm) by wind provides a potential source of particles to the
compounds. The sea salt contribution represents particles in the atmosphere (Loosmore and Hunt, 2000).
form of fresh sea salt. It is calculated as the sum of measured
chloride ion concentration plus the sea salt fraction of The contribution of marine aerosol varies between 4 to 7% for
concentrations of Na+, Mg2+, K+, Ca2+, SO42− based on the sampling sites presenting a fairly regular basis. This evidences that
composition of seawater and ignoring atmospheric transforma‐ the contribution of this component is due to a regional scale
tions (Seinfeld and Pandis, 2006): phenomena. The levels of organic matter and elemental carbon
(OM+EC) are higher in HE–01 and BE–02 and representing between
SS=Cl–+ssNa++ssMg2++ssK++ssCa2++ssSO42– (5) 65 and 69% of the total mass of particles PM2.5. These results show
the importance of the contribution of combustion processes to the
where ssNa+=0.556 Cl−, ssMg2+=0.12 ssNa+, ssK+=0.036 ssNa+, composition of fine particles. The trace metals represent the
ssCa2+=0.038 ssNa+, and ssSO42−=0.252 ssNa+ (Terzi et al., 2010). smallest contribution for all monitoring sites.
The EC contribution was reported as measured by thermal optical
reflectance (TOR). The SI contribution was calculated as the sum of 3.2. Temporal variations in PM10 and PM2.5 chemical composition
nss–SO42−, NH4+, and NO3−, where nss–SO42− is total measured SO42−
minus the sea salt fraction of SO42−. Trace elements were also Two different behaviors were found in the time series analysis
converted to their common oxides. Trace elements only represent for PM10 and PM2.5 in the Metropolitan Area of Costa Rica. For
a small percentage of the total PM mass, however they were also MO–01, HE–01 and SJ–03, PM10 and PM2.5 concentrations were
added to the analysis because they have a great environmental lower during the dry season (January–April) and increased
importance due to their toxicity and anthropogenic origin. TE gradually in the rainy season (Figure 3). This situation can be
contribution was calculated using the following equation explained as during the dry season, the average wind speed in the
(Marcazzan et al., 2001): Metropolitan Area increases as a result of the influence of Alisios
winds movement from the Caribean Sea favoring the dispersion of
TE=1.47 V+1.29 Mn+1.27 Ni+1.25 Cu+1.08 Pb+1.31 Cr (6) pollutants. During the rainy season, Alisios winds reduce their
intensity and allow the entry of Pacific breezes (generating rains in
The coefficients in Equation (6) represent a gravimetric factor the zone) reducing the wind speed by 35%. This situation
to obtain the corresponding oxides. The gravimetric and decreases the pollutant dispersion in the metropolitan area. For
reconstructed mass concentrations show a generally strong BE–02, higher concentrations were reported in dry season since
correlation, with a coefficient of determination (R2) range between this sampling point is located in the wind exit of the metropolitan
0.65–0.74, indicating overall good agreement between the recon‐ area. As a result of this situation if pollutant dispersion decreases
structed mass and the gravimetric mass. The major components of in the metropolitan area, the concentrations of particulate matter
PM10 (Figure 2) in the four sampling sites were organic matter would be enhanced in BE–02. This explanation is in accordance
(OM) (27.3–48.6%), elemental carbon (EC) (5.4–12.8%) and crustal with the correlation coefficient observed between HE–01 and BE–
material (23.5–35.6%). The major components of PM2.5 were 02 (r=–0.768).

Figure 2. Chemical mass closure of PM10 and PM2.5 at high traffic commercial and industrial sampling sites in the
metropolitan area of Costa Rica.
Herrera Murillo et al. – Atmospheric Pollution Research (APR) 186

Figure 3. Temporal variation of PM10 and PM2.5 concentrations at BE‐02 and HE‐01 sampling sites in the
metropolitan area of Costa Rica, January‐July 2011.

Figure 4 shows the temporal variation of major components in increase the contribution of resuspension process to the chemical
PM10 and PM2.5 for SJ–03 sampling site. Because the variations composition of particles.
were similar to the other sites, the results for these sampling
stations are not plotted. Carbonaceous particle concentration Sulfate concentrations showed a clear seasonal trend. The
(organic matter and elemental carbon), both coarse and fine mean concentrations in dry season were slightly higher than in
fractions showed a clear seasonal pattern. The OC and EC in PM10 rainy season because of the stronger solar radiation in dry season,
have showed lower concentrations in dry season (December–April) which promotes the oxidation of SO2 to SO42−. On the other hand,
and higher values in rainy season (May–November). By comparing nitrate levels were characterized by a clear maximum in rainy
the results for dry and rainy seasons, at the same sampling sites, it season and a minimum in dry season, both for PM10 and PM2.5
is evident that the average OC and EC concentrations for PM2.5 and fractions. However, sulfates showed the opposite pattern. This is
PM10 during the rainy season were around 1.3–2.4 times higher probably due to the low thermal stability of NH4NO3 in dry season
than dry season. This can be explained by a reduction in the when, under warm conditions, the formation of gaseous HNO3 is
pollution dispersion in the metropolitan area of Costa Rica in rainy favored. The concentration of sulfate in coarse and fine fraction
season. In the same way, temperatures during rainy season are was similar, which indicated that SO42− was mostly present as
lower than dry season helping the condensation process that ammonium sulfate (Harrison et al., 2003). In the case of nitrate,
produces secondary organic aerosols. the ratio PM2.5/PM10 was not as high as expected, may be due to
the presence of NaNO3 and Ca(NO3)2 in the coarse fraction (Yin and
The crustal matter and sea salt concentrations presented their Harrison, 2008). Ammonium levels at PM10 exhibited a clear
highest values in dry season due to the influence of Alisios winds minimum in dry season and the fine fraction showed no seasonal
that have speeds around 30 km h–1 favoring the movement of air variation.
mass from Caribean Sea. The occurrence of high speed winds can
Herrera Murillo et al. – Atmospheric Pollution Research (APR) 187

3.3. Organic and elemental carbon 3.4. Sources contribution

The origin of OC and EC can be evaluated by the relationship The sampling sites for PM10 and PM2.5 were divided into three
between OC and EC (Turpin and Huntzicker, 1995; Chow et al., categories (industrial, commercial and residential) in the PMF
1996). OC–EC strong correlations (r) of 0.64–0.77 and 0.78–0.88 analysis. The respective error estimates of masses, OC, EC, ionic
were observed for PM10 and PM2.5 respectively, in all the sampling compounds and elements were computed, and incorporated in the
sites. These correlations indicated the presence of common PMF model which was applied for each category separately.
dominant sources for OC and EC (e.g. industrial combustion, motor
vehicle exhaust) because the relative rates of EC and OC emissions For the industrial sampling site (BE–02), eight sources were
would be proportional to each other. OC and EC had their identified as a result of PMF analysis. However, in commercial sites
strongest correlation in rainy season (0.78–0.84) implying that the (SJ–03 and HE–01), six sources including: secondary sulfate,
majority of OC was primary and secondary organic compounds secondary nitrate, secondary organic, crustal, traffic and sea salt
(SOC) formation might be minor. Lower values during dry season were identified successfully. A good correlation (r=0.837 and 0.852
(0.69–0.72) can be due to a possible contribution of particles for PM10 and PM2.5, respectively) was obtained between predicted
produced by resuspension process and the variations in SOC versus measured PM10 mass concentration showed that the model
formation. The SOC is estimated by means of the following adequately reproduced the observed mass. In the residential zone
equation (Chow et al., 1996): sampling site (MO–01), only five sources were successfully
identified: crustal, sea salt, traffic, secondary sulfate and secondary
 OC  nitrate.
OC sec  OCtot  EC   (7)
 EC  prim The first source in Table 4 was identified as vehicle exhaust
based upon the abundances of EC, OC, SO42– and certain amount of
where OCsec is the secondary OC, and OCtot is the measured total Fe. Both diesel and gasoline powered vehicles generate large
OC. The primary organic carbon (POC) could be calculated from the amounts of carbonaceous compounds but the ratios of OC to EC in
formula EC(OC/EC)prim. However, the primary ratio of OC/EC is the emission profile can be quite different. Watson et al. (1994)
usually not available because it is affected by many factors such as noted that the OC/EC mass concentration ratio was 2.2 for gasoline
the type of emission source as well as its variation in temporal and fueled vehicle exhaust, 1.2 for diesel–fueled vehicle exhaust, and
spatial scales, ambient temperature, and carbon determination 1.1 for a mixture of vehicle types in roadside test. The common
method, etc. In many cases, (OC/EC)prim was represented by the source attributed to these metals is resuspension of road dust due
observed minimum ratio [(OC/EC)min], and assumptions regarding to traffic activity, mainly wear and tear of tires, brake wear and oil
the use of this procedure were discussed in detail by Castro et al. burning (Kulshrestha et al., 2009). The mean contributions to PM10
(1999). and PM2.5 by traffic at residential site (MO–01), were calculated to
be 7.1 and 4.7 μg m–3 respectively and at high traffic commercial
–3
The annual average concentrations of estimated SOC in the site (HE–01) these were estimated to be 10.2 and 5.1 μg m .
metropolitan area of Costa Rica contribute with 48% and 56% of Lower contribution of vehicle exhaust was obtained for the
the OC in PM10 and PM2.5, respectively. Compared with rainy industrial site (BE–02) (Table 5).
season results, there is an overall trend towards lower SOC levels
but with a higher percentage of SOC in the TOC at each site during The second source identified by PMF was distinguished by
the dry season. Higher temperatures and more intense solar large contribution values for Ni and V, which are good indicators
radiation during the dry season summer months provide favorable for residual oil combustion (Chow and Watson, 2002). The
conditions for photochemical activity and SOC production. contribution of these sources represents 7.5 and 11.8% of the PM10
and PM2.5 mass concentration at BE–02.

Figure 4. Temporal variation in PM10 and PM2.5 chemical mass closure for SJ‐03 sampling site.
Herrera Murillo et al. – Atmospheric Pollution Research (APR) 188

Table 4. Profiles of particulate sources resolved by PMF analysis for the industrial site BE‐02. Values listed here correspond
to mass percentage of each component
Vehicle Residual Sea Crustal Secondary Secondary Secondary
Sources Industrial
Emission Oil Salt Material Organics Sulfate Nitrate
Species PM2.5
2‐
SO4 2.91 15.11 0.94 10.39 12.72 60.2 1.37 0.38

NO3 0.09 0.22 0.55 9.22 0 0.11 33.1 0.27
+
NH4 0 0.93 0 0.77 0 15.33 8.17 0.11
OC 53.9 28.84 0 0.12 83.45 16.28 48.5 42.7
EC 39.7 43.92 0.86 1.18 0 7.15 5.11 9.2
V 0.28 2.07 0 0 0 0.05 0.1 9.9
Pb 0.05 0.33 0 0 0.33 0 0.04 20.5
Ni 0 2.39 0 0 0 0 0 0
Na 0 0.01 34.1 1.25 0 0.03 0 0
Cl 0.03 0 55.27 0 0 0 1.07 0
Mn 0.05 0 0.05 3.79 0 0 0 0
Al 0 0.02 0.01 12.56 0.87 0 0 0
Ca 0.17 0 3.11 21.77 1.09 0 0.97 0
Mg 0 0 2.96 15.02 0 0.02 1.16 0
K 0 2.85 1.08 8.17 0 0.12 0 0
Fe 1.55 0.03 0 16.43 0.87 0.07 0 7.3
Cu 0 0.07 0 0.11 0 0.14 0 14.10
PM10
2‐
SO4 3.04 14.73 1.1 5.53 11.85 51.4 2.04 0.07

NO3 1.28 1.45 0.91 4.01 0.47 0.85 14.6 0.35
+
NH4 1.11 2.03 0.11 0.33 0.9 20.6 5.46 0.10
OC 45.7 25.1 0 0.54 72.4 9.69 53.8 35.5
EC 30.8 37.6 0.44 0.74 3.11 2.44 3.21 6.7
V 0.13 2.49 0 0 0.09 0 0.06 7.1
Pb 0.55 2.08 0 0 0.04 0 0 13.9
Ni 0 3.42 0 0 0 0 0 0
Na 0 0 38.1 3.93 0 2.09 6.89 0
Cl 0 0 51.9 1.87 0 0 0.4 0
Mn 1.74 0.31 0.14 1.45 0 1.15 1.28 0
Al 1.38 0 0.07 15.8 3.21 0 0 0
Ca 2.67 0 3.73 20.3 0 4.65 5.1 0
Mg 2.09 0.08 3.46 18.4 1.28 0 1.07 0
K 0 4.65 0.07 8.61 2.57 3.46 4.92 12.5
Fe 6.57 0.44 0 18.8 0 1.59 0.45 8.4
Cu 0.88 0 1.02 0 0 0.77 0 17.5

Table 5. Contributions of each source identified with PMF in each category of sampling sites
Oil Sea Crustal Secondary Secondary Secondary
g m
‐3
Traffic Industrial
Combustion Salt Material Organic Sulfate Nitrate
4.4 3.8 5.22 10.45 11.81 7.5 5.2 2.4
PM10
(8.7) (7.5) (10.3) (20.6) (23.3) (14.8) (10.2) (4.7)
BE‐02
2.7 4.1 1.94 3.4 9.27 5.67 4.09 3.5
PM2.5
(7.8) (11.8) (5.6) (9.8) (26.7) (16.4) (11.8) (10.1)
10.23 6.45 12.87 8.34 8.1 6.42
PM10
(19.5) (12.3) (24.6) (15.9) (15.5) (12.2)
HE‐01
5.12 2.11 4.54 9.17 6.65 5.32
PM2.5
(15.6) (6.4) (13.8) (27.9) (20.2) (16.2)
8.28 4.79 7.63 7.21 4.63 3.87
PM10
(22.7) (13.2) (21.0) (19.8) (12.7) (10.6)
SJ‐03
4.04 1.7 2.92 5.83 4.29 3.23
PM2.5
(18.4) (7.7) (13.3) (26.5) (19.5) (14.7)
7.11 3.42 6.71 3.85 2.54
PM10 0
(30.1) (14.5) (28.4) (16.3) (10.7)
MO‐01
4.73 1.85 3.55 3.92 2.33
PM2.5 0
(28.9) (11.3) (21.7) (23.9) (14.2)
Herrera Murillo et al. – Atmospheric Pollution Research (APR) 189

The third source was characterized with large amounts of Cl–, of PM speciation obtained at the metropolitan area of Costa Rica
Na and Mg, signature of fresh sea salt. These three species, in the show that the PM10 fraction mainly consists of carbon particles
four sampling sites, accounted for (51.9–55.3)%, (34.5–38.1)% and (32.7–59.4%), mineral dust (23.5–35.6%), sea spray (5–8%), and
2.9–3.5% of the total mass of this aerosol source, respectively, secondary inorganic aerosols (mostly Na, Ca, K, Mg sulfate and
which is in close agreement with their corresponding percentage nitrate, 20–25%). The PM2.5 fraction mainly comprises OM+EC
composition in seawater. The contribution of this source decreases (44.5–69.9%) with an important fraction of secondary inorganic
in average from 10.3% in dry season to 6.1% in rainy season at all aerosols (mostly ammonium nitrate and sulfate, 31%) and very
the sampling sites, due to wind speed reduction of the Alisios air minor proportions of mineral dust and sea spray.
mass coming from the Caribean Sea.
Average PM2.5 and PM10 OC and EC concentrations during the
The fourth source has large contributions from Al, Ca, Fe and rainy season were around 1.3–2.4 times higher than dry season.
Mg, suggesting crustal or dust aerosols. This is most likely related This can be explained by a reduction in the pollution dispersion in
to exposed soil, unpaved roads and construction activities, as well the Metropolitan area of Costa Rica during the rainy season. The
as other crustal materials. The airborne soil factor accounted for annual average concentrations of estimated SOC in the
20.6% and 9.8% of the total PM10 and PM2.5 mass concentration in Metropolitan Area of Costa Rica contribute with 48% and 56% of
BE–02. Higher contributions were found in the residential site with the OC in PM10 and PM2.5 respectively. There is an overall trend
28.4% and 21.7% for PM10 and PM2.5. Similar to the sea salt, the towards lower SOC levels but with a higher percentage of SOC in
crustal material contribution showed an average reduction of 46% the TOC at each site during the dry season.
in rainy season for all the sampling sites.
In the same way, crustal matter and sea salt concentrations
The fifth source profile was related with secondary organic presented their highest values in dry season due to the influence of
aerosols and includes OC and sulfate. The OC presented in the Alisios winds that favoring the movement of air masses from
source profile, apart from the gas–to–particle conversion of Caribean Sea.
organic gases with relatively low vapor pressure and the chemical
processes in cloud and fog droplets provided an alternate PMF model identified the principal sources of PM10 and PM2.5
explanation that once the secondary inorganic aerosol is formed, in each site. Residential site showed a higher traffic and crustal
the acidic surface could catalyze heterogeneous reactions of contribution than the industrial site. The PM mass concentrations
organic carbonyl species, leading to potentially multifold increases in BE–02 site are influenced by two additional sources: oil
in secondary organic aerosol mass (Jang et al., 2002). This source combustion and industrial activities.
–3
contributes with 11.8 and 9.3 μg m for PM10 and PM2.5 concentra‐
tions. Lower contributions were obtained for commercial sites may Acknowledgments
be due to a reduction in volatile organic compound emissions in
these sites. This study is part of a Binational Cooperation Project
developed by the National University in Costa Rica and the
The source 6 assigned to secondary sulfate is traced by sulfate National Ecology Institute in Mexico. Collaboration was also
(which accounts for about 60% of the PM2.5 mass in the factor provided by both, Environmental and Health Secretary of Costa
profile) and ammonium. OC is also a major contributor (16%) but it Rica.
has been found also in other factors. This factor showed high
correlation with sulfate (r=0.71) and moderate correlation with Supporting Material Available
ammonium (r=0.57) for PM2.5 samples. Secondary sulfate was the
largest contributor to PM2.5 (24–19%), especially for commercial The Supporting Material contains results of SRM 8785
and residential sites. This factor is often considered as a fingerprint reference material analysis for organic carbon and elemental
being representative of secondary sulfate (Kim et al., 2004). carbon percentages (Table S1a), Mean OC, EC and TC
concentrations obtained at the MACR (Table S1b), Source
The source 7 was determined as secondary nitrate, since contribution in each sampling site for PM2.5 samples according to
nearly the 82% of NO3− and OC existed in this factor. The calculated PMF analysis results (Figure S1). This information is available free
contributions of secondary nitrate to PM2.5 at BE–02 in dry and of charge via the Internet at http://www.atmospolres.com.
–3
rainy season were 0.85 and 1.47 μg m respectively. This temporal
behavior can be attributed to the higher stability of ammonium References
nitrate in warm seasons (Mariani and De Mello, 2007). The
observed ratio of NO3−/NH4+ for source profile in BE–02 (4.1) is Cabada, J.C., Pandis, S.N., Robinson, A.L., 2002. Sources of atmospheric
higher than the stoichiometric ratio of NO3−/NH4+ for NH4NO3 carbonaceous particulate matter in Pittsburgh, Pennsylvania. Journal of
(3.44). This result can be explained through the formation of the Air & Waste Management Association 52, 732‐741.
NaNO3 and Ca(NO3)2, especially in dry season. Secondary nitrate Castro, L.M., Pio, C.A., Harrison, R.M., Smith, D.J.T., 1999. Carbonaceous
accounted for 10–12% and 12–16% of total mass in PM10 and PM2.5 aerosol in urban and rural European atmospheres: estimation of
respectively. secondary organic carbon concentrations. Atmospheric Environment
33, 2771‐2781.
The last source has significant values for Cu, K, Pb and OC. Chan, L.Y., Kwok, W.S., 2001. Roadside suspended particulates at heavily
These elements are often emitted from a variety of combustion trafficked urban sites of Hong Kong ‐ seasonal variation and
processes such as heavy fuels. In addition, the presence of K in this dependence on meteorological conditions. Atmospheric Environment
mode shows that wood/biomass burning (Cabada et al., 2002) also 35, 3177‐3182.
contributed to this mode. This source contributes with 4.7 and
–3 Chow, J.C., Watson, J.G., 2002. Review of PM2.5 and PM10 apportionment
10.1 μg m of PM10 and PM2.5 mass concentration at industrial site
for fossil fuel combustion and other sources by the chemical mass
BE–02.
balance receptor model. Energy & Fuels 16, 222‐260.
4. Conclusions Chow, J.C., Watson, J.G., Lu, Z.Q., Lowenthal, D.H., Frazier, C.A., Solomon,
P.A., Thuillier, R.H., Magliano, K., 1996. Descriptive analysis of PM2.5
The mean annual PM10 and PM2.5 concentrations in and PM10 at regionally representative locations during
commercial and industrial sites were higher than Air Quality SJVAQS/AUSPEX. Atmospheric Environment 30, 2079‐2112.
Standards established by Costa Rican and USA regulations. Results
Herrera Murillo et al. – Atmospheric Pollution Research (APR) 190

Colette, A., Menut, L., Haeffelin, M., Morille, Y., 2008. Impact of the Morawska, L., Zhang, J.F., 2002. Combustion sources of particles. 1. health
transport of aerosols from the free troposphere towards the boundary relevance and source signatures. Chemosphere 49, 1045‐1058.
layer on the air quality in the Paris area. Atmospheric Environment 42, Paatero, P., Hopke, P.K., 2003. Discarding or downweighting high‐noise
390‐402. variables in factor analytic models. Analytica Chimica Acta 490, 277‐
Finlayson‐Pitts, B.J., Pitts, J.N.J., 2000. Chemistry of the Upper and Lower 289.
Atmosphere, Academic Press, San Diego, pp. 349‐360. Paatero, P., 2000. User’s guide for positive matrix factorization programs
Gao, Y., Nelson, E.D., Field, M.P., Ding, Q., Li, H., Sherrell, R.M., Gigliotti, PMF2 and PMF3, Part 1: tutorial, pp. 65‐87.
C.L., Van Ry, D.A., Glenn, T.R., Eisenreich, S.J., 2002. Characterization of Paatero, P., 1997. Least squares formulation of robust non‐negative factor
atmospheric trace elements on PM2.5 particulate matter over the New analysis. Chemometrics and Intelligent Laboratory Systems 37, 23‐35.
York ‐ New Jersey harbor estuary. Atmospheric Environment 36, 1077‐
1086. Polissar, A.V., Hopke, P.K., Poirot, R.L., 2001. Atmospheric aerosol over
Vermont: chemical composition and sources. Environmental Science &
Harrison, R.M., Jones, A.M., Lawrence, R.G., 2003. A pragmatic mass Technology 35, 4604‐4621.
closure model for airborne particulate matter at urban background and
roadside sites. Atmospheric Environment 37, 4927‐4933. Pope, C.A., Dockery, D.W., 2006. Health effects of fine particulate air
pollution: lines that connect. Journal of the Air & Waste Management
Hien, P.D., Bac, V.T., Thinh, N.T.H., 2004. PMF receptor modelling of fine Association 56, 709‐742.
and coarse PM10 in air masses governing monsoon conditions in Hanoi,
Northern Vietnam. Atmospheric Environment 38, 189‐201. Raes, F., Van Dingenen, R., Vignati, E., Wilson, J., Putaud, J.P., Seinfeld, J.H.,
Adams, P., 2000. Formation and cycling of aerosols in the global
Hopke, P.K., Paatero, P., 2002. Discarding or downweighing high‐noise troposphere. Atmospheric Environment 34, 4215‐4240.
variables in factor analytic models. Annual Conference of the American
Association for Aerosol Research, Charlotte, NC, pp. 89‐96. Seinfeld, J.H. Pandis, S.N., 2006. Atmospheric Chemistry and Physics: From
Air Pollution to Climate Change, John Wiley & Sons, New York, pp. 408‐
Hueglin, C., Gehrig, R., Baltensperger, U., Gysel, M., Monn, C., Vonmont, H., 440.
2005. Chemical characterisation of PM2.5, PM10 and coarse particles at
urban, near city and rural sites in Switzerland. Atmospheric Solomon S., Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B.,
Environment 39, 637‐651. Tignor, M., Miller, H.L., 2007. Climate Change: The Physical Science
Basis, Contribution of Working Group I to the Fourth Assessment
INEC (Institute of Statistics and Census), 2011. National Population Census. Report of the Intergovernmental Panel on Climate Change, Cambridge,
Preliminary Report, San Jose, Costa Rica, 67 pages. 996 pages.
Jang, M.S., Czoschke, N.M., Lee, S., Kamens, R.M., 2002. Heterogeneous Terzi, E., Argyropoulos, G., Bougatioti, A., Mihalopoulos, N., Nikolaou, K.,
atmospheric aerosol production by acid‐catalyzed particle‐phase Samara, C., 2010. Chemical composition and mass closure of ambient
reactions. Science 298, 814‐817. PM10 at urban sites. Atmospheric Environment 44, 2231‐2239.
Kaufman, Y.J., Koren, I., 2006. Smoke and pollution aerosol effect on cloud Turpin, B.J., Lim, H.J., 2001. Species contributions to PM2.5 mass
cover. Science 313, 655‐658. concentrations: revisiting common assumptions for estimating organic
Kim, E., Hopke, P.K., Larson, T.V., Maykut, N.N., Lewtas, J., 2004. Factor mass. Aerosol Science and Technology 35, 602‐610.
analysis of Seattle fine particles. Aerosol Science and Technology 38, Turpin, B.J., Huntzicker, J.J., 1995. Identification of secondary organic
724‐738. aerosol episodes and quantitation of primary and secondary organic
Kim, E., Larson, T.V., Hopke, P.K., Slaughter, C., Sheppard, L.E., Claiborn, C., aerosol concentrations during SCAQS. Atmospheric Environment 29,
2003. Source identification of PM2.5 in an arid Northwest U.S. City by 3527‐3544.
positive matrix factorization. Atmospheric Research 66, 291‐305. Watson, J.G., Chow, J.C., Lowenthal, D.H., Pritchett, L.C., Frazier, C.A.,
Kulshrestha, A., Satsangi, P.G., Masih, J., Taneja, A., 2009. Metal Neuroth, G.R., Robbins, R., 1994. Differences in the carbon
concentration of PM2.5 and PM10 particles and seasonal variations in composition of source profiles for diesel‐powered and gasoline‐
urban and rural environment of Agra, India. Science of the Total powered vehicles. Atmospheric Environment 28, 2493‐2505.
Environment 407, 6196‐6204. WHO (World Health Organization), 2006. Health Risks of Particulate Matter
Lammel, G., Leip, A., 2005. Formation of nitrate and sulfate in the plume of From Long‐Range Transboundary Air Pollution, Copenhagen, 99 pages.
Berlin. Environmental Science and Pollution Research 12, 213‐220. Yadav, S., Rajamani, V., 2006. Air quality and trace metal chemistry of
Lazaridis, M., Semb, A., Larssen, S., Hjellbrekke, A.G., Hov, O., Hanssen, J.E., different size fractions of aerosols in N‐NW India ‐ implications for
Schaug, J., Torseth, K., 2002. Measurements of particulate matter source diversity. Atmospheric Environment 40, 698‐712.
within the framework of the European Monitoring and Evaluation Yatkin, S., Bayram, A., 2008. Source apportionment of PM10 and PM2.5 using
Programme (EMEP) I. first results. Science of the Total Environment positive matrix factorization and chemical mass balance in Izmir,
285, 209‐235. Turkey. Science of the Total Environment 390, 109‐123.
Loosmore, G.A., Hunt, J.R., 2000. Dust resuspension without saltation. Yin, J.X., Harrison, R.M., 2008. Pragmatic mass closure study for PM1.0, PM2.5
Journal of Geophysical Research‐Atmospheres 105, 20663‐20671. and PM10 at roadside, urban background and rural sites. Atmospheric
Marcazzan, G.M., Vaccaro, S., Valli, G., Vecchi, R., 2001. Characterisation of Environment 42, 980‐988.
PM10 and PM2.5 particulate matter in the ambient air of Milan (Italy). Yu, H., Kaufman, Y.J., Chin, M., Feingold, G., Remer, L.A., Anderson, T.L.,
Atmospheric Environment 35, 4639‐4650. Balkanski, Y., Bellouin, N., Boucher, O., Christopher, S., DeCola, P.,
Mariani, R.L., De Mello, W.Z., 2007. PM2.5‐10, PM2.5 and associated water‐ Kahn, R., Koch, D., Loeb, N., Reddy, M.S., Schulz, M., Takemura, T.,
soluble inorganic species at a coastal urban site in the metropolitan Zhou, M., 2006. A review of measurement‐based assessments of the
region of Rio de Janeiro. Atmospheric Environment 41, 2887‐2892. aerosol direct radiative effect and forcing. Atmospheric Chemistry and
Mazzei, F., D'Alessandro, A., Lucarelli, F., Nava, S., Prati, P., Valli, G., Vecchi, Physics 6, 613‐666.
R., 2008. Characterization of particulate matter sources in an urban Yuan, C.S., Lee, C.G., Liu, S.H., Chang, J.C., Yuan, C., Yang, H.Y., 2006.
environment. Science of the Total Environment 401, 81‐89. Correlation of atmospheric visibility with chemical composition of
Kaohsiung aerosols. Atmospheric Research 82, 663‐679.
Copyright of Atmospheric Pollution Research is the property of Turkish National Committee
for Air Pollution Research & Control (TUNCAP) and its content may not be copied or
emailed to multiple sites or posted to a listserv without the copyright holder's express written
permission. However, users may print, download, or email articles for individual use.

You might also like