Content uploaded by Alessandro Borgini
Author content
All content in this area was uploaded by Alessandro Borgini on Feb 21, 2019
Content may be subject to copyright.
Technical note
Estimation of particle mass concentration in ambient air using
a particle counter
A. Tittarelli
a
,
*
, A. Borgini
a
, M. Bertoldi
a
, E. De Saeger
b
, A. Ruprecht
a
, R. Stefanoni
a
,
G. Tagliabue
a
, P. Contiero
a
, P. Crosignani
a
a
Cancer Registry and Environmental Epidemiology Unit, National Cancer Institute, Via Venezian 1, 20133 Milano, Italy
b
European Commission, Joint Research Centre Institute for Environment and Sustainability, Transport and Air Quality Unit,
Via Fermi 1/I, 21020 Ispra (Varese), Italy
article info
Article history:
Received 11 March 2008
Received in revised form 28 July 2008
Accepted 28 July 2008
Keywords:
Air pollution
PM
Particle counter
Size distribution
abstract
Particle count may have advantage over particle mass concentration for assessing the
health effects of airborne particulate matter. However, health effects have mainly been
investigated with mass-measuring instruments, so it is important to assess relationships
between the variability of particle number, as determined by an optical particle counter,
and the variability of particle mass as measured by traditional mass-measuring instru-
ments. We used a light scattering particle counter to monitor the concentration of
particulate matter in ambient air in a northern Italian city continuously from August 2005
to July 2006. Six channels were calibrated to count particles in the size range 0.3–10
m
m
and above. Particles under 0.3
m
m cannot be detected by the instrument. The particle
counter was placed alongside the mass-measuring instruments of the Environmental
Protection Agency of the Region of Piemonte (ARPA). Particle numbers were transformed
into masses and compared with PM
10
and PM
2.5
data obtained from the ARPA instruments.
Daily average values were compared. The correlation between the two methods was good
for both PM
10
(R
2
¼0.734) and PM
2.5
(R
2
¼0.856); differences between means were
significant only for PM
2.5
. These findings suggest that a light scattering particle counter
might be suitable for assessing particulate matter variability in epidemiological studies on
effects of air pollution, though further investigations are necessary.
Ó2008 Elsevier Ltd. All rights reserved.
1. Introduction
Air pollution is an important determinant of human
health and the deleterious effects of airborne particulate
matter are of major concern. Epidemiological studies have
shown that the level of particulate air pollution is associ-
ated with adverse short-term (Samet et al., 2000; Jaffe
et al., 2003) and long-term health effects (Dockery et al.,
1993;Pope et al., 2002;Gauderman et al., 2004).
In all the above-cited studies exposure to particulate
matter (PM) was assessed by determining mass concen-
trations (
m
gm
3
) of particles of aerodynamic diameter less
than 10
m
m (PM
10
) or less than 2.5
m
m (PM
2.5
). However, it
has been suggested that the number rather than the mass
per unit volume of fine particles in air might be more
closely correlated with adverse health effects (Wichmann
et al., 2000). Particle number is an important indicator of
air quality (Gomis
ˇc
ˇek et al., 2004). Ruuskanen et al. (2001)
suggested that both particle number and mass concentra-
tions should be measured to provide a comprehensive
assessment of urban air quality, as well as to investigate
associations between air pollution and adverse health
outcomes.
The most important advantages of particle counters are
their mobility, low cost, ease of use and their ability to
measure particle concentrations over short time intervals
(1 s). They can therefore be used to assess spatial and
*Corresponding author. Tel.: þ39 02 23903542; fax: þ39 02 23902762.
E-mail address: andrea.tittarelli@istitutotumori.mi.it (A. Tittarelli).
Contents lists available at ScienceDirect
Atmospheric Environment
journal homepage: www.elsevier.com/locate/atmosenv
1352-2310/$ – see front matter Ó2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.atmosenv.2008.07.056
Atmospheric Environment 42 (2008) 8543–8548
temporal variations in particle concentrations (Weijers
et al., 2004) and provide good approximations of real
exposure in various situations (Gouriou et al., 2004).
However, as health effects have mainly been investi-
gated with mass-measuring instruments, it is important to
determine the relationship between particle number, as
determined by a particle counter, and particle mass as
measured by traditional mass-measuring instruments.
The aim of this study was to assess whether the vari-
ability of particle counting correlated satisfactorily with the
variability of concentration measurements and whether
there were differences in relation to meteorological vari-
ables. Our work was also stimulated by suggestions derived
from the literature concerning the use of particle counters to
assess air pollution concentration (Tuch et al., 1997; Weijers
et al., 2004; Gomis
ˇc
ˇek et al., 2004) and to investigate the
relationship between particle number and particle mass
(Harrison et al.,1999; Yanoskyet al., 20 02; Hoek et al., 2008).
Such a study is a first step in assessing the validity of
counting particles of different sizes, which is expected to be
used for studies relating air pollution to health outcomes.
The present study was performed in Turin, a city of 902 569
inhabitants in north-west Italy, over the period of one year.
2. Experimental
2.1. The particle counter
The device we used was a six-channel particle counter
(model 9012-2, Met One Instruments, Inc., Rowlett, Texas,
USA) which employs light scattering technology and a laser
diode optical sensor to detect and count particles in six size
ranges. The instrument continuously samples air at
2.83 L min
1
. The average number of particles counted per
litre per minute in each channel is recorded by the data
acquisition system.
The instrument was calibrated by the manufacturer
using traceable polystyrene latex particles, following
the method prescribed by the US National Institute of
Standards and Technology (web site http://ts.nist.gov/
traceability, accessed 28 February 2008); the six channels
were set to count particles of the following range of
diameters:
(1) 0.3–0.5
m
m
(2) >0.5–0.7
m
m
(3) >0.7–1.0
m
m
(4) >1.0–2.5
m
m
(5) >2.5–10
m
m
(6) >10
m
m
Particles under 0.3
m
m diameter cannot be detected by
the instrument.
The device was installed at the Region of Piemonte
Environmental Protection Agency (ARPA) station of
Lingotto, within 1 m of instruments measuring masses of
PM
10
and PM
2.5
. All instruments were 3 m above the
ground. The Lingotto station is situated in a park in a resi-
dential area of Turin and is classified as an urban back-
ground station.
The particle counter was tested once a month by
checking the zero reading while purging the instrument
inlet with dust-free nitrogen. Flow rate was measured once
a week using a flow meter with the manufacturer’s cali-
bration certificate; the pump was adjusted to compensate
for small changes in the flow when necessary. To reduce
wear on the instrument it collected data for 12 min in each
hour, being turned on and off automatically at the begin-
ning and end of this period. The data were divided into 12
one-minute intervals. Measurements were carried out from
August 2005 to the end of July 2006.
2.2. Data check and number-mass transformation
The accuracy of hourly estimations using only 12 one-
minute measurementswas assessed in a pilot study in which
we compared the mass estimation using all 60 available
one-minute measurements with that obtained using 12
measurements. The agreement between the two was excel-
lent (R¼0.96, data not shown in detail), and it is not there-
fore expected that the smaller number of measurements is
likely to introduce a bias in the comparison.
Twelve one-minute values were obtained each hour.
The measurements of the first one-minute period were
always discarded as they proved not to be reliable. When
the measured flow rate differed from the set flow rate of
2.83 L min
1
, a correction factor was applied to the data
gathered during the preceding week.
Outliers were identified using an algorithm that rejec-
ted values five times higher or lower than the mean of the
10 preceding measurements, even if they belonged to the
previous twelve-minute period, or the mean of the 10
successive measurements including those in the successive
twelve-minute period.
The algorithm used to transform particle numbers to
mass assumed particles were spherical (Wittmaack, 2002)
and had a density of 1.65 g cm
3
, as suggested by Tuch et al.
(2000) and Weijers et al. (2004). For each channel, to
determine the mass per
m
gm
3
we apply an average
0
50
100
150
200
250
300
350
Aug-05
Sep-05
Oct-05
Nov-05
Dec-05
Jan-06
Feb-06
Mar-06
Apr-06
May-06
Jun-06
Jul-06
µg m-3
particle counter
ARPA beta attenuation
Fig. 1. Daily averages of PM
10
values determined by the particle counter and
particle mass (ARPA) methods over the period from 1 August 2005 to 31 July
2006.
A. Tittarelli et al. / Atmospheric Environment 42 (2008) 8543–85488544
particle diameter, which was calculated as the arithmetic
mean of the stated size interval for the first to fourth
channels. For the fifth channel, collecting a wider size range
(2.5–10
m
m), a value closer to the lower extreme was used
(i.e. 4.47). This number had been estimated empirically
using a sample of real data, considering that particle
number decreases exponentially as size increases.
PM
10
(
m
gm
3
) was calculated as the sum of mass for
channels 1–5. Similarly, PM
2.5
(
m
gm
3
) was calculated as
the sum of mass for channels 1–4.
2.3. PM mass data
PM
10
and PM
2.5
were measured continuously at the
ARPA station at Lingotto, using devices that are checked
daily. PM
10
was measured using a beta attenuation SM200
instrument (Opsis, Furulund, Sweden) operating in mass
mode. PM
10
values were determined every 2 h. PM
2.5
was
measured by the European gravimetric reference method
using a Charlie HV instrument (TCR Tecora, Corsico, Italy).
The PM
2.5
mass measurements were available as daily
values.
2.4. Meteorological data
Hourly values of meteorological variables were supplied
by the meteorological station of the Turin Giardini Reali
(Royal Gardens, about 6 km from the Lingotto station).
Daily averages of the following were used: air temperature
(
C), relative humidity (%), wind speed (m s
1
) and rainfall
(mm).
2.5. Data processing and statistical analyses
Particle masses were calculated from the particle
counter data as daily averages because the ARPA PM
2.5
measurements are available as daily averages and because
daily averages are generally used in epidemiological
studies. The two-hourly ARPA PM
10
values were also con-
verted to daily averages. A daily average was considered
valid only if 75% of the hours were covered.
Logarithmic transformation was performed to render
PM distributions normal (Lu and Fang, 2003). Differences
between means were evaluated by the two-tailed t-test.
Correlations between our PM mass estimates and the PM
values supplied by ARPA were assessed by linear regres-
sion, considering ARPA values as the independent variable
and our mass estimation as the dependent variable.
Correlations between values for each channel, and
correlation between the fine (PM
2.5
) and coarse (PM
(10–2.5)
)
0
50
100
150
200
250
300
350
0 50 100 150 200 250
ARPA beta attenuation (
µg m
-3
)
particle counter (
µg m
-3
)
Fig. 2. Correlation between mean daily values of PM
10
determined by the
particle counter and those determined by particle mass (ARPA) method over
the period from 1 August 2005 to 31 July 2006.
particle counter
ARPA gravimetric
0
50
100
150
200
250
Aug-05
Sep-05
Oct-05
Nov-05
Dec-05
Jan-06
Feb-06
Mar-06
Apr-06
May-06
Jun-06
Jul-06
µg m-3
Fig. 3. Daily averages of PM
2.5
values determined by the particle counter
and particle mass (ARPA) methods over the period from 1 August 2005 to 31
July 2006.
0
50
100
150
200
250
0 20 40 60 80 100 120 140 160 180 200
gravimetric
ARPA (
µg m
-3
)
particle counter (
µg m
-3
)
Fig. 4. Correlation between mean daily values of PM
2.5
determined by the
particle counter and those determined by particle mass (ARPA) method over
the period from 1 August 2005 to 31 July 2006.
Table 1
Influence of meteorological variables on PM
10
estimated by the particle
counter and particle mass (ARPA) methods
No. of
days
Mean PC
a
(
m
gm
-3
)
Mean
ARPA
b
(
m
gm
-3
)
pvalue
(t-test)
R
2
b
c
Rain (mm) 0 224 52.73 58.69 0.017 0.755 0.942
>0 78 46.20 40.32 0.184 0.659 1.139
Temperature
(
C)
<13.5 145 74.92 74.37 0.898 0.732 1.013
13.5 157 28.39 35.17 <0.001 0.820 0.761
Relative
humidity
(%)
d
<70 129 33.83 50.43 <0.001 0.798 0.698
70 94 78.36 70.07 0.096 0.786 1.110
Wind speed
(m s
1
)
<1.0 259 46.25 48.56 0.290 0.723 0.992
1.0 7 32.28 40.71 0.256 0.799 0.756
All days 305 51.07 53.97 0.179 0.734 0.969
a
PM
10
mass estimated by particle counter.
b
PM
10
mass measured by ARPA beta attenuation instrument.
c
coefficient of linear regression (y¼
b
x), with ARPA values as inde-
pendent variable and PC values as dependent variable.
d
Excluding rainy days.
A. Tittarelli et al. / Atmospheric Environment 42 (2008) 8543–8548 8545
components of particulate matter, estimated by the
particle counter, were assessed by Pearson’s correlation
coefficient (R).
Correlation coefficients were calculated for dichoto-
mized values of temperature, relative humidity, rainfall and
wind speed. The t-test was used to assess the significance of
differences between the means of the distributions of each
group. All the analyses were performed with Stata/SE
version 8.2.
3. Results and discussion
3.1. Particle number and mass
We considered particle count measurements obtained
in 365 days from 1 August 2005 to 31 July 2006. A total of
150 335 valid one-minute estimates were available. From
these 365 daily averages were calculated, 32 of which were
excluded as more than 25% of hourly measurements were
missing, leaving 333 valid daily averages. For the PM
10
analyses ARPA data were missing for 28 of these days and
305 daily averages were used. For PM
2.5
analysis, 325 values
were used, as only eight corresponding PM
2.5
values from
ARPA were missing.
Over the entire study period, our estimate of the average
PM
10
was 51.07
m
gm
3
(standard deviation [SD] 52.25)
compared to 53.97
m
gm
3
(SD 35.49) measured by ARPA.
The t-test indicated that the difference between the means
of the two distributions was not significant (p¼0.179).
Daily values were in good agreement, as illustrated in
Figs. 1 and 2, with goodness of fit R
2
¼0.734.
Fig. 3 shows daily PM
2.5
values estimated by the particle
counter in comparison with those measured by ARPA. Our
estimate of average PM
2.5
over the entire period was
36.06
m
gm
3
(SD 32.73) compared to 40.28
m
gm
3
(SD
31.63) measured by ARPA. The correlation was better
(Fig. 4) for PM
10
, with R
2
¼0.856. However, the means of
the two distributions differed significantly (t-test,
p¼0.001).
The algorithm used to convert number to mass may
have had an influence on the mean values we found.
Particles are not in general perfectly spherical (Taylor,
2002); furthermore, variations in particle density with size
(Wittmaack, 2002) and time of day (Morawska et al., 1999)
are also likely. However, using different values for the
particle density will change the estimated mass values, but
will not affect correlations between the number of particles
and the estimated mass.
3.2. Influence of meteorological variables
Meteorological variables were dichotomized to investi-
gate their effects on the estimates of PM
10
(Table 1) and
PM
2.5
(Table 2). Median values were used as cut-offs for
temperature (13.4
C) and relative humidity (70%). There
were 88 rainy days (0.1 mm rain) and 8 windy days (wind
speed 1.0 m s
1
) over the study period.
Relative humidity (not analysed on rainy days) had
a major influence on the agreement between the two
measures: when relative humidity was high, PM
10
masses
derived from particle counts were higher (though not
significantly) than those estimated by ARPA, whereas PM
2.5
masses from particle counts were significantly lower than
those reported by ARPA; when humidity was low, both our
PM masses were significantly lower than ARPA values. A
likely explanation is that particles are hygroscopic (Wittig
et al., 2004) and increase in size on absorbing water
resulting in mass overestimation (Wilson and Suh, 1997).
On rainy days the mass estimated by the particle
counter was higher and correlation was lower (both for
PM
10
and PM
2.5
), presumably also due to water absorption
by particles.
Temperature had differing influences on PM
10
and on
PM
2.5
: for PM
10
the means of the two distributions differed
significantly at higher (13.5
C) temperatures, although
the correlation was slightly stronger; for PM
2.5
the two
means differed significantly at lower (<13.5
C) but not
higher temperatures; the correlation remained good for
both temperature intervals. This could well be a chance
Table 2
Influence of meteorological variables on PM
2.5
estimated by the particle
counter and particle mass (ARPA) methods
No. of
days
Mean
PC
a
(
m
gm
-3
)
Mean
ARPA
b
(
m
gm
-3
)
pvalue
t-test
R
2
b
c
Rain (mm) 0 241 37.58 44.12 <0.001 0.867 0.852
>0 81 31.43 28.70 0.147 0.840 1.106
Temperature
(
C)
<13.5 155 51.98 60.26 0.0 01 0.855 0.872
13.5 167 21.04 21.73 0.357 0.857 0.921
Relative
humidity
(%)
d
<70 137 24.37 31.45 <0.001 0.843 0.749
70 103 55.06 61.02 0.017 0.881 0.895
Wind speed
(m s
1
)
<1.0 269 31.87 33.41 0.135 0.848 0.963
1.0 8 22.87 25.37 0.462 0.890 0.908
All days 325 36.06 40.28 0.001 0.856 0.880
a
PM
2.5
mass estimated by particle counter.
b
PM
2.5
mass measured by ARPA gravimetric instrument.
c
coefficient of linear regression (y¼
b
x), with ARPA values as inde-
pendent variable and PC values as dependent variable.
d
Excluding rainy days.
Table 3
Distribution of particle numbers in the five particle size channels, with relative mass transformations
Channel Diameter (
m
m) Min (10
3
cm
3
) Max (10
3
cm
3
) Mean (10
3
cm
3
) Mean (
m
gm
3
) % of mass
1 0.3–0.5 13 517 507 529 199027 11.00 21.24
2>0.5–0.7 1131 256899 37403 6.98 13.47
3>0.7–1.0 247 73111 7139 3.79 7.31
4>1.0–2.5 124 29 743 3054 14.14 27.29
5>2.5–10 9 2146 206 15.93 30.68
Total PM
2.5
35.91 69.32
Total PM
10
51.84 100.00
A. Tittarelli et al. / Atmospheric Environment 42 (2008) 8543–85488546
finding, but may be worth investigating at various
temperature intervals.
Wind speed had a small effect. On windy days correla-
tions between both PM
10
and PM
2.5
were slightly better
than on non-windy days. As expected, PM concentrations
by both measurements were considerably lower on windy
days (wind speed 1.0 m s
1
) than non-windy days.
3.3. Relationship between particles of different sizes
As shown in Table 3, the number of particles was
inversely related to the diameter. Around 199 cm
3
parti-
cles were counted in the first channel (0.3–0.5
m
m)
decreasing to around 0.2 cm
3
particles in the fifth channel
(>2.5–10
m
m).
After transformation into mass (last two columns, Table
3), it was found that the fourth plus the fifth channels
(>1.0–10
m
m) provided the greatest contribution (almost
60%) to total PM
10
. Considering the first three channels
(particles 1
m
m), the first (0.3–0.5
m
m) provided the
greatest contribution to PM
10
(mean 21.2% of total),
because of the high number of particles counted in that
channel.
Analysis of the correlations (Pearson’s R) between each
of the channels (Table 4) showed that the intermediate
channels (second, third and fourth, counting particles from
0.5 to 2.5
m
m) correlated most strongly with each other.
Values for the first channel (0.3–0.5
m
m) correlated well
(R¼0.74) with those of the second channel (0.5–0.7
m
m),
but less well with all the others. Channel 5, counting
coarser particles (>2.5–10
m
m) correlated strongly with
channel 4 (R¼0.90) and channel 3 (R¼0.80).
4. Conclusions
One of the major limitations of the type of particle
counter we used is that it cannot detect particles under
0.3
m
m diameter, implying the underestimation of total
particle number and possible underestimation of mass in
our measurements. Such underestimation might
contribute to the differences found between the two
methods, but would not significantly affect the correlation.
It is important to note that the differences between the
two sets of estimates do not constitute a limitation of the
particle counting method. They could be due to the optical
properties of particles, their chemical composition or their
morphology. Particle counting provides information addi-
tional to that provided by mass measurement which is
likely to be useful for analysing the health effects of
particulate matter (Weijers et al., 2004).
Our findings do not suggest that particle counters
should substitute conventional instruments, but – though
further investigations are necessary – do suggest that they
provide valid particle number data of particular importance
in epidemiological studies for the evaluation of health
effects from particle matter in ambient air.
Acknowledgements
The authors are grateful to Donald Ward for revising the
English. They are also indebted to the technicians (Loretta
Badan, Maria Bondı
`, Sabina D’Attilio, Pier Paolo Fin, Fran-
cesco Romeo) of ARPA, and Mauro Grosa, director of the
Forecast and Environmental Monitoring Area of ARPA Pie-
monte, for precious on-site support.
References
Dockery, D.W., Pope III, C.A., Xu, X., Spengler, J.D., Ware, J.H., Fay, M.E.,
Ferris, B.G., Speizer, F.E., 1993. An association between air pollution
and mortality in six US cities. New England Journal of Medicine 329,
1753–1759.
Gauderman, W.J., Avol, E., Gilliland, F., Vora, H., Thomas, D., Berhane, K.,
McConnell, R., Kuenzli, N., Lurmann, F., Rappaport, E., Margolis, H.,
Bates, D., Peters, J., 2004. The effect of air pollution on lung devel-
opment from 10 to 18 years of age. New England Journal of Medicine
351, 1057–1067.
Gomis
ˇc
ˇek, B., Hauck, H., Stopper, S., Preining, O., 2004. Spatial and
temporal variation of PM
1
,PM
2.5
,PM
10
and particle number
concentration during the AUPHEP-project. Atmospheric Environment
38, 3917–3934.
Gouriou, F., Morin, J.P., Weill, 2004. On-road measurements of particle
number concentrations and size distributions in urban and tunnel
environments. Atmospheric Environment 38, 2831–2840.
Harrison, R.M., Jones, M., Collins, G., 1999. Measurements of the physical
properties of particles in the urban atmosphere. Atmospheric Envi-
ronment 33, 309–321.
Hoek, G., Kos, G., Harrison, R., de Hartog, J., Meliefste, K., ten Brink, H.,
Katsouyanni, K., Karakatsani, A., Lianou, M., Kotronarou, A.,
Kavouras, I., Pekkanen, J., Vallius, M., Kulmala, M., Puustinen, A.,
Thomas, S., Meddings, C., Ayres, J., van Wijnen, J., Hameri, K., 2008.
Indoor-outdoor relationships of particle number and mass in four
European cities. Atmospheric Environment 42, 156–169.
Jaffe, D.H., Singer, M.E., Rimm, A.A., 2003. Air pollution and emergency
department visits for asthma among Ohio Medicaid recipients,
1991–1996. Environmental Research 91, 21–28.
Lu, H.C., Fang, G.C., 2003. Predicting the exceedances of a critical PM
10
concentration – a case study in Taiwan. Atmospheric Environment 37,
3491–3499.
Morawska, L., Johnson, G., Ristovski, Z.D., Agranovski, V., 1999. Relation
between particle mass and number for submicrometer airborne
particles. Atmospheric Environment 33, 1983–1990.
Pope III, C.A., Burnett, R.T., Thun, M.J., Calle, E.E., Krewski, D., Ito, K.,
Thurston, G.D., 2002. Lung cancer, cardiopulmonary morality, and
long-term exposure to fine particulate air pollution. Journal of
American Medical Association 287, 1132–1141.
Ruuskanen, J., Tuch, T., Ten Brink, H., Peters, A., Khlystov, A., Mirme, A.,
Kos, G.P.A., Brunekreef, B., Wichmann, H.E., Buzorius, G., Vallius, M.,
Kreyling, W.G., Pekkanen, J., 2001. Concentrations of ultrafine, fine
and PM
2.5
particles in three European cities. Atmospheric Environ-
ment 35, 3729–3738.
Samet, J.M., Dominici, F., Curriero, F.C., Coursac, I., Zeger, S., 2000. Fine
particulate air pollution and mortality in 20 US cities, 1987–1994.
New England Journal of Medicine 343, 1742–1749.
Taylor, M.A., 2002. Quantitative measures for shape and size of particles.
Powder Technology 124, 94–100.
Tuch, T., Brand, P., Wichmann, H.E., Heyder, J., 1997. Variation of particle
number and mass concentration in various size ranges of ambient
aerosols in Eastern Germany. Atmospheric Environment 31,
4193–4197.
Tuch, T., Mirme, A., Tamm, E., Heinrich, J., Heyder, J., Brand, P., Roth, C.,
Wichmann, H.E., Pekkanen, J., Kreyling, W.G., 2000. Comparison of
two particle size spectrometers from ambient aerosol measurements
Table 4
Correlations between channels (Pearson’s R)
Channel (
m
m) 1 2 3 4 5
1 (0.3–0.5) 1.00 0.74 0.57 0.48 0.40
2(>0.5–0.7) 1.00 0.95 0.88 0.69
3(>0.7–1.0) 1.00 0.97 0.80
4(>1.0–2.5) 1.00 0.90
5(>2.5–10) 1.00
A. Tittarelli et al. / Atmospheric Environment 42 (2008) 8543–8548 8547
in environmental epidemiology. Atmospheric Environment 34,
139–149.
Weijers, E.P., Khlystov, A.Y., Kos, G.P.A., Erisman, J.W., 2004. Variability of
particulate matter concentrations along roads and motorways
determined by a moving measurement unit. Atmospheric Environ-
ment 38, 2993–3002.
Wichmann, H.E., Spix, C., Tuch, T., Wolke, G., Peters, A., Heinrich, J.,
Kreyling, W.G., Heyder, J., 2000. Daily mortality and fine and
ultrafine particles in Erfurt, Germany. Part 1: role of particle
number and particle mass. Research Report Health Effects Institute
98, 5–86.
Wilson, W.E., Suh, H.H., 1997. Fine particles and coarse particles:
concentration relationships relevant to epidemiologic studies. Journal
of Air and Waste Management Association 47, 1238–1249.
Wittig, A.E., Anderson, N., Khlystov, A.Y., Pandis, S.N., Davidson, C.,
Robinson, A.L., 2004. Pittsburgh air quality study overview. Atmo-
spheric Environment 38, 3107–3125.
Wittmaack, K., 2002. Advanced evaluation of size-differential distribu-
tions of aerosol particles. Journal of Aerosol Science 33, 1009–1025.
Yanosky, J.D., Williams, P.L., MacIntosh, D.L., 2002. A comparison of two
direct-reading aerosol monitors with the federal reference method
for PM
2.5
in indoor air. Atmospheric Environment 36, 107–113.
A. Tittarelli et al. / Atmospheric Environment 42 (2008) 8543–85488548