&ER&
United States
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 2771:
Research and Development
EPA-600/S4-81-021 July 1981
Project Summary
Development of Methods and
Analysis of Particulate
Sampler Data from RAPS
Sandor J. Freedman and William M. Yeager
Methods were formulated, tested
and demonstrated to statistically
analyze and graphically process ana-
lytical data derived from dichotomous
particulate samplers operating at ten
Regional Air Monitoring System
(RAMS) stations in the greater St.
Louis area.
Statistical techniques and computer
programs are documented in this
report and the sample outputs are
presented for the following tasks:
Time series plots of total mass
and sulfate 24-hr averages for
high volume and dichotomous
sampler instruments
Means and standard deviations of
total mass and sulfate by station
by quarter for high volume and
dichotomous samplers
Correlations between high volume
and dichotomous sampler mea-
surements of total mass and sul-
fate at each station
Scatterplots of high volume vs.
dichotomous sampler measure-
ments for total mass and sulfate
at each station
Frequency distributions for the
fine/total ratio of total mass, sul-
fate, and lead at each station for
dichotomous sampler measure-
ments
Correlations between the fine
and coarse fractions for 23 ele-
ments plus total mass at each
station for dichotomous sampler
measurements
Correlations among six selected
elements and total mass at each
station by fraction for dichoto-
mous sampler measurements
This report was submitted by System
Sciences, Inc. in fulfillment of Con-
tract No. 68-02-2495 with the U.S.
Environmental Protection Agency.
The technical effort reported commenced
on January 13, 1977, and was com-
pleted on September 12, 1980.
This Project Summary was developed
by EPA 's Environmental Sciences Re-
search Laboratory. Research Triangle
Park, NC. to announce key findings of
the research project that is fully docu-
mented in a separate report of the
same title (see Project Report ordering
information at back).
Introduction
A recently developed technique, using
the dichotomous sampler coupled with
X-ray fluorescence analysis of samples,
has facilitated the determination of the
mass and elemental in concentrations
for particulate matter collected from
ambient air in two different size ranges.
The data consist of measurements
taken in the years 1975, 1976, and
1977 at ten sites in the greater St. Louis
area. Continuous sampling was con-
ducted, and periods between filter
changes ranged from 2 to 12 hours. The
sites coincided with 10 of the 25 sites of
the Regional Air Monitoring System
(RAMS), a part of the Regional Air
Pollution Study (RAPS). Site locations
-------
ranged from urban to rural areas at
varying distances from a variety of
industrial point sources as well as
sources of emissions from mobile vehi-
cles. Data generated by the RAMS
monitoring program include paniculate
mass and sulfate from a standard hi-
volume sampler as well as a variety of
measurements of other atmospheric
conditions.
The data provide a unique opportunity
for investigating relationships between
air quality and emissions as well as the
spatial and temporal characteristics of
ambient pollutant concentrations. The
work reported herein consists of the
development and implementation of
computer-assisted methodology for pre-
sentation and analysis of dichotomous
particulate sampler data. The analyses
include the following:
Time series plots, scatter plots, and
correlations of 24-hour averages of
total mass and sulfate for high
volume and dichotomous sampler
data at each station
Means and standard deviations of
total mass and sulfate by station by
quarter for high volume and dichoto-
mous samplers
Frequency distributions for the
fine/total ratio of total mass, sulfate,
and lead at each station for dichot-
omous sampler measurements
Correlations between the fine and
coarse fractions for 23 elements
and total mass at each station for
dichotomous sampler measure-
ments
Monitoring
The RAMS network consisted of 25
stations strategically located through-
out the greater St. Louis area. Dichot-
omous samplers and hi-volume samplers
were operated at ten of the sites (See
Table 1). These sites lay along four
concentric circles centered in down-
town St. Louis with radii of 4,10,20 and
40 kilometers, respectively. The sites
represented a diverse range of sampling
environments from heavily trafficked,
inner city areas to industrial locations to
agricultural settings. Choice of site
locations was based primarily upon the
extent of surrounding pollutant emis-
sion sources (stationary or mobile) and
the significance of topographical fea-
tures capable of influencing local air
flow.
The automated dichotomous sampler
operates on the principle of virtual
impaction. That is, particles are impacted
into a slowly pumped void thus elimi-
nating the problem of particle bounce on
collection surfaces. The device collected
particles in two size ranges: fine«2.4/um)
and coarse (2.4 to 20 yum). Filters from
the dichotomous sampler were sent to
Lawrence Berkeley Laboratory (LBL),
Berkeley, CA for mass and elemental
analysis.
X-ray fluorescence spectroscopy was
employed at LBL to determine the ele-
mental composition of samples in both
fine and coarse fractions. The composi-
tion of a sample was determined by irra-
diation with a beam of X-ray and obser-
vation of the emissions of characteristic
K and L X-rays. The procedure was well
suited for analysis of elements heavier
than Mg in atomic weight. The following
elements were analyzed: Al, Si, P, S, Cl,
K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, Ga,
As, Se, Br, Rb, Sr, Cd, Sn, Sb, Ba, and Pb.
A beta gauge was used to determine
total mass for both fine and coarse frac-
tion samples. The beta gauge method
for aerosol mass determination had an
accuracy of 12 jig/m3 for 2-hr samples
and 2 ug/m3 for 12-hr samples. The
reproducibility of the X-ray fluorescence
measurements was found to be within
5% for major elements.
Hi-volume particulate samplers were
also operated at each of the 10 dichoto-
mous sampler sites in the RAMS net-
work. Samples were collected for 24-
hour periods every third day. Hi-volume
samples were analyzed by Rockwell In-
ternational for total particulate mass, sul-
fate and nitrate concentrations. Dichot-
omous samplers and hi-volume samplers
were operated simultaneously side-by-
side; pollutant data collected simul-
taneously were matched for compari-
son of the two types of instruments.
Results
Tables 1 and 2 show comparisons of
the high volume and dichotomous sam-
pler results. For both total mass and sul-
fate, the high volume instrument collects
significantly greater amounts than does
the dichotomous sampler. Overall, the
ratio of high volume and dichotomous
sampler results is 1.87 ± .03 for total
mass, 1.47 ± .02 for sulfate. There are
no significant differences between sta-
tions in the ratios for either mass or
sulfate. The central sites have the largest
concentration of both mass and sulfate.
In general, the mass concentrations
were highest during the second quarter
(April-June); the sulfate concentrations
were highest during the third (July-
September). For total mass the overall
correlation coefficient is 0.671, varying
from 0.377 at station 120 to 0.860 at
Table 1. Linear Regression Analysis Hi-Volume Mass vs. Dichotomous Total Mass
Station
Number
103
105
106
108
112
115
118
120
122
124
All
No. of
Cases
120
118
95
91
115
111
104
97
112
77
1040
Mean
HV/DS
Ratio
1.93
2.04
1.76
1.96
1.73
1.57
2.18
1.71
1.72
2.17
1.869
S.D.
HV/DS
Ratio
0.85
0.78
0.82
1.09
O.71
0.52
1.33
1.09
0.92
1.32
0.970
Corr.
Coeff.
0.637
0.646
0.666
0.708
0.722
0.860
0.514
0.377
0.594
0.433
0.671
Intercept*
(ug/m3)
43.99
49.46
29.40
31.50
21.11
14.33
37.02
31.06
22.27
28.33
27.04
95% Conf.
Interval
-16.28~104.26
+ 0.24- 98. 68
-20.88- 79.68
-24.23- 87.23
-37.26- 79.48
-12.56- 41.22
-15.01- 89.05
-28.80- 90.92
-26.49- 71.03
-23.56- 80.22
-25.81- 79.89
Slope*
0.902
0.786
0.967
1.051
1.143
1.074
0.809
0.682
0.861
0.919
1.020
95% Conf.
Interval
0.703-1.101
0.615-0.957
0.744-1.190
0.830-1.272
0.938-1.348
0.953-1.195
0.544-1.074
0.340-1.024
0.640-1.082
0.478-1.360
0.951-1.089
'The intercept of the regression line is the predicted value of the HIVOL mass when the dichotomous sampler mass is zero.
"The slope of the regression line is the predicted change in HIVOL mass for unit change in dichotomous sampler mass.
-------
Table 2. Linear Regression Analysis Hi-Volume Sulfate vs. Dichotomous Sulfate
Station
Number
103
105
106
108
112
115
118
120
122
124
All
No. of
Cases
132
141
103
105
143
129
110
106
121
78
1168
Mean
HV/DS
Ratio
1.47
1.39
1.35
1.47
1.39
1.35
1.53
1.80
1.53
1.45
1.47
S.D.
HV/DS
Ratio
0.65
0.34
0.45
0.66
0.83
0.46
0.75
1.35
0.66
0.55
0.72
Corr.
Coeff.
0.857
0.964
0.937
0.811
0.818
0.904
0.854
0.637
0.702
0.952
0.838
Intercept
(M/m*)
4.574
2.815
2.983
4.647
2.927
3.418
4.725
3.705
3.711
2.695
3.620
95% Conf.
Interval
-1 .893*1 1 .041
-0.796* 6.426
-3.435* 9.401
-3.990*13.284
-6.576*12.430
-1.979* 8.815
-1.870*11.320
-9.418*16.828
-8.715*16.137
-2.932* 8.322
-4.481*11.721
Slope*1
0.791
0.939
0.916
0.778
0.928
0.808
0.706
0.949
0.888
0.951
0.864
95% Conf.
Interval
0.708*0.874
0.896*0.982
0.848*0.984
0.668*0.888
0.819*1.037
0.741*0.875
0.624*0.788
0.725*1.173
0.724*1.052
0.881*1.021
0.832*0.896
'The intercept of the regression line is the predicted value of the HIVOL mass when the dichotomous sampler mass is zero.
"The slope of the regression line is the predicted change in HIVOL mass for unit change in dichotomous sampler mass.
station 115. The correlation coefficients
for sulfate are higher, with an overall
value of 0.838, varying from 0.637 at
station 120 to 0.964 at station 105.
Differences between the measure-
ments by high volume and dichotomous
samplers may be at least partially
explained by differences in the design of
the two instruments. The high volume
sampler had essentially no upper cutoff
for particle size, but the dichotomous
sampler rejected particles with diam-
eters greater than 20 um. Although
particle size distributions were not
measured by the RAMS instruments,
previous studies have shown that aero-
sol particles have a bimodal size distri-
bution, with different sources'for each
mode. The fine particle mode (< 2 ton in
diameter) is primarily due to combustion
products. The coarse particle mode (> 3
//m in diameter) results from dispersion
of mechanically generated particles,
especially soil. The greater total mass
measured by the high volume samplers
may be due to particles larger than 20
ion diameter in St. Louis aerosol. Known
sources of dust in the area include road
dust, rock quarries, cement manufac-
turing, and wind erosion of soil. Applica-
tion of a chemical element balance
model to the dichotomous sampler data
from St. Louis indicated that crustal
components account for 43% of the total
mass.
Another difference between the high
volume and the dichotomous samplers
is in the filter media. The former used
glass fiber filters while the latter used
cellulose ester membrane filters. Glass
fiber filters are known to collect artifact
SO4 in the presence of SOZ gas. This
may account for the larger amounts of
sulfate collected in the high volume
samplers. The higher correlations be-
tween the high volume and dichotomous
measurements of sulfate probably result
from the fact that most of the sulfate
(89%) is in the fine fraction, which is not
affected by differences in inlet design.
Other differences between the two
instruments include the lack of symmetry
in the gable roof of the volume sampler,
which makes its collection efficiency
sensitive to wind direction. The dichoto-
mous sampler used inlets having axial
symmetry. The high volume samplers
were operated from midnight to mid-
night on a three or six day cycle; filters
remained in the instrument for several
hours after the air flow was shut off,
which could have allowed some addi-
tional particles to have settled onto
them.
The analysis of the dichotomous
sampler data by size fraction showed
that the total mass was approximately
equally divided between the fine and the
coarse fractions. There was no signifi-
cant site dependence in the ratios of fine
to total mass and elemental concentra-
tions. Overall, the ratio was 0.518 ±
.001 for mass, 0.885 ± .001 for sulfur
and 0.799 ± .001 for lead. The correla-
tion between fine and coarse fractions
was highest for bromine and titanium,
lower for lead and sulfur, and even
lower for total mass. The correlations
among elements were highest for sulfur
with total mass in the coarse fraction,
for iron with total mass in the fine
fraction, and for lead with bromine in
both fractions. The high correlation
between lead and bromine is expected,
since both are due to auto emissions.
Sandor J. Freedman and William M. Yeager are with System Sciences, Inc.,
Chapel Hill, NC 27514.
Thomas G. Dzubay is the EPA Project Officer (see below).
The complete report, entitled "Development of Methods and Analysis of Panicu-
late Sampler Data from RAPS," (Order No. PB 81-179 327; Cost: $17.00,
subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
> US GOVERNMENT PfUNTINO OFFICE. 1(61 -757-OU/7Z07
-------
United States Center for Environmental Research
Environmental Protection Information
Agency Cincinnati OH 45268 Protect
Agency
EPA 335
Official Business
Penalty for Private Use $300
PS 0000329
U S ENVIR PROTECTION AGENCY
REGION 5 LIBRARY
230 S DEARBORN STREET
CHICAGO IL 60604
------- |