EPA-450/3-77-021c
August 1977
AN IMPLEMENTATION PLAN
FOR SUSPENDED
PARTICIPATE MATTER
IN THE PHOENIX AREA
VOLUME III. MODEL
SIMULATION OF TOTAL
SUSPENDED PARTICULATE
LEVELS
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
-------
EPA-450/3-77-021c
AN IMPLEMENTATION PLAN
FOR SUSPENDED PARTICULATE
MATTER IN THE PHOENIX AREA
VOLUME III. MODEL SIMULATION
OF TOTAL SUSPENDED
PARTICULATE LEVELS
by
George Richard, Jim Avery, and Lai Baboolal
TRW Environmental Engineering Division
One Space Park
Redondo Beach, California 90278
Contract No. 68-01-3152
EPA Project Officer: Dallas Safriet
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
"Office of Air and Waste Management
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
August 1977
-------
This report is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are
available free of charge to Federal employees, current contractors and
grantees, and nonprofit organizations - in limited quantities - from the
Library Services Office (MD-35), Research Triangle Park, North Carolina
27711; or, for a fee, from the National Technical Information Service,
5285 Port Royal Road, Springfield, Virginia 22161.
This report was furnished to the Environmental Protection Agency by
Environmental Engineering Division of TRW, Inc., One Space Park,
Redondo Beach, California, in fulfillment of Contract No. 68-01-3152.
Prior to final preparation, the report underwent extensive review and
editing by the Environmental Protection Agency. The contents reflect
current Agency thinking and are subject to clarification and procedural
changes.
The mention of trade names of commercial products does not constitute
endorsement or recommendation for use by the Environmental Protection
Agency.
Publication No. EPA-450/3-77-021c
11
-------
TABLE OF CONTENTS
1.0 INTRODUCTION AND SUMMARY 1-1
1.1 Results 1-1
1.2 Conclusions 1-4
2.0. FORMULATION OF SOURCE RECEPTOR RELATIONSHIP 2-1
2.1 Selection of the Air Diffusion Model 2-1
2.1.1 Phoenix Multiple Box Model 2-2
2.1.2 Denver Brown Cloud Model 2-3
2.1.3 Climatological Dispersion Model, COM ....... 2-7
2.1.4 Atmospheric Transport and Diffusion Model, ATOM . 2-9
2.1.5 Manna's Urban Model 2-11
2.2 Modifications to COM 2-12
2.3 Model Inputs 2-16
2.3.1 Meteorology Data 2-16
2.3.2 Emissions Parameters and Pollutant Half Life . . . 2-17
3.0 ADJUSTMENT OF AIR QUALITY DATA 3-1
3.1 Variation of TSP with Monitor Height 3-2
3.2 Representativeness of Monitor Environment 3-6
3.3 Completeness of TSP Data . . . 3-8
3.4 Summary of Bias of Air Quality Data 3-8
4.0 MODEL PARAMETERIZATION 4-1
4.1 Background Levels of TSP 4-1
4.2 Assignment of Empirical Constants 4-3
iii
-------
TABLE OF.CONTENTS - continued
Page
5.0 FORECASTED BASELINE TSP LEVELS FOR 1975 AND 1985 5-1
5.1 Baseyear TSP Levels 5-1
5.2 Projected Baseline TSP Levels 5-3
6.0 REFERENCES 6-1
iv
-------
1.0 INTRODUCTION AND SUMMARY
Under contract to the Environmental Protection Agency, TRW
Environmental Engineer!nq has developed control strategies for total
suspended participates in the Phoenix area. The data base and metho-
dology developed for Phoenix have been extended into a general technical
support document for application to areas with fugitive dust problems.
This report is the third of four technical support documents prepared
for the project.
The relationships between ambient total suspended particulate,
(TSP) and emission" levels were established in this study. An air
quality model, comprised of the COM and a simple rollback relation, was used
to forecast suspended particulate levels for small and larger particle
sizes. Present (1975) and future (1985) suspended particulate concen-
trations arising form individual source categories were simulated.
Controls were established after several iterations of air quality fore-
casts which were necessary in order to attain standards.
The present section (Section 1) introduces the study and summarizes
the major results and conclusions. Section 2 discusses the choice of
model. Section 3 identifies factors affecting the representiveness
of the emissions and air quality data base, and discusses how the bias
in the data base might be eliminated. Section 4 details the model
parameterization procedure, including assignment of empirical co-
efficients for each of the monitor sites. Section 5 presents the
modeling results for 1975 and 1985.
1.1 RESULTS
Particulate concentrations in the Phoenix area were simulated using
a superposition model which combined the results of the Climatological
Dispersion Model (COM) with those of linear rollback (LR). It was as-.
sumed in this study that particulates smaller than 20ym aerodynamic
diameter could be adequately modeled with the COM model, and those
greater than 20um could be adequately treated in the context of the LR
1-1
-------
modeling concept. In the latter, ambient particulate levels in a given
grid square are assumed proportional to the average emissions within
that grid.
The mathematical relation underlying the superposition principle
is of the form
Xi = aliCi + a2iEi+B
where X is the estimated TSP concentration at receptor i, C. is the COM
simulated level of particulates in the size range 0-20 microns at receptor
i, E.J is the emission level of particulates greater than 20 microns in
the grid square containing receptor i, B is the background level of
TSP, and a,, and cu. are empirical coefficients obtained by comparing
the simulated values with the measurements.
Microscopy analyses of hi-vol filters were used to establish the
contribution of each particle, size range to the TSP levels. These
analyses, although limited, confirm the substantial effect of large par-
ticles on TSP levels. Seventy percent (70%) of the particulate mass found
on hi-vol sampler filters throughout the study area was comprised of par-
ticles greater than 20 micron in diameter. This finding, along with the
model baseyear TSP simulations and observed TSP levels was used to es-
tablish empirically the a's at each of the sampler site.
An analysis of the representativeness of air quality data revealed
that the TSP measurements could be significantly biased because of three
major factors: 1) variation in monitor heights at the different sites;
2) site-specific sources affecting the monitor in a manner atypical of
the general area, and 3) incompleteness of measurements. The first
factor can result in substantial understatement of TSP levels for monitors
higher than the standard exposure height; the second factor can result
in significant overstatement of the air quality in the general area, and
the third factor can result in a bias of variable nature depending
on the site and annual meteorology. Sufficient information is not avail-
able in Phoenix to assess the quantitative relationship between these
1-2
-------
factors and measured TSP levels. Hence, air quality adjustments are not
possible.
Uncertainties associated with the emissions data base compound the
probable bias entering the model parameterization. It is not within the
scope of this study to evaluate the sensitivity of the air quality simu-
lation to the numerous influence factors affecting the data base. This
study has sought to identify some of the major sources of bias, and to
indicate areas where the model results could be used while at the same
time recognizing some shortcomings due to data bias.
TSP was estimated at the various sites for 1975 and 1985 with the
empirical model using the compiled emissions inventory. The results
(Table 5-1) show that in 1975 four particulate sources dominated: un-
paved roads, entrained street dust, construction activities and wind
erosion. By 1985 (Table 5-2), only the first three sources are likely
to dominate. Wind erosion will not be a major source then because of
less available open soil surfaces. By 1985, most of the monitor sites
will be affected more by entrained street dust than by any other source. Ex-
ceptions will occur at the more rural sites (St. Johns, Chandler, N.
Scottsdale, S. Phoenix) where unpaved road emissions will cause the
greatest impact on TSP levels.
As a result of the net changes in emission source magnitudes and
distribution, TSP levels in 1985 will decrease significantly at 11 of
the 13 monitoring sites considered (Table 5-2). These TSP levels will
be 16 to 50% less than the corresponding 1975 levels, depending on the
site. In most cases, the air quality improvements will be due primarily
to the forecasted regional development. This development
will reduce the proportion of open surfaces and diminish the magnitudes
of several local sources which currently affect certain samplers. At two
of "the monitors, TSP levels will increase due to persisting sources which
will be unchanged by expected development plans.
1-3
-------
The modeling method used shows good simulation capability. COM
is on firm theoretical ground and applies strictly to the smaller
particles. Linear rollback .is on weaker theoretical grounds. Consider-
ing the difficulties inherent in modeling particulates over the entire
range of particle sizes,the superposition scheme used in this study is
recommended for particulate modeling until such time as better methods
are developed.
1.2 CONCLUSIONS
The COM adequately characterizes the transport and diffusion of
smaller particulates (<20ym) in the atmosphere but is inadequate in the
study of larger particulates for which gravitational settling is an
important factor. A proper treatment of this problem must consider
gravitational settling.
Microscopy analysis of high volume filter samples can provide
useful data for model improvement. .Particle size distribution
data at the various sites may be used to determine size de-
pendent empirical constants (model parameterization). This technique
is especially suitable for larger particle regimes where the effect
of large particle emissions are so localized that only very site
specific adjustments are appropriate.
In Phoenix, the particle size distribution observed at the various
monitor sites appears to be very similar. Within a small variation (about
10%), particles 20ym and larger comprise 70% of the TSP by mass for each
of the hi-vol sites and filters examined.
Various influence factors (i.e., height of monitor, completeness
of data) may affect the representativeness of the air quality data base
and therefore its utility for model application. Although the data
base examined is small, it appears that height of the monitor may affect
values of observed TSP dramatically. Exposure concentrations at ground
level may be 30 to 40% greater than those recorded at monitors 5 or
6m above the ground.
.In areas where TSP levels are caused principally by fugitive dust,
normal growth patterns may have significant impact on future TSP levels
1-4
-------
and distributions primarily because fugitive dust is a short range
problem which tends to be diminished by local development. The
localized effect of fugitive sources and the changing distribution of
TSP levels due to regional development suggest .the importance of air
monitor placement and the need for "hot spot" monitoring. Emissions
density maps may be useful in identifying these TSP maxima.
Considering the state-of-the-art of particulate modeling the
superposition approach of combining COM with linear rollback appears
adequate. As emissions become better defined the modeling approach
"lust be likewise refined. This study suggests the need for developing
source-receptor models which will incorporate dry and wet deposition
and gravitational settling.
1-5
-------
2.0 FORMULATION OF SOURCE RECEPTOR RELATIONSHIP
This chapter discusses the choice of a suitable air quality model
for total suspended particulates in the Phoenix area. Available
diffusion models are.reviewed to evaluate their potential applicability
for emissions sources in the study area. A standard diffusion
model is selected, and modifications to the model are proposed to ac-
cour.t for the diverse spectrum of dispersion characteristics exhibited
by fugitive dust sources. Major inputs required to parametize the
forrulated model are discussed.
2.1 SELECTION OF THE AIR DIFFUSION MODEL
Averaging time is an important consideration in model selection.
Tne federal air quality standards define both a short term (24-hour)
and long term (annual) concentration. However, there is reasonable
cause to restrict our analysis to only the long term levels. First,
considerably greater control is required to attain the primary annual
standard at each station than is required to attain the primary 24-hour
standard, provided episodes due to duststorms are excluded (see Table 2-1)
Second, uncertainties with the data base also affect model selection.
Uncertainties are introduced at several stages of the air monitoring
measurements, emissions inventory compilation, and model formulation.
The analytical limitations inherent in the modeling of short term averages
do not warrant the additional effort at this time.
The discussion of the following sections includes a review of four
air diffusion models considered as potential tools for estimating annual
concentrations in the Phoenix area. The Phoenix Multiple Box (Berman and
I-eLaney, 1975), the Denver Brown Cloud (Middleton and Brock, 1975) and
Henna's Urban Model (Hanna, 1971) are reviewed and assessed to be inap-
plicable for particulate modeling in the Phoenix area. The Atmospheric
Transportand Diffusion Model, (Culkowski and Patterson, 1976) ATOM, is
reviewed and deemed applicable provided certain modifications are made.
Recormiended is the Climatological Dispersion Model (COM) for the long-
term (annual) averages. Some modifications to the COM are required for
an accurate representation of air quality from all particle size ranges.
2-1
-------
TABLE 2-1. SUMMARY OF 1973-1975 AIR QUALITY VIOLATIONS FOR
TSP IN PHOENIX AREA
Stations
Reporting
Central Phoenix
South Phoenix
Arizona State
Glendale
North Phoenix
N Scot/Paradise
Scottsdale
Mesa
Downtown
St. Johns
Sun City
Paradise Valley
Chandler
Carefree
3
TSP Concentration ug/m
Annual
139
. 170
156 -
97
127
143
110
124
199
145
84
191
136
41
Expected Second
Highest 24-Hra '.
370
320
390
22.0
340
450
225
250
450
630
200
430
320
135
Percentage Emission Reductions
to meet primary Standards" based
on linear rollback
Annual
58.7
67.8
64.1
32.8
53.6
60.1
43.7
52.1
73.3
60.8
16.6
72.0
57.5
i
24-Hour
32.3
20.6
36.1
--
25.8
45.2
--
45.2
61.6
--
48.8
20.6
!
3Based on statistically computed expected concentrations (from distributions
derived from historical data.for 1973 to 1975 [23] and assuming 60 measurements per
Annual primary standard = 75 vici
?4-Hour prirr.ary standard = 260- pg/n
2-2
-------
The model used, with modifications, is discussed in Section 2.2.
2.1.1 Phoenix Multiple Box Model
To study transport and diffusion of air pollutants over the
greater Phoenix area, Berman and DeLaney (1975) selected a multiple box
model. In particular, Gaussian concepts witri a single wind field could
not be used because of the spatially varying wind field which exists
in the area. In their model, which is based on the method of Reiquam
(1970), the wind field is an input in the basic equation.
Mass of Pollutant = Mass Imported + Mass Emitted + Mass Remaining
in each box from in in
Adjacent Box the Box the Box
And, when the boxes are of constant height, the concentration, X, in
a grid square (i,j) at time, t, is given by
(2-1)
where Q and S are the advective and emission rates, respectively, V
the volume of the (i,j) cell, and Rl , R2, and p are the residuals of
Q, S and X remaining at time t.
Stated more compactly, the above equation may be written as
Mass of Pollutant
, . = r(residual)(rate)
in each box v /v
The residuals are functions of the mean horizontal winds in a
given cell. The above equations do not account for vertical diffusion.
To include this, the residual terms were divided by a dilution factor.
These factors were obtained from Slade and Rag] and .
However, there are several problems with this model. First, it
would be strictly inapplicable to particulates. since it does not take
into account particle size. Second, it is essentially a short term
-------
model from which long term averages may be obtained by aggregating
('brute'force1 method) the hourly simulation -- a process which is both
time consuming and costly. Third, it employs factors Cdilution) which
are derived from data gathered elsewhere. For these reasons this
model cannot be applied to the present problem.
2.1.2 Denver Brown Cloud Model
Another approach which did take into account the changing particle
sizes is the Denver Brown Cloud Model (Middleton and Brock, 1974). In
this model, the evolution of the particle size spectrum,assuming some
initial distribution's investigated as an air parcel traverses over
a given source distribution. The primary physical mechanisms modeled
are coagulation, condensation and deposition. Neglected in the model
are the urban "heat island effects," the "chimney effect" and complex
air circulations other than drainage flows.
In this model, the evolution of the density function n(x,r,t) for
* i
an aerosol with convective transport is described by
dn (x,T,t) +v 7n(x,r,t) = V K -Vn(x,T,t)
at ' .-,-.'.. . : - -
1 fx dx' b(x-x',x) n (x-x',T,t) n (x',T,t)
/oo
- n(x,r,t) / dx' b(x',x) n (x',T,t)
*^j>
a [>(x) n (n,~r",t)] + 2 tai
- TT OX
+ ?(x) vn(x,r,t) + Z "A, (x,T,t) + xijjj (x,r",t)
P J (2-2)
where n(x»r,t) represents the number of aerosol particles of mass, x,
between the mass interval x and x+ dx and at position f at time t; and f
velocity of the air mass, K the eddy diffusivity tensor, G(x) the sed-
imentation velocity of a particle of mass x, $ (s,r,t) the rate of pro-
duction of particles of mass x at r,t from primary sources, v.,.(x,?",t)
the rate of production of particles of mass x at r,t by homogeneous
2-4
-------
nucleation of the j th chemical species, b(x,x) the coagulation co-
efficient and i'(x) and a(x) the condensation coefficients which account
for heterogeneous nucleation.
The first term on the right-hand side of the equation accounts
for the turbulent dispersion of the aerosols, the second and third
terms account for coagulation, the fourth and fifth terms treats the~
heterogeneous nucleation involving the j-th chemical species.
To be applied to the problem at hand, the above equation must
be averaged over time and space. In this process, the volume average
of the density function then becomes
= / n(x,r,t)dr
(2-3)
where the time dependence is dropped on the left-hand side in the
interest of brevity.
If, subsequent to the above averaging, the above volume integral
is converted to a surface integral the evolution of the density function
becomes
aar =]_ /"* b(x-x',x) dx'
- f° b(x,x') dx'
+ 2 ti W1 (x)
P
(2-4)
In this equation, the homogeneous nucleation term has been omitted
because it is small compared to the contribution from the other terms,
treats the removal of aerosol from a given cell viz.
this term replaces that involving £ (x) in the previous equation, while
the last term accounts for the direct production of aerosols from all
2-5
-------
primary sources. These factors depend strongly on the local mixing
conditions and as such are terrain dependent. Therefore, these factors
could be expected to be site specific.
The next stage in the model development involves specifying the
coagulation coefficient, b(x,x"). It is assumed that the particles
interact through Brownian coagulation only. In the treatment of con-
densation, the effects due to particle size dispersion are ignored
relative to diffusional transfer of chemical species to the surface
of a particle. Furthermore, it is assumed that the only significant
mechanism for secondary source input involves the conversion of hLSO,.
The deposition coefficient, , is expressed in terms of the
deposition velocity, V(x,t), and the mixing height, H(t), as
= V(x,t)/H(t)
where V(x,t) = v(x,t) + V (x).
(2-5)
The gravitational settling velocity, "VQ(x) , of particles of effective
radius R and density p, is given by
VQ(x) = | n R3 pg/6nyR (2
where g = acceleration due to gravity and y = viscosity of air; and,
v(x,t) is related to the wind speed, U(x), by means of
v(x,t) = U(x)L(t)
where L(t) is a time factor which specifies the diurnal variation
in the mean flow.
Five different source types (traffic dust, construction dust,
point sources, stationary combustion sources and transportation) were
used to construct the source term. Furthermore, since no information
on source size distribution was available, a log normal distribution
in the mass was assumed. An empirical relationship for the time de-
pendent vertical mixing was used in the model. Advection was accounted
2-6
-------
for in a valley flow factor in the source term. A typical night-time
aerosol distribution derived from actual measurements was used for
the initial size distribution.
From the various simulation runs, the following conclusions
were drawn for Denver:
1. The ambient particle size distribution is very sensitive
to the choice of dry deposition and primary source input
rate parameters.
2. The episode aerosol is strongly source dominated with
photochemistry acting only as a minor secondary source;
hence, the submicron particles increase mainly by source
injection and by coagulation, while the large particles
are influenced by source injection, by deposition as
well as, by coagulation with combustion nuclei.
3. The wavelength dependent light scattering ability of the episode
aerosol has been shown to be a possible contribution to the
"brown cloud" effect.
The model, as presented above, is only a preliminary one. Even so,
its complexity and data input requirements make it impractical for present
application to the Phoenix area. However, this type of modeling approach
offers much hope provided the prescribed data becomes available in the
future.
2.1.3 Climatological Dispersion Model, COM
COM was also considered for application to this study. This model was
i)
developed by Busse and Zimmerman (1973). Essentially, it is a regional
model which accepts the emissions inventory in the form of gridded input.
Long term concentrations are obtained by inputting the joint frequency
distribution functions of the surface winds. Turbulence is parameterized
in terms of the standard Pasquill-Gifford Scheme. Recognition of the
diurnal variation in mixing layer height is made by an algorithm which
2-7
-------
assigns a separate height for.each stability class. Also, the variation in
the horizontal wind with height is modeled according to the wind power law.
Pollutant removal, by whatever means such as coagulation, sedimentation,
Brownian diffusion, is handled only in a gross way through an exponential
decay term, the equations describing the average concentration and relevant
parameters are in general quite complicated and are not repeated here in
great details especially since the model has been in existence for a rela-
tively long time and is'a commonly used model.
In the COM model the average concentration X, due to area sources
a - - - ~~
at a particular.receptor is given by
'16
ic /"~
X
= ii r
2- 1
66
q (6) -S : <£ (k,l,m,) S(6, z; u., P )
k A m
k=l
(2-7)
where k = index for wind direction sector
= /QUv
qk = QUv e) d6 for the k sector .
Q ( 9 » £) = emission rate of the area source per unit area and
unit time
6 = distance from the receptor to an infinitestirnal area
source
e = angle relative to polar coordinater centered on receptor
1 - index identifying wind speed class
m = index identifying Pcsquill stability category
> (k,c,m) = joint frequency function
z = height of receptor above ground level
un = representative wind speed
P = Pa squill stability category
2-8
-------
S(5,z; u?, P ) = dispersion function
For point sources, toe average concentration X due to t\ point sources
x 16 £ E £ «0 0.8 L
where o (5) = vertical dispersion function
h = effective emissions height
L = afternoon mixing height
1/2 = assumed pollutant half life, hours
2.1.4 Atmospheric Transport and Diffusion Model, ATDM
The ATDM (Culkowski and Patterson, 1976) was also reviewed for
this project. It is based on the standard Gaussian equation which
2-9
-------
has been modified to include the effect of aerodynamic roughness on
dispersion. The ATOM also models terminal and deposition velocities, in-
corporates a tilting plume for the heavy particulates, and includes an
episodic calculation of exposure maxima. Wetfall and dryfall deposition nates
are both included in the model.
Equations (2-7) through (2-10) apply to the dispersion and transport
of small particles (<^'5 m aerodynamic diameter). The equations may be
modified to account for the larger particles. These modifications will
assume that plume dilution takes place through dry deposition, and wash-
out.
Dry Deposition
Dry deposition rate is given by
D = vg S (P , z; u£,pm) ; (2-11)
where v is the deposition velocity.'
The rate of change of effective sources strength as a function of down-
wind distance from the source is
-/'
W
Equation (2-12) may be substituted in (2-11) to give
Q = Q0 exp [- V ) ^ S'( 5; z; u£, Pjj
dx' (2-131
Washout
Washout may be described by the equation:
lit = ' XQ (2-14)
2-10
-------
where x is the washout coefficient. Equation (2-14) may be integrated
to give
-A*
where the time of flight, t, is set equal to x/u.
The results given in equations (2-14) and (2-15) may be incorporated
in equation (2-7). Dry deposition may be estimated from the expression
D = FD VD Xa <2-1£)
where: FD = fraction of time in which only dry deposition occurs
VD = deposition velocity
and XQ is given by equation (2-7)
Similarly washout may be estimated by
where F = fraction of time in which both washout and dry deposition
w
are occurring
A = washout coefficient
Dry deposition and washout may be incorporated into the basic
Gaussian, equation as prescribed in equations 2-7 and 2-8.
This was actually done in ATOM. However, in its present .state the
model is most suitable for point sources and would have to be rewritten
for regional application involving aggregates of sources. This would
involve substantial amounts of time and resources and was not pursued
any further.
2.1.5 Hanna's Urban Model.
A "Simple Method of Calculating Dispersion from Urban Area Sources"
v-
was proposed by Hanna (1971). In this model, the surface concentration
2-11
-------
was assumed directly proportional to the local area source strength and
inversely proportional to the wind speed.when all source strennth are
approximately the same. Or,
X = C -^ (2-18)
where X = surface concentration
QQ = source strength -
u = wind speed, and
C is a function of atmospheric stability.
This model gave good results for sulfur dioxide simulation in the
Chicago area. It has also been applied in a recent study of the TSP
problem in Reno and Las Vegas13. Relatively high correlations be-
tween observed and predicted TSP levels were obtained, with the model tending
to over-predict measured levels-. The model is unable to account for grava-
tational settling and deposition of particulates.
2.2 'MODIFICATIONS TO COM
Even though COM was found to be most applicable for the Phoenix par-
ticulate problem, there were still some serious shortcomings which had
n
to be addressed. One study ** of high-vol filters in the Phoenix
area showed that large particles (greater than 20 microns) accounted
for roughly 70% of the particulate mass by weight. This implies that
only 30% of the particulate matter at receptor level could be treated
as a dispersive gas; and, the other 70% would have to be treated
differently. Moreover, the experimental data supported the observation
that the large particles were directly associated with local nearby
sources while the smaller particles were derived from the region as
a whole.
It was,therefore, necessary to treat emissions from local
sources in the air quality modeling effort differently from those area
2-12
-------
wide sources with smaller particle sizes which COM could represent
accurately. To facilitate this modified modeling approach, the emissions
model was altered so that it would prepare a gridded inventory for each
of four particle size ranges. The particle size ranges were selected
based on approximate cutoff points in dispersive behavior. The four
size ranges are as follows: . 0-10 microns; 11-20 microns; 21-70 microns;
and greater than 70 microns.
The COM is useful for the first two ranges governed primarily by
dispersion forces but not useful for the latter ranges where gravitation-i"!
settling becomes the dominant force. Figure 2-1 illustrates the.sett!ing
effect for different wind speeds. For the COM, the effect of gravita-
tional settling in the smaller size ranges was approximated with the
assignment of concentration decay constants. The decay rate for the
11-20 micron range is significantly larger than that of the 0-10
micron range. The assignment of decay rates for incorporation to the
COM is discussed in Section 2.3.1
Particles greater than 70 microns in size are ignored in the air
quality model. The diffusion of these particles will be determined
almost exclusively by gravity effects. Their travel distance is only
a few meters and generally not enough to impact the air quality monitors,
except for a few cases where nearby local sources may be situated very
near the monitor.
The modeling of air quality for particles in the .21-70 micron
range was accomplished by a simple rollback scheme which assumed that
concentration within a given grid is directly proportional to emissions
within that grid. In this scheme the mathematical relation is of the form
X = oE
where X is the concentration of particles larger than 20ym at a receptor
in a given grid, E the emissions of particles larger than 20ym within
the grid and a an empirical constant.
The overall air quality model now assumes the form:
X = cijC + a2E + B
2-13
-------
'f. Impeded Settling :^'xx^x:
6 8 10 12
REFERENCE WIND SPEED (mph)
FIGURE 2-1. PARTICLE SETTLING/SUSPENSION REGIMES (MRI, 1974)
2-14
-------
where: X = Total suspended participate concentrations
C = COM concentration estimates for 0-10 and
11=20 micron ranges
E = Emissions of particles >20 micron in grid containing
receptor
B = Background TSP level
ai= Empirical coefficient for COM
a2= Empirical coefficient for rollback model
The background term in the above equation derives from.tv/o different
sources. The first is due to suspended matter advected into the Phoenix
area from other regions. The second is due to natural local sources
which have been neglected from the emissions inventory.
The Carefree site showed the same 70% large particles as the other
sites, while TSP levels there were only about one third larger than those
of the background stations (42yg/m3 vs. 30yg/m3). There are some anthro-
pogenic sources near Carefree to account for the larger TSP levels,
but not enough to account for the 70% observed. Therefore, some fraction
of the background must be due to particles greater than 20 microns. Un-
fortunately, microscopy analysis^ was not performed on any station
which could be considered purely natural background, so there are no
data to indicate what percent of the background is large and what is
small. Hence, it was assumed that 50% of the background was less than
20 microns, and 50% greater than 20 microns.
With the nature of the background thus defined, and the particle size
distribution on the filters known, it is a straightforward task to
estimate ai and 02 in the above equation. Since 30% of the total sus^
pencted particulates measured are less than 20'. microns, and 50% of
the background particulates are less than 20 microns, the following
relationship must hold :
0.3X = aiC + 0.5B
2-15
-------
Also, since 70% of the measured particulates are greater than 20
microns: .
0.7X = a2E + 0.5B
and*
o! = (0.3X -0.5B)/C
o2 = (0.7X -0.5B)/E
The numerical procedure for the assignment of a, and cty and their
empirical values are presented in Section 4.2. The determination of
a, and«2 must be performed individually for each receptor, so that for
the ith receptor :
X. -al1C1>.«21E1 + B
2.3 MODEL INPUTS
The COM requires various meteorological and source emissions input data
which must be prepared in forms suitable for model application.
2.3.1 Meteorology Data
Meteorology data for the study were obtained from the National
Climatic Center (NCC) in Asheville, North Carolina. The NCC provided
both the joint frequency function and mixing height data. The joint
frequency function is a combined frequency of occurrence for three
meteorological parameters as defined by COM: six stability classes,
six wind speed classes, and sixteen wind directions. A monthly annual
(day/night) star program run was made by NCC for both 1975 and for the
entire 1973-1975 period, The 1975 data were used for model parameter-
ization (i.e., to determine the empirical a coefficients) while the
1973-1975 averaged data were used for 1980 and 1985 particulate simu-
lation.
The mixing height data were prepared from two different sets of
NCC inputs using both surface observations and upper air data. The
mixing heights were calculated form upper air data collected at Tucson
2-16
-------
(the nearest upper air station to Phoenix), and the surface observations
at Phoenix. Mean mixing heights used are shown in Table 2-2, and
mean daily temperature in Table 2-3.
2.3.2 Emissions Parameters and Pollutant Half Life
The COM requires as input the emissions and their diurnal behavior,
plume heights, source configuration; and pollutant half life.
Diurnal Assignment
The diurnal distribution of emissions must be specified in the
COM. To estimate this distribution, the five largest sources were
analyzed for their emission patterns. A weighted average of these
patterns produced a 78-22 percent day-night split for 1975 and an
81-19 percent split for 1985. The 1985 daytime figure is larger pri-
marily because of greater emissions due to construction activities fore-
cast to occur then.
Plume Height of Sources
Plume heights for all point sources were given in the NEDS data.
An assumed plume height of 10 meters was used for all area sources.
TABLE 2-2
MIXING HEIGHTS FOR PHOENIX, 1975.
QUARTER AVERAGE AFTERNOON AVERAGE NOCTURNAL
(METERS) (METERS)
1 1685 269
2 3287 463
3 4363 743
4 2688 366
Annual Average 3006 460
2-17
-------
TABLE 2-3.
MEAN DAILY TEMPERATURE AT PHOENIX
QUARTER
1
2
3
4
Annual Average
Determination of
1975
(°F)
55
75
91
63
71
Pollutant Half-life
1980-85 (Historical Data)
(6F)
56
76
88
61
71
The pollutant half-life is required for the estimation of the decay
term used in the COM diffusion model for the 10-20 ym range. Half-life
refers to the time elapsed before the, ambient concentration of a given
size particulate is reduced by one-half due to physical removal mecha-
nisms (e.g., dry deposition and gravitational settling). The following
derivation of half-life is based upon the IITRl* study in Phoenix;
however, the procedure can be readily applied to other areas. The
computational technique is based on Van der Hoven's dry deposition
formulation (given in Slade, 1968). First, it is assumed that a 15 \im
diameter particle is representative of the 10-20 pm range. Then for an
average wind speed of 2.41 m/s (annual mean value for Phoenix) and a
terminal fall speed of 1.69 cm/s (corresponding to a 15 ym diameter
particle), Van der Hoven's expression for reduction of the source
strength due to dry deposition may be used to determine the distance at
which the effective source strength has been reduced to half its orig-
inal value. The time that it takes a parcel of air, embedded in the
mean flow, to travel that distance may then be used_as the half-life for
2-18
-------
particles in the 10-20 ym size range. This half-life value may then be
used in the exponential decay term of the COM.
The results of the calculations, using the technique outlined
above, are shown in Table 2-4.
Table 2-4. Half-life for Physical Removal
Mechanism in the COM for a 15 ym
Particle and a Mean Wind Speed of 2.4 m/s.
Stability Half-life (minutes)
Ti
A 00*
B »*
C 691.2
D 62.2
E 42.2
F 27.7
*Not calculated, but can graphically be shown to be essentially infinite.
Because half-life (and the resulting decay term in COM) varies with
both stability and wind speed, the user must decide whether to use
separate values for the various wind speed/stability categories of COM
or to use a single composite value. For Phoenix, a single composite
value was used on the basis that since this is only an approximate
technique, a more complex analysis is not justified. The composite
value */as derived from a weighted average of the half-life times given
in Table 2-4.
2-19
-------
The weights used to determine the composite value were a function
of two factors: (1) the percent frequency of each stability and (2)
the relative contribution to the predicted concentration given by the
model for each stability class. The latter contribution to the weighting
term was approximated from xu/Q curves (for example, those given by
Turner, 1970). Table 2-5 gives numerical values associated with the two
factors that determine the weights. The weighting factors themselves,
given in column 4 of this table, are the product of columns 2 and 3.
Table 2-5. Annual Weighting Factors
Stability
Class
.A
B
C
D
E .
F
Annual
Frequency
of Occurrence
0.02
Q.ll
0.18
0.22
0,18
0,29
Stability Class Contribution
to Concentration at 500m
Relative to Class F
0.02
0.08
0.18
0.40
0.60
1.00
Half -life
Weighting
Factor-W.
0.0004
0.0088
0.032
0.088
0.108
0.290
The mean value for the half-life term in the decay constant was then
evaluated using
2-20
-------
6 W,
/_L_ V 1-1 ^
V VA ! ..
where the W. are from Table 2-5 and T. are the half-life values given
in Table 2-4. The inverse half-life time was used because the physical
removal mechanism in the COM is proportional to this expression (see
Section 2.1.3). Based on these data, the composite half-life was found
to be approximately thirty-seven minutes.
Emissions by Particle Size Categories
Particle size distributions of the various emission source cate-
gories are documented in a previous phase of the study (TRW, 1976b).
Owing to the general lack of information available to characterize the
particle size of the various sources, substantial uncertainty is
associated with the size distribution estimates. Figures 2-2, 2-3, and
2-4 are approximations for the particle size distributions of various
source categories, drawn as probable fits to the limited data. The
curves assembled for these plots were utilized to compile the particle
size distributions corresponding to the size regimes selected for
input to the COM. The distributions for the various sources are summa-
rized in Table 2-6.
2-21
-------
ro
i
ro
ro
100..
90..
80. _
70
o
o
e 60..
cu
1 501
10
40..
o
s_
(O
Q_
30
20
TO
0
Motor Vehicles
Aircraft
Point Sources
Area Sources
0 10 20 30 40 50 60 70 80 90 100
Percentage of all particles (by weight) less than stated size
Figure 2-2. Particle Size Distribution of Conventional Emission Sources
-------
ro
t
ro
CO
90..
80 ..
I/I
o 70
o
Vehicles
Cattle Feed Lots
Off Road Vehicles
Agricultural Tilling
010 20 30 40 50 60 70 80 90 .100
Percentage of all particles (by weight) less than stated size
Construction
Resuspension
Aggregate Storage
Figure 2-3. Particle Size Distribution of Anthropogenic Fugitive Dust Emission Sources
-------
ro
i
ro
100-j-
90..
80..
«, 70
c
o
o
= 60
S-
-------
TABLE 2-5. PARTICLE SIZE DISTRIBUTIONS
FOR VARIOUS EMISSION SOURCE CATEGORIES*
FRACTION OF ALL PARTICLES IN STATED SIZE RANGE
ro
ro
en
Source Category
0-1 Oy <20y
10-20y
30y
20-30y
70y
30-70y
ANTHROPOGENIC
Motor Vehicles
Ag. Tilling
Aggregate Sto.
Cattle Feed Lots
Off Road Vehicles
Construction
Resuspension
WIND BLOWN
Unpaved Roads
Agriculture
Undisturbed Desert
Tailing Piles
Disturbed Soils
CONVENTIONAL
Motor Vehicles
Aircraft
Point Sources
Area Sources
.41
.62
. 1.00
.41
.41
.66
.66
.68
.68
.68
1.00
.68
.91
.91
.99
.99
.41
.62
1.00
.40.
.41
.66
.66
.68
.68
.68
1.00
.68
.91
.91
.99
.99
.52
.74
1.00
.52
.52
.89
.89
.90
.90
.90
1.00
.90
.93
.93
.99
.99
.11
.12
in
.11
.23
.23
.22
.22
.22
-
.22
.02
.02
-
-
.60
.80
1.00
.60
.60
1.00
1.00
1.00
1.00
1.00
1.00
1.00
.94
.94
.99
.99
.08
.06
-
.08
.08
.11
.11
.10
.10
.10
-
.10
.01
.01
-
9.
.85
.93
1.00
.85
.85
1.00
1.00
1.00
1.00
1.00
1.00
1.00
.97
.97
1.00
1.00
.25
.13
-
.25
.25
-
_
-
-
-
-
.03
.03
.01
.01
Based on interpolation using Figures 2-3, 2-4, and 2-5.
-------
3.0 ADJUSTMENT OF AIR QUALITY DATA
In providing air quality and emissions data as input to the model,
a major objective is to obtain the most representative data set for model
parameterization. Obtaining such a data set might require several adjust-
ments of the data base. In the limit, data adjustments would be performed
as part of an iterative process to be conducted concurrently with succesive
trials of the model. However, because of significant uncertainties
associated with most of the data base, this iterative process is a costly
and impractical task. Within the scope of a practical program, opportunity
for adjustment of the data base is limited. Instead, it is more feasible
to identify the degree of representativeness of the data set, to determine
the factors affecting this representativeness, and to interpret the im-
plications of these findings-for the final air quality forecasts. The
latter approach has been adopted for the present study.
The following sections document the suitability of the data set for
use in the model, and discuss potential adjustments which should be
considered to make the data set more representative. Three major con-
siderations are involved in the selection of representative monitoring
data; 1) fieight of the monitor above ground level 2) representativeness
of site location, and 3) completeness of the data.
3.1 VARIATION OF TSP WITH HEIGHT OF MONITOR
An aspect of model parameterization related to particle size concerns
the variation of TSP with height. A clear dependence of particle size
distribution with height would suggest a variation of TSP with height
3-1
-------
as well. If the TSP variation can be clearly established, it may be
possible to adjust observed levels of TSP at different elevations to a
common reference elevation to facilitate a more meaningful parameteriza-
tion of the particulate model.
Table 3-1 shows the variation of TSP with height as measured during
the study conducted by the IIT Research Institute . For the
windy day, sampling variation of concentration with monitor height is
somewhat erratic, with no consistent trend observed for the five sites
considered. For days of more typical wind velocity, a consistent pattern
was noted at most of the stations. Concentration of TSP declined signif-
icantly with height, the average concentration of all stations decreasing
over 40% from 3 meters to 30 meters elevation. This result is consistent
with the particle size distribution variation observed for the same
monitoring conditions. Table 3-2 shows that the weight percent of parti-
cles greater than 15v in size decreases significantly with height, with
the average weight percent of these larger particles diminishing about
30% from 3 meters to 30 meters elevation.
IITRI also conducted particle size'-analysis of :hi-vol filter.
samples for two selected days in late 1975. One of these days was
characterized by typical low wind speeds, while the other (September 27)
was characterized by substantial gusts and an average wind speed of
9.8 mph. The overall observed size distributions (Table 3-3) were found to
be consistent in some respects with those obtained by the Anderson measure-
ments shown in Table 3-2. For example, a substantial portion of the par- .
ticulate mass (nearly 70%) is comprised of particles of 20y diameter or
more. However, the hi-vol microscopy examination indicates no clear dif-
ferences in particle size distributions for the windy and calm days, and
there appears to be no significant variation of particle size distribution
with monitor height (as noted previously with the Anderson measurements).
The presence of substantial portions of larger particles on the
hi-vol filters and Anderson samplers, both at lower and higher elevations,
indicates the presence of local source influence at each of the various
sampling stations. Variability of the particle size distributions from
one monitor to another may also be due in part to local source influences.
In many cases, the monitor is in the plume of these sources. Because
the effect of local sources at a single receptor point is likely to be
a-2
-------
TABLE 3-1. TSP AS FUNCTION OF ELEVATION AT VARIOUS MONITOR SITES .(UTRI, 1976)
TOTAL PARTICULATES (SUM OF ANDERSON MEASUREMENTS)
HEIGHT
ABOVE
GROUND
3m
10m
30m
WINDY CONDITIONS
NOV. 18, WIND = 4m/sec.
AMER INDIAN
GRAD RESERVA-
SCHOOL PARKER PAGE TION MESA
163 96 110 59 131
96 35 52 84 269
79 78 82 216 70
AVERAGE
112
107
105
CALM
NOV. 17,21,25 WIND = Im/sec.
AMER INDIAN
GRAD RESERVA-
SCHOOL PARKER PAGE TION MESA
173 133 142 133 165
93 116 99 86 55
76 51 89 99 81
AVERAGE
149
90
79
CO
I.
CO
-------
TABLE 3-2. PARTICLE SIZE GREATER THAN 15y FOR SAMPLES AT SELECTED MONITOR
SITES IN PHOENIX, NOVEMBER 17, 18, 21, 25 of 1975 [4]
PERCENTAGE OF PARTICLES (BY WEIGHT) GREATER THAN 15y
HEIGHT
ABOVE
GROUND
3m
10m
30m
TOTAL
PARTICLES
(cm)
WINDY CONDITIONS
NOV. 18, WIND = 4m/sec.
\MER INDIAN
3RAD RESERVA-
SCHOOL PARKER PAGE TION MESA
64 58 29 56 59
65 29 85 51 58
65 51 73 62 59
211 164 * * *
AVERAGE
54
58
62
187
CALM
NOV. 17,21, 25, WIND = Im/sec.
AMER INDIAN
GRAD RESERVA-
SCHOOL PARKER PAGE TION MESA
44 54 42 34 67
44 28 44 26 38
34 29 37 48 29
247 196 266 203 212
AVERAGE
48
34
34
225
CO
-------
TABLE 3-3. PARTICLE SIZE DISTRIBUTION FOR SUSPENDED PARTICULATES
MEASURED IN PHOENIX SEPT. 27 & NOV 14, 1975 [4]
PERCENTAGE OF PARTICLES (BY WEIGHT) IN SIZE RANGE
SEPT 27 (WINDY)
MONITORS AT 20 FEET
MONITORS AT 15 FEET
MONITORS AT 5 FEET
NOV 14 (CALM)
MONITORS AT 20 FEET
MONITORS AT 15 FEET
MONITORS AT 5 FEET
<2y
.07
.06
.05
.14
.14
.13
2-8y
2.8
2.2
2.0
2.8
2.5
2.7
8-20y .
32.2
33.0
28.3
32.4
29.5
34.9
>20p
65.0
64.6
69.7
64.7
67.9
62.6
co
en
-------
highly variable, an average behavior at any one monitor can only be
leiernined by extensive sampling and analyses over a significant
ceriod of time. In this study, the data base is probably too small to show.
any systematic variation of particle size distribution with elevation. It
will not be feasible, therefore, to adjust air quality measurements at
different elevations to a common level of representativeness. Parametri-
zation of the model must, therefore, be performed with the actual observed
values of TSP, whatever the elevation of the observation. Deviations of
the observed levels of TSP with the forecasted model values (calculated
for a single elevation at all receptors) may be due in part, to actual TSP/
height relationships effective at the various monitor sites.
3.2 REPRESENTATIVENESS OF MONITOR ENVIRONMENT
The monitor site review of this study (TRW, 1976a) revealed that
air quality at some of the monitor sites was not representative of air
, <.
quality of the general area surrounding the site. Instead, these sites
were influenced by local sources in a manner atypical of the general
area, and, therefore, may be only representative of '"site specific air
quality." Table 3-4 summarizes these latter sites which were determined
after reviewing the air quality relative to surrounding sources. Those
sites which are representative only of site specific air quality may be
deleted from the data base, or they may be included if the significance
of the local intervention can be assessed. ','.
3-6
-------
TABLE 3-4. SUMMARY OF POTENTIAL MONITOR SITES WITH ONLY SITE SPECIFIC REPRESENTATIVENESS
MONITOR SITE
CHARACTERIZATION
OF GENERAL AREA
SITE SPECIFIC SOURCES
SOURCES IN GENERAL
AREA
PROBABLE SIGNIFICANCE OF
ATYPICAL LOCAL SOURCES
St. Johns
Rural/Residential.
Indian Reservation
in open desert.
Soil dust from unpaved roads,
residence yards, and open
fields.
Soil dust from unpaved
roads.
Significant Impact on
monitors measurements.
North Phoenix
Suburban/Residential.
Soil dust from unpaved road-
ways and unpaved parking lots.
Soil dust from unpaved
roadways and vacant lots.
Doubtful 1f significance
of site specific source
can be determined without
special study. Measure-
ments may have to be deleted
from data base.
Mesa
Suburban/Res 1denti al-
Commercial.
Soil dust from unpaved road-
ways and parking lots, soil
yards.
Soil dust from unpaved
parking lots and road-
ways, disturbed vacant
lots.
Impact of site specific
sources probably significant,
but special study needed to
assess the degree of Impact.
Downtown
Phoenix
Urban/Commercial.
Soil dust from unpaved roads
and parking areas, motor
vehicle exhaust.
Soil dust from unpaved
roads and parking areas,
motor vehicle exhaust.
Impact of site specific
sources probably significant,
but special study needed to
assess the degree of Impact.
-------
3.3 COMPLETENESS OF TSP DATA
Because of the highly variable meteorology throughout the year in
Phoenix, concentrations of TSP also vary substantially. If measurements
are incomplete for a significant period of the year, the calculated
geometric means may be significantly biased from the acutual mean. Table
3-5 shows the completeness of measurements conducted during 1975 for
each of the monitor sites. The sites lacking complete measurements were
characterized by absence of data from the second, third and fourth
quarters. The general pattern of TSP quarterly variation in 1975 showed
TSP minima in the first quarter and maxima in the last quarter. This
trend was, however, somewhat indefinite, and it is unlikely that it
held at all stations. Because no definite quarterly pattern of TSP can
be clearly assigned to any given monitor station (historically the
variation changes dramatically due to annual meteorology fluctuations),
it is not possible to evaluate the bias of the available data .due to the
data gaps.
3.4 SUMMARY OF BIAS OF AIR QUALITY DATA
Table 3-6 summarizes the impact of various influence factors on
the representativeness of air quality measurements made at the different
monitor sites in 1975. The factor of greatest impact on observed TSP
levels is monitor height. Nine of the 16 monitors recording in 1975
are situated at elevations exceeding the five foot reference height.
TSP levels reported from these nine monitors tend to underestimate the
true exposure levels at the five foot level. The factor of next greatest
concern for TSP levels is data completeness. Five monitor sites exhibit
substantial data gaps for nearly three fourths of the year, placing the
measured values reported from these stations in serious dcubt. The bias
from the data gaps is indeterminate, and probably significant. Four
of the sites experience a bias toward high readings because of atypical
site specific sources affecting TSP there. The purpose of the deter-
mination of probable impact of the various factors on representativeness
of air quality data is for interpretation of the final air quality fore-
casts. Because data are not available to quantify the impact of these
3-8
-------
TABLE 3-5. COMPLETENESS OF TSP MEASUREMENTS FOR PERIOD OF 1975
NUMBER OF MEASUREMENTS IN 1975
MONITOR
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Central Phoenix
South Phoenix
Arizona State
Glendale
West Phoenix
North Phoenix
Scottsdale/Paradise
Scottsdale
Mesa
Downtown Phoenix
St Johns
Sun City
Paradise Valley
Carefree
Chandler
Guadalupe
Litchfield
1ST QTR.
10
9
15
9
0
7
14
15
15
5
15
13
12
14
13
14
14
2ND QTR.
14
8
15
9
0
12
10
13
14
0
15
0
0
13
0
12
11
3RD QTR.
15
11
15
5
0
14
13
15
14
0
15
0
0
14
0
12
0
4TH QTR.
15
12
11
.9
0
14
16
11
9
0 ..
14
o
0
14
0
15
0
co
1
to
-------
TABLE 3-6. MATRIX OF PROBABLE IMPACT OF FACTORS INFLUENCING REPRESENTATIVENESS
OF 1975 TSP DATA
CO
_J
o
MONITOR
2
3
4
5
7
8
9
10
11
12
13
14
15
16
17
18
Central Phoenix
South Phoenix
Arizona State
Glendale
North Phoenix
Scottsdale/Paradise
Scottsdale
Mesa
Downtown Phoenix
St. Johns
Sun City
Paradise Valley
Carefree
Chandler
Guadalupe
Litchfield
PROBABLE BIAS OF INFLUENCE
HEIGHT OF MONITOR3'
Low
Low
Low
Low
Low
Low
Low
Low
- -
'
Low
__^
FACTOR ON ANNUAL GEOMETRIC
REPRESENTATIVENESS
OF SITE ENVIRONMENT
___
.
High
High
High
High
. . .
'
.
MEAN MEASURED AT STATION
COMPLETENESS OF DATA
.
- '
Unclear
---
Unclear
Unclear
___
Unclear
Unclear
a.
A height of 5 feet is assumed as reference height.
-------
factors, it is not possible to adjust the air quality data to more repre-
sentative figures. Instead, the qualitative assessment here will be em-
ployed to indicate areas where the model parameterization and forecasts
should be used with qualifications as to their representativeness.
3-11
-------
4.0 MODEL PARAMETERIZATION
The air quality model chosen for this study was discussed earlier
in Section 2.0. Quantification of the empirical constants in the model
is discussed in this Section. In particular, the background term B is
discussed in Section 4.1 and the coefficients a-,, and a.?- in Section 4.2,
Before assigning numerical values to .the constants, it is useful
to explain the computer system used in performing the emissions and .
air quality modeling. Figure 4-1 is a schematic diagram portraying
the development of the parameterized model. In the first step, the
Emissions Simulator Program (TRW, 1976b) produces both a printed output
of total emissions together with a graphical disaggregation of emissions.
Next, COM simulations are made using, the 0-lOpm and ll-20ym emissions
data as well as the appropriate meteorology. The 21-70 micron particle
emissions are not used until the final step. The COM output and the
emissions in the 21-70 micron range are input to a parameterization
program, which is discussed in Section 4.2. The final product is a
parameterized TSP model capable of simulating future air quality given
emissions and meteorology.
4.1 BACKGROUND LEVELS OF TSP
Four monitoring sites, sufficiently removed from urban Phoenix,
were chosen to determine background levels. This means that only the
natural sources in the area affect the readings at these sites, plus
whatever suspended particulates are transported from other areas.'. For
the-purposes of this study, a background value was interpreted and
used as if there were no means of controlling it.
Background levels for 1973, 1974 and 1975 ere shown in Table 4-1.
The historical data shown for 1973 and 1974 wer^ jsed to determine a
weighted average for 1975 which lacked any TSP c*-.-a at Grand Canyon
3
and Petrified Forest. A weighted average of 2r. .-- per m was used for
1975 as well as for 1985, the projection year.
4-1
-------
Raw Data
for Emission
Catagories
Emissions Simulator
(Produce Emissions Grid)
Meteorology Data
Printed Output
of Emissions
Binary Output
of Emissions/grid
for 4 Particle Sizes
I
Emissions for
0-10 and 11-20
micrometer I
range i
COM
(Calculate TSP for
0-10 and 11-20 micro-
meters). -
Decay Constant
for 11-20
Micrometer
Range
1
Emissions for
21-70-
micrometer ',
range
Parametrlzation
(Assign Empirical
Coefficients and
Background)
Alii.Quam;
Estimates
Figure 4-1 Computer Modeling System
4-2
-------
TABLE 4-1. BACKGROUND LEVELS OF TSP
SITE
Grand Canyon
Petrified Forest
Organ Pipe
Monte zuma
Average
1973
(yg/m3)
22
26
34
28
28
1973
17
23
23
27
23
1975
(yg/mj)
N/A
N/A .
31
34
32
4.2 ASSIGNMENT OF EMPIRICAL CONSTANTS
The air quality model was presented earlier in Section 2.1, but
is restated here for convenience. The basic equation is
Xi = "11 Ci + *2i Ei + B ' ,
, (4-1)
where
X.j = Total suspended particulate concentration
C. = COM calculated concentration of 0-10 and 11-20 micron
particles
E.J = Emissions of particles >20y in the grid square of the
receptor
B = Background TSP
a-j.j= Empirical coefficient related to COM
a2i= empirical coefficient related to rollback model
i = denotes the receptor under consideration
The constants a-,, and .do,- are defined by:
aii= (0.3 X1 - 0.5 B)/Ci
ot2i= (0.7 X. - 0.5 B)/Ei (4-2)
4-3
-------
Table 4-2 summarizes the results of a computer run for 1975, in-
cluding the emissions model and COM for the two small particle size
ranges. Columns 1,2, and 3 contain the COM simulations based on emis-
sions from small particles, column 4 the actual observed air quality, and
column 5 the emissions of particle 21-70 microns within the grid square
of each receptor. Columns 7 and 9 are the contribution of TSP from
particles 0-20y in size and from particles 21-70y in size, respectively..
The coefficients a,, and o^. are shown in columns 6 and 8 and are computed
from equation.4-2 above. X.. is shown in column 11 and C. and E. are in
columns 3 and 5, respectively.
A plot of observed TSP levels versus those levels predicted by the
COM model is shown in Figure 4-2. The observed levels are defined to
be the concentration of sub-20 micron particles measured at the monitor
sites, or 30% of the TSP. A linear regression of the plot of Figure 4-2
yields the equation y = 20 + .49 X, where y is the observed level and X
the CDM-predicted concentration for particles 0 to 20 micron size. The
intercept (20) is found to be relatively close to the background level
assumed for sub-20 micron particulates (15), and the slope (.49) is within
the range of "usual" calibration for COM predicted pollutant concentrations,
This indicates the modified COM treats the diffusion behavior of 0-20
micron particles reasonably well.
There are several factors which may cause poor correla-
tion of model predicted values with observed values. First, there is
probable bias of the observed values for true representative concentra-
tions due to variations in monitor height, completeness of data, and
representativeness of the monitor site environment. Second, there is
probable bias in the emissions data base due to numerous uncertainties
underlying the development of the fugitive dust emissions inventory.
Finally, there are limitations associated with the assumptions of the
model itself. While the implications of any one particular limitation
on the predictability achieved by the model may be assessed, the simul-
taneous intervention of many influence factors known to be affecting
the model results make any attempt to explain the variations unfeasible.
4-4
-------
TABLE 4-2 EMPIRICAL COEFFICIENTS DETERMINED FOR PHOENIX COM/ROLLBACK MODEL
10 11
AIR QUALITY RECEPTORS
2 C. Phoenix
3 S. Phoenix
4 Arizona St.
5 Glendale
7 N. Phoenix
8 N. Scott/Par Va.
9 Scottsdale
10 Mesa
11 Downtown
12 St. Johns
13 Sun City
14 Par. Valley
15 Carefree
16 Chandler
COM
0-1 On
' (ug/m3)
43.4.
23.5
45.3
35.1
39.5
33.0
40.3
34.6
51.0
14.1
29.0
38.8
7.7
23.9
COM
ll-20u
(ug/m )
6.4
2.3
7.1
4.4
6.1
5.4
5.7
5.0
7.4
1.0
3.6
6.4
0.6
3.0
1C
CDIT
(ug/m3)
49.8
25.8
52.4
39.5
45.6
38.4
46.0
39.6
58.4
15.1
32.6
45.2
8.3
26.9
OBSERVED
(ug/m3)
112
144
169
101
121
149
115
117
200
145
88
184
42
119
EMISSIONS IN GRID a,.
(21-70u) n
(tons/day)
3.
1.
4.
2.
>3-
2.
3.
, 3.
4.
1.
4.
3.
0.
3.
69
73
35
59
81
97
71
28
35
66
74
92
71
75
0.37
1.09
0.68
0.39
0.47
0.77
0.42
0.51
0.77
1.89
0.35
0.89
-.29
0177
°2i Ci
(ug/m3)
18.6
28.2
35.7
15.3
21.3
29,7
19.5
20.1
45.0
28.4
11.4
40.2
-2.4
20.7
"21 "21 Ei
' ug/m3 3
T/day (ug/m
17.18 63.4
49.60 85.8
23.75 103.3
21.51 55.7
18.29 69.7
30.07 89.3
17.65 65.5
20.40 66.9
28.74 125.0
52.11 86.5
9.83 46.6
29.03 113.8
20.28 14.4
18-21 68.3
B y^
)( ug/m3) (yg/m3)
30
30
30
30
30
30
30
30
30
30
30,
30
30
30
112
144
169
101
121
149
115
117
200
145
88
184
42
119
I
U1
-------
80-
70.
OJ
+J
(D
O
t-CO
4-> E
IQ O!
Q- 3-
t- f
O)
C >
O QJ
i- r
U
T- Q-
t/) O
f)
«*-
O 0)
JO
\f>
O
O 3
(U
-------
The coefficient a^. has an average value of 27. This is a dimen-
sional coefficient, unlike a,, which is nondimensional, and cannot be
expected to assume a value of unity. Also there is no distinct pattern
for agj as there was for a^, probably because ag-j reflects the influence
of local emissions on TSP. This influence is undoubtedly different for
each grid square because of the numerous variations for local source
distributions around a given monitor. Figure 4-3 illustrates the scatter
of data for observed versus calculated levels of particulates greater
than 20 micron in size.
It can be seen that the most significant contribution to the COM
prediction of concentration is from the 0-10 micron range, while the
11-20 micron range contributes nearly an order of magnitude less (col-
umns 1, 2, and 3 of Table 4-2). This difference is due to the shorter
half-life for larger particles (see Section 2.3). Additional analysis
showed that COM predicts neligible contributions to air quality from
the 21-70 micron particles.
The empirical constants shown in Table 4-2 were the ones actually
used for the air quality projections which are discussed next.
4-7
-------
140 .
E
"c5> 12° -
c
_^3 ^^^
1 l/»
j-'ol
SI 100 -
O)
i-0-
O 00
0>«f-
+-» 0
|g . 8°- -
+J OJ
J- .0
Q. O
4->
°"S 60 -
c e
0 3
^ l/>
*^
C J-
* * /in
o -iJ 40 -
C O)
O E
c_> C
J- O
Ol S- on
M u 20 -
^ 1
oz:
o
CM
0
LEGEND FOR MONITOR SITES:
2 CENTRAL PHOENIX
3 SOUTH PHOENIX
4 ARIZONA STATE
- 5 GLENDALE !!
7 NORTH PHOENIX
8 N, SCOTTSDALE/PARADISE V,
9 SCOTTSDALE . *14 .
10 MESA
- 11 DOWNTOWN PHOENIX -d
12 ST, JOHNS *
13 SUN CITY
14 PARADI-SE V,
15 CHANDLER *12 *8
3
7
^ . * . . . ...
10. ,|.5
13
5
Figure 4-3.
' 1 2.3 4 5 6
Total Emissions of Particulates Greater Than 20 Micron Diameter in Grid
Square Enclosing Monitor Site, gm/sec.
Observed Concentration of Particulates Greater Than 20 Micron
Diameter in Grid Square
'4-8
-------
. 5.0 FORECASTED BASELINE TSP LEVELS FOR 1975 AND 1985
s
' The baseline emission levels corresponding to the baseyear and
1985 are translated into air quality forecasts using the source re-
ceptor relationship discussed previously. The model is used to evaluate
contributions of each of the source categories to TSP levels, and the
impact of source changes on air quality. The 1975 TSP simulation is dis-
cussed in Section 5.1 and the 1985 forecast in 5.2.
5.1 BASEYEAR TSP LEVELS
Table 5-1 shows the domination of 1975 TSP levels by the four
major emission sources that year. The model predicts that nearly all
the TSP level (excluding background) at 12 of the 13 sites monitoring
in 1975 was caused by emissions from unpaved roads, entrained street
dust, construction activities, or wind erosion. The exception was the
Sun City site where off-road vehicles were responsible for most of the
TSP levels. Monitors which were most dramatically affected by wind
erosion emissions tended to be located in the rural areas under develop-
ment, such as the Paradise Valley and North Scottsdale/Paradise Valley
sites. Other sites within cities were also significantly affected by
wind blown dust emissions. These sites were generally surrounded by
numerous vacant lots and/or dirt residence yards. Entrained dust also
had an impact on urban sites. The sites at Central Phoenix, Arizona
State, North Phoenix, Scottsdale, Mesa and Downtown Phoenix were more
affected by dust entrained off streets than any other-single source.
Emissions from unpaved roads contributed significantly to TSP at each
of the sites, but were particularly dominant at the South Phoenix, St.
Johns, and Chandler sites.
5.2 PROJECTED BASELINE TSP LEVELS
Air quality forecasts were made for 1985 using the projected
emissions and annual daily meteorology. The projected emissions were
based on anticipated developments in 1985. These forecasts are shown
for each of the monitoring locations in the study area in Table 5-2.
5-1
-------
TABLE 5-1. IMPACT OF. MAJOR SOURCES ON TSP LEVELS
CONTRIBUTION OF SUSPENDED PARTICULATES
FROM FOUR' MAJOR SOURCES (ug/mj) Dp
MONITOR SITE
Central Phoenix
S. Phoenix
Arizona State
Glendale
N. Phoenix
N.Scotts/Para. V.
Scottsdale
Mesa
Downtown
St. Johns
Sun City
Paradise Valley
Chandler
TSP
IN 1975,
112
144
169
101
121
149
115
117
200
145
88
184
119
UNPAVED
ROADS
25
75
35
30
26
. 24
27
32
42
93
15
42
64
ENTRAINED
DUST
31
20
59
17
28
8
33
35
70
2
12
14
10
ff
CONSTRUCTION WIND »;
ACTIVITIES EROSION
.4
2
7
7
7
14
6
8
8
0
3
17
7
19
15
33
15
28
71
16
10
40
-, 18
2
78
5
:RCENTAGE OF TSP LEVEL
JNTRIBUTED FROM FOUR
UOR SOURCES & BACKGROUND
96.3
98.2
96.4
97.2
97.8
98.3
96.5
97.7
94.1
98.3
55.2
98.1
96.6
en
i
ro
-------
TABLE 5-2. FORECASTED IMPROVEMENT IN TSP LEVELS DUE TO ANTICIPATED DEVELOPMENT IN THE PHOENIX AREA
SUSPENDED PARTICIPATES, ug/nT
cn
i
CO
T575
Observed
MONITOR SITE (TOTAL)
C. Phoenix
: S. Phoenix
Arizona St.
Glendale
N. Phoenix
N. Scott/Paradise
Scottsdale
Mesa
Downtown
St. Johns
Sun City
Paradise Valley
Chandler
112
144
169
101
121
149
115
117
200
145
88
184
119
RBI
Forecast
(TOTAL)
87
101
132
65
83
101
93
95
155
157
74
93
160
Percentage Reduction
in TSP Unpaved
1975 to 1985 1975
22.3
29.8
21.9
35.6
30.4
32.2
19.1
18.8
22.5
-8.3
15.9
49.4
-34.5
25
75
35
30
26
24
27
32
42
93
15
42
64
Roads
1985
8
32
12
9
8
32
10
13
15
116
6
14
91
Resuspension
1975 1985
31
20
59
17
28
8
33
35
70
2
12
14
10
37
24
68
20
32
9
42
45
82
0
17
17
12
Construction
1975 1985
4
2
7
7
7
14
6
8
8
2
3
17
7
5
9
9
2
7
25
5
4
10
16
25
23
Percentage of TSP Contributed
by 3 major sources and background
1975 1985
80
88
78
83
75
51
83
90
75
66
68
56
93
92
94
90
94
93
95
94
97 .
89
96
93
93
97
-------
Significant improvements in air quality occur at 11 of the 13 monitoring
sites. Baseline 1985 TSP levels are from 16 to 50% less than 1975 levels,
depending on the site. Two of the 13 sites are forecasted to attain the
primary air quality standard (75 pg/m ). In many cases, a significant
portion of the air quality gains.over baseyear levels is due to base-
line development planned for the area. This development will change the
distribution of emission sources, eliminate local sources near the monitors,
and diminish the magnitude of many sources. Although total dust emissions
from unpaved roads are expected to increase slightly from 1975 to 1985,
the distribution of these emissions changes substantially, such that they
are more widely spread in the rural areas, and greatly reduced in the
city areas. Wind erosion emissions are estimated to decrease greatly
in 1985 due to 1) a decrease in wind erosion sources (i.e., vacant
property), and 2) the expectation of typical meteorology in 1985 based
on historical averages (the historical data show 1975 to be relatively
more windy than other years). Contributions to TSP from entrainment
of street dust are expected to increase slightly by 1985, especially at
monitors located within the city areas. As a result of the net changes
in emission source magnitudes and distribution, TSP levels will decrease
significantly at 11 of the 13 monitoring sites under consideration,
(Table 5-2).
5-4
-------
REFERENCES
1. Berman, N.S. and DeLaney, J.R. (1975) : Atmospheric Modeling for
Phoenix, Arizona, Arizona State University, Tempe, Arizona
ERC-R-75009.
2. Busse, A.D. and Zimmerman, J.R.. (1973): User's Guide for the Climato-
logical Dispersion Model, U.S. Department of Commerce. NTIS PB227346.
3. Heffner, J.L.: Taylor, A.D. and Ferker, G. (1975): A Regional-
Continental Scale Transport, Diffusion and Deposition Model - Part 1:
Trajectory Model and Part II: Diffusion-Deposition Models U.S. Depart-
ment of. Commerce. NTIS COM-75-11094
4. IITRI (1976): "Field Air Sampling Study -- Phoenix, Arizona". Pre-
pared by R.H. Snow, R. G. Draftz and J. Graf of IIT Research Institute
for the U. S. Environmental Protection Agency, Contract No 68-01-3163,
Task No, 5, April.
5. Massor, C. (1976), Appendix.B* Factors Affecting Atmospheric Trans-
port of-Fugitive Dust". Sent by D. Safreit to G. Richard.
6. Middleton, P.B. and Brock, J.R. (1975): Atmospheric Aerosol Dynamics: .
The Denver Brown Cloud (to be published)
7. Midwest Research Institute, "Development of Emission Factors for
Fugitive Dust Sources", prepared for the Environmental Protection
Agency, June 1974.
Pasquill, F. (1971): Atmospheric Diffusion, John Wiley & Sons, 2nd
Edition, New York, 429 pp.
9. Personal Communication with Jean Graf of IIT Research Institute about
work performed under EPA Grant No. R803078-02-0, June, 1976.
10. Regional Air Pollution Study (St. Louis)
11. TRW, 1976a: An Implementation Plan for Suspended Particulate Matter
in the Phoenix Area. Air Quality Review : Prepared by TRW Environmental
Engineering Division for the U.S. Environment Protection Agency,
November.
12. TRW, 1976b: An Implementation Plan for Suspended Particulate Matter
in the Phoenix Area. Air Emissions Inventory. Prepared by TRW Environ-
mental Engineering Division for the U.S. Environmental Protection
Agency, November.
13. PEDCo - Environmental, "Nevada Particulate Control Study for Air
Quality Maintenance Areas," Prepared for U.S. Environmental Protection
Agency, 1976.
6-1
-------
14. Slade, D.H. (Ed.), Meteorology and Atomic Energy 1968, U. S. Energy
Research and Development Administration, TID-24190, July 1968.
15. Ragland, K.W., "Multiple Box Model for Dispersion of Air Pollutants
from Area Sources", Atmospheric Environment 7_t 1017 (1973).
16. Hanna, S.R. A Simple Method of Calculating Dispersion from Urban
Area Sources. J. Sir Poll. Cont. Assoc., 12_, 774-777 (Dec. 1971).
17. Culkowski, W.M. and M.R. Patterson, 1976: A Comprehensive Atmospheric
Transport and Diffusion Model. ORNL/NSF/EATS-17.
6-2
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO. 2.
EPA 450/3-77-021 c
4. TITLE AND SUBTITLE
An Implementation Plan for Suspended Parti
in the Phoenix Area, Volume III, Model Sim
Total Suspended Particulate Matter
7. AUTHOR(S)
George Richard, Jim Avery, Lai Baboolal
9. PERFORMING ORGANIZATION NAME AND ADDRESS
TRW
Environmental Engineering Division
One Space Park
Redondo Beach, California
12. SPONSORING AGENCY NAME AND ADDRESS
U. S. Environmental Protection Agency
Office of Air Quality Planning and Standar
Research Triangle Park, N.C. 27711
3. RECIP
5. REPO
rulat.P Matter Au
ulation of 6-PERF<
8. PERF
10. PRO
11. CON
13. TYP
ds 14-SPO
lENT'S ACCESSION-NO.
RT DATE
gust 1977
DRMING ORGANIZATION CODE
3RMING ORGANIZATION REPORT NO.
GRAM ELEMENT NO.
TRACT/GRANT NO.
68-01-3152
E OF REPORT AND PERIOD COVERED
Final
MSORING AGENCY CODE
200/04
15. SUPPLEMENTARY NOTES Volume I, Ai r Quality Analysis - EPA 450/3-77-021 a; Volume II,
Emission Inventory - EPA 450/3-77-021 b; Volume III, Model Simulation of Total
Suspended Particulate Matter Levels - EPA 450/3-77-021 c; Volume IV, Control Strategy
is. ABSTRACT Formulation - EPA 450/3-77-021d.
This document is one volume of a four-volume report presenting an implementation
plan for control of suspended particulate matter in the Phoenix area.
17. ' KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
Particulate Matter
Total Suspended Particulate
Emission Sources
Control Methods
Fugitive Dust
Air Quality Measurements
Modeling
18. DISTRIBUTION STATEMENT
Release Unlimited
b.lDENTIFIERS/OPEN END!
19. SECURITY CLASS (This
Unclassified
20. SECURITY CLASS (This
Unclassified
ED TERMS c. COSATl Field/Group
Report) 21. NO. OF PAGES
59
page) 22. PRICE
EPA Form 2220-1 (9-73)
------- |