EPA/600/R-99/030

Compute Motor Vehicle Emissions Estimates

The mobile source emission processor computes day-specific, gridded, hourly motor vehicle
emissions estimates of VOC total organic gases (TOG), carbon monoxide (CO), and oxides of
nitrogen (NOX). The actual emission estimate computations performed and maintained in the
MEPPS data base are as follows:

C TOG for gasoline (noncatalyst/catalyst composite) evaporative diurnal processes;

C TOG for gasoline (noncatalyst/catalyst composite) evaporative nondiurnal processes;
C TOG for gasoline (noncatalyst/catalyst composite) exhaust operations;
C NOX for gasoline (noncatalyst/catalyst composite) exhaust operations;
C CO for gasoline (noncatalyst/catalyst composite) exhaust operations;
C TOG for diesel (noncatalyst/catalyst composite) exhaust operations;
C NOX for diesel (noncatalyst/catalyst composite) exhaust operations; and
C CO for diesel (noncatalyst/catalyst composite) exhaust operations.

The mobile emission processor computes motor vehicle emissions estimates based on three types
of VMT files:

C On-network VMT;
C Off-network VMT; and
C Public land survey quarter sections VMT.

The differences in the VMT types are:

C On-network VMT are network link-specific;

C Off-network VMT are county/area type/facility type-specific; and
C Public land survey quarter sections VMT are county/polygon/area type/facility type-
specific.

The mobile-emission processor estimates motor vehicle emissions for all combinations of VMT.
Regardless of the VMT type that is used to generate motor vehicle emission estimates, all motor-
vehicle emission estimates have some degree of uncertainty. However, the most desirable motor-
vehicle emissions estimates usually are generated from on-network, link-specific data, followed
by public land survey quarter sections data, and finally off-network data. Motor vehicle
emission estimates that are generated from on-network data are not more certain, but that they are
more spatially representative.

Before the mobile-source emission processor computes the motor vehicle emission estimates, it
computes a fleet composite emission factor. Equations 4-44 through 4-67 show how the
processor computes the fleet composite emission factors. In all cases, the emission factors are
extracted from the Mobile lookup tables based on the appropriate indices. Please refer to the

4-58
EPA/600/R-99/030

section on generating the Mobile emission factors lookup tables earlier in this section for
additional information.

If the user has provided non-I/M-specific VMT, the mobile source emission processor computes
the fleet composite emission factors by vehicle classification for temperature and speeds
combinations through the application of Equations 4-35 through 4-42.

hcdifs,t = Gjhcdij,s,t * vmixj (4-35)

hcexfs,t = Gj hcexj,s,t * vmixj (4-36)
noxfs.t = Gjnoxxj,s,t * vmixj (4-37)
coexfs,t = Gj coexj,s,t * vmixj (4-38)
hcevfs,t = Gj hcevj,s,t * vmixj (4-39)
dhcexfs,t = Gj dhcexj,s,t * vmixj (4-40)
dnoxfs,t = Gj dnoxxj,s,t * vmixj (4-41)
dcoexfs.t = Gj dcoexj,s,t * vmixj (4-42)

where: j is the vehicle classification index

s is the speed index
t is the temperature index
hcdif is the hydrocarbon diurnal fleet composite emission factor for all vehicle types
(grams/mile)
hcdi is the non-I/M hydrocarbon diurnal emission factor (grams/mile)
vmix is the vehicle mix profile
hcexf is the hydrocarbon exhaust nondiurnal fleet composite emission factor for
gasoline vehicles (grams/mile)
hcex is the non-I/M hydrocarbon exhaust nondiurnal emission factor for gasoline
vehicles (grams/mile)
noxf is the oxides of nitrogen exhaust nondiurnal fleet composite emission factor for
gasoline vehicles (grams/mile)
noxx is the non-I/M oxides of nitrogen exhaust nondiurnal emission factor for
gasoline vehicles (grams/mile)
coexf is the carbon monoxide exhaust nondiurnal fleet composite emission factor for
gasoline vehicles (grams/mile)
coex is the non-I/M carbon monoxide exhaust nondiurnal emission factor for
gasoline vehicles (grams/mile)
hcevf is the hydrocarbon evaporative nondiurnal fleet composite emission factor for
gasoline vehicles (grams/mile)
hcev is the non-I/M hydrocarbon evaporative nondiurnal emission factor for gasoline
vehicles (grams/mile)
dhcexf is the hydrocarbon exhaust nondiurnal fleet composite emission factor for
diesel vehicles (grams/mile)
dhcex is the non-I/M hydrocarbon exhaust nondiurnal emission factor for diesel
vehicles (grams/mile)

4-59
 EPA/600/R-99/030

dnoxf is the oxides of nitrogen exhaust nondiurnal fleet composite emission factor
for diesel vehicles (grams/mile)
dnoxx is the non-I/M oxides of nitrogen exhaust nondiurnal emission factor for
diesel vehicles (grams/mile)
dcoexf is the carbon monoxide exhaust nondiurnal fleet composite emission factor
for diesel vehicles (grams/mile)
dcoex is the non-I/M carbon monoxide exhaust nondiurnal emission factor for diesel
vehicles (grams/mile)

If the user has provided I/M-specific VMT data, the processor computes the fleet composite
emission factors by vehicle category for temperature and speed combinations through the
application of Equations 4-43 through 4-50.

hcdifs,t = Gj ihcdij,s,t * vmixj (4-43)

hcexfs,t = Gj ihcexj,s,t * vmixj (4-44)
noxfs,t = Gj inoxxj,s,t * vmixj (4-45)
coexfs,t = Gj icoexj,s,t * vmixj (4-46)
hcevfs,t = Gj ihcevj,s,t * vmixj (4-47)
dhcexfs,t = Gj idhcexj,s,t * vmixj (4-48)
dnoxfs,t = Gj idnoxxj,s,t * vmixj (4-49)
dcoexfs,t = Gj idcoexj,s,t * vmixj (4-50)

where: ihcdi is the I/M hydrocarbon diurnal emission factor (grams/mile)

ihcex is the I/M hydrocarbon exhaust nondiurnal emission factor for gasoline
vehicles (grams/mile)
inoxx is the I/M oxides of nitrogen exhaust nondiurnal emission factor for gasoline
vehicles (grams/mile)
icoex is the I/M carbon monoxide exhaust nondiurnal emission factor for gasoline
vehicles (grams/mile)
ihcev is the I/M hydrocarbon evaporative nondiurnal emission factor for gasoline
vehicles (grams/mile)
idhcex is the I/M hydrocarbon exhaust nondiurnal emission factor for diesel vehicles
(grams/mile)
idnoxx is the I/M oxides of nitrogen exhaust nondiurnal emission factor for diesel
vehicles (grams/mile)
idcoex is the I/M carbon monoxide exhaust nondiurnal emission factor for diesel
vehicles (grams/mile)

If the user has provided both I/M and non-I/M VMT data, the mobile source emission processor
computes the fleet composite emission factors by vehicle category for temperature and speed
combinations through the application of Equations 4-51 through 4-58 This set of equations can
be applied only to on-network VMT data because the I/M percentages are input to the mobile
source emission processor on a link-specific basis.

4-60
EPA/600/R-99/030

hcdifs,t = Gj ((ihcdij,s,t * imvmtl) + (hcdij,s,t * (1 - imvmtl))) * vmixj (4-51)

hcexfs,t = Gj ((ihcexj,s,t * imvmtl) + (hcexj,s,t * (1 - imvmtl))) * vmixj (4-52)
noxfs,t = Gj ((inoxxj,s,t * imvmtl) + (noxxj,s,t * (1 - imvmtl))) * vmixj (4-53)
coexfs,t = Gj ((icoexj,s,t * imvmtl) + (coexj,s,t * (1 - imvmtl))) * vmixj (4-54)
hcevfs,t = Gj ((ihcevj,s,t * imvmtl) + (hcevj,s,t * (1 - imvmtl))) * vmixj (4-55)
dhcexfs,t = Gj ((idhcexj,s,t * imvmtl) + (dhcexj,s,t * (1 - imvmtl))) * vmixj (4-56)
dnoxfs,t = Gj ((idnoxxj,s,t * imvmtl) + (dnoxxj,s,t * (1 - imvmtl))) * vmixj (4-57)
dcoexfs,t = Gj ((idcoexj,s,t * imvmtl) + (dcoexj,s,t * (1 - imvmtl))) * vmixj (4-
58)

where: imvmt is the fraction of vehicles that are under an I/M program; and
l is the link identifier index.

Many indices have been left out of Equations 4-35 through 4-58. The composite emission
factors that are generated are gridded, hourly values. Because temperature is a gridded, hourly
value, the temperature index t in Equations 4-31 through 4-54 implies that the composite
emissions factor are gridded, hourly values.

Once the fleet composite emission factors have been computed, the mobile source emission
processor computes the motor vehicle emission estimates through the application of Equations 4-
59 through 4-66.

dhc_vee = hcdif * divmt * 10**-3 (4-59)

hc_xee = hcexf * ovmt * 10**-3 (4-60)
noxxee = noxf * ovmt * 10**-3 (4-61)
co_xee = coexf * ovmt * 10**-3 (4-62)
hc_vee = hcevf * ovmt * 10**-3 (4-63)
dhc_xee = dhcexf * ovmt * 10**-3 (4-64)
dnoxxee = dnoxf * ovmt * 10**-3 (4-65)
dco_xee = dcof * ovmt * 10**-3 (4-66)

where: dhc_vee is the gridded, hourly diurnal hydrocarbon emission estimate (kilograms)
divmt is the gridded, hourly diurnal VMT (miles)
hc_xee is the gridded, hourly exhaust hydrocarbon nondiurnal emission estimate for
gasoline vehicles (kilograms)
ovmt is the gridded, hourly nondiurnal VMT (miles)
noxxee is the gridded, hourly exhaust oxides of nitrogen nondiurnal emission
estimate for gasoline vehicles (kilograms)
co_xee is the gridded, hourly exhaust carbon monoxide nondiurnal emission estimate
for gasoline vehicles (kilograms)
hc_vee is the gridded, hourly evaporative hydrocarbon nondiurnal emission estimate
for gasoline vehicles (kilograms)

4-61
 EPA/600/R-99/030

dhc_xee is the gridded, hourly exhaust hydrocarbon nondiurnal emission estimate for
diesel vehicles (kilograms)
dnoxxee is the gridded, hourly exhaust oxides of nitrogen nondiurnal emission
estimate for diesel vehicles (kilograms)
dco_xee is (the gridded, hourly exhaust carbon monoxide nondiurnal emission
estimate for diesel vehicles (kilograms)

The mobile-source emission processor computes the final motor vehicle emission estimates file
by summing the emission estimates computed by Equations 4-59 through 4-66 over each state
identifier, county identifier, east-west grid cell identifier, north-south grid cell identifier, process
type (EV for evaporative, EX for exhaust), technology type (1 for gasoline, 2 for diesel), and
pollutant identifier (HC for hydrocarbon, CO for carbon monoxide, NOX for oxides of nitrogen).

The PART5 Model

The current version of PART5 was released in 1995 by the U.S. EPA Office of Mobile Sources.
The description given here is taken principally from the user’s guide, A Draft User’s Guide to
PART5: A Program for Calculating Particle Emissions from Motor Vehicles (U.S. EPA, 1995b),
which provides information in addition to that given in this section. The guide is available from
the Office of Mobile Sources Internet web site at http:www.epa.gov/omsw/models.htm The
PART5 is a FORTRAN program used to model emission factors needed to estimate emissions
from gasoline and diesel powered on-road vehicles. It calculates particle emission factors in
grams per mile for particle sizes 1 to 10 micrometers. The emission factors include exhaust
particulate matter and components, brake wear, tire wear, and re-entrained road dust. The
program includes default data for most inputs, but allows user-supplied data for most items. The
interactive aspects of running PART5, as it is provided by the Office of Mobile Sources, have
been subsumed into MEPPS to allow the user to enter and edit data, specify program control
flags, and run the model via a series of windows within the EMPRO mobile source processor.

Controls specified within PART5 may vary spatially and temporally. PART5 output report flags
are not germane within MEPPS because the outputs are automatically processed with emissions
from other emission sources in MEPPS internal format. They flags used include the following
items:

C Vehicle fleet mix. The user may specify of whether default or user-supplied VMT
vehicle fleet mixes are used. The VMT mix is the fraction of total VMT for all vehicles
contributed to by each vehicle class. The default VMT is based on national averages and
trends over the years. The trends reflect the sales shift from automobiles to light-duty
trucks, and the increasing use of diesel engines in both light and heavy-duty trucks. VMT
mix may vary spatially. The following are the descriptions of the vehicle classes and
corresponding Federal Highway Administration (FHA) and gross vehicle weight (GVW)
used in PART5.

4-62
EPA/600/R-99/030

Vehicle Class FHA Class GVW (lbs)

1 = LDGV (light-duty gasoline vehicle)
2 = LDGT1 (light-duty gasoline truck, I) 1 <6000
3 = LDGT2 (light-duty gasoline truck II) 2A 6001-8500
4 = HDGV (heavy-duty gasoline truck) 2B-8B >8500
5 = MC (motor cycle)
6 = LDDV (light-duty diesel vehicle) 1 <6000
7 = LDDT (light-duty diesel truck) 2A 6001-8500
8 = 2BHDDV (class 2B heavy duty diesel vehicle) 2B 8501-10000
9 = LHDDV (light heavy-duty diesel vehicle) 3,4,5 10001-
19500
10 = MHDDV (medium heavy-duty diesel vehicle) 6,7,8A 19501-
33000
11 = HHDDV (heavy heavy-duty diesel vehicle) 8B 33000+
12 = BUSES (buses)

C Mileage accumulation rates. The user may supply or use default mileage accumulation
rates and vehicle registration distributions. The mileage accumulation rate is the
expected number of miles that a typical vehicle (for each class) is expected to travel in a
year, divided by 100,000. The rates must be provided for each vehicle type for each of 25
years (12 years for motor cycles).

C Inspection and maintenance. The user specifies whether an inspection and maintenance
program is assumed for gasoline vehicle vehicles only.

C Reformulated gasoline. The specifies whether the use and effects of reformulated
gasoline on particulate emissions are assumed. The effects are partially based on the
sulfur content of gasoline used. An average sulfur weight percent of 0.034 is used for any
gasoline used in all years prior to 2000. If reformulated gasoline is used for the year 2000
or later, the sulfur weight percent is assumed to be 0.0138.

The purpose of MEPPS is to estimate temporally and spatially-varying emission data on a

gridded basis. This is consistent with some additional input data which PART5 requires be
entered on spatial basis, which are defined as PART5 scenarios. In EMPRO mobile source
processing the user is prompted to edit existing or new scenario data. The information includes:

C Region, calendar year, speed cycle (type of driving, ie. cruise), and average speed.

C Percent of unpaved road silt, paved road silt loading in gm /m2, and optionally, the
average number of wheels per vehicle. Modeled dust emission factor are highly sensitive
to the unpaved road silt percentage which is extremely variable. Consequently, measured
data are advised when possible.

4-63
 EPA/600/R-99/030

C The number of days each year with greater than 0.01 inches of precipitation for use in
modeling re-entrained dust. This is climatological average data. In the future, the
emission processing system may be modified to use more spatially and temporally
accurate modeled precipitation data by grid cell taken from MM5 and MCIP for the
PART5 calculations.

C A scenario name may be applied. In MEPPS a FIPS code a geographic identifier is

entered which is used to apply PART5 emission factors computed for the county level to
the appropriate modeling grid cells.

C PART5 expects the user to define a maximum particle size cutoff (maximum allowed is
10 micrometers). In MEPPS the maximum size is defaulted to 10 micrometers to ensure
that the full range of particle sizes are represented in the regional emission estimates.

C Average vehicle weight (lbs)

PART5 Emission Factors

The emission factors calculated by PART5 include the particulate pollutant compounds of lead,
sulfate, soluble organic fraction particulate matter, remaining carbon portion particulate matter,
and total exhausted particulate. The lead and sulfate are formed from the lead and sulfur
contained in the fuel. The soluble organic fraction consists primarily of hydrocarbons coming
from unburned or partially burned fuel and lubricating oil. The remaining carbon portion
consists of soot-like carbon (elemental carbon) and trace amounts of other components from the
fuel and lubricating oil. The total exhaust particulate is the sum of these four categories. In
addition to these categories of exhaust emissions, idle exhaust emissions (for heavy diesel
vehicles only), brake wear, tire wear, fugitive dust, indirect sulfate, and gaseous sulfur dioxide
are calculated by PART5.

The model calculates the emission factors for the 12 vehicle classes previously described and a
fleet-wide average (estimated by VMT weighting of the emission factors for each of the 12
vehicle classes. The factors are composites of the emission factors for the 25+ years prior to the
year of interest, in order to allow for the effects of older vehicles. The composite emission
factor for each vehicle class is calculated by weighting the emission factor calculated for each
model year by the travel fraction for that model year, and then summing the 25 weighted factors.
The travel fraction of a model year is the fraction of VMT by a vehicle of that model year out of
the total number of miles traveled by all model years’ vehicles in that vehicle class.

25
EFCOMPv = EF m,v * TF m,v (4-67)
m=1

where: EFCOMPv is the composite emission factor for vehicle class v,

4-64
EPA/600/R-99/030

EF m,v is the emission factor for vehicle class v, model year m,

TF m,v is the travel fraction for vehicle class v, model year m.

The overall travel fraction of a vehicle class represents the fraction of the total number of VMT
of that class of the total highway VMT by all 12 classes. The VMT fractions for each vehicle
class are multiplied by the corresponding composite emission factors (EFCOMPv), and the sum
of the adjusted emission factors is reported as the emission factor for all vehicles.

EFALL = ' EFCOMPv * TFCLASSv (4-68)

where: EFALL is the weighted emission factor for all vehicles,

TFCLASSv is the VMT of vehicle class v.

The emission factor for all vehicles, EFALL, represents the grams/mile of emissions

Lead Emission Factors for gasoline-fueled vehicles

Lead particulate emission factors for gasoline-fueled vehicles assume that almost all lead in the
fuel is exhausted. Therefore, the emission factors (grams/mile) depend on the lead content of
fuel and fuel economy of the vehicle (miles/gallon). The factors allow for the fact that the lead
content of leaded fuel is much greater than for unleaded fuel. The assumption is made that
because diesel fuel has negligible lead content, the lead emissions from diesel vehicles will be
negligible. The following formulae are used to compute lead emission factors for all gasoline-
fueled automobiles and motorcycles, respectively.

LEADm,v = PLNOCTm,v * VLNOCTm.v + PUNOCTm,v * VUNOCTm,v +

PLYSCTm,v * VLYSCTm,v + PUYSCTm,v, * VUYSCTm,v (4-69)

where: m denotes a specified model year

v denotes a specified vehicle class
LEADm,v is the lead particulate emissions for any given vehicle (grams/mile)
PLNOCTm,v is the emissions for a non-catalyst, leaded fuel vehicle (grams/mile)
PUNOCTm,v is the emissions for a non-catalyst, unleaded fuel vehicle (grams/mile)
PLYSCTm,v is the emissions for a catalyst, leaded fuel vehicle (grams/mile)
PUYSCTm,v is the emissions for a catalyst, unleaded fuel vehicle (grams/mile)
VLNOCTm,v is the emissions for a non-catalyst, leaded fuel vehicle (grams/mile)
VUNOCTm,v is the emissions for a non-catalyst, unleaded fuel vehicle (grams/mile)
VLYSCTm,v is the fraction of catalyst leaded fuel vehicles
VUYSCTm,v is the fraction of catalyst, unleaded fuel vehicles

The emission rate is adjusted for speed by the factor FEC such that:

CLEADm,v = FEC * LEADm,v (4-70)

4-65
 EPA/600/R-99/030

where: CLEADm,v is the lead emissions for a vehicle of model year m and vehicle class v,
which has been adjusted for the effect of speed (grams/mile)

FEC is 1/SCFc (4-71)

SCFc is the speed correction factor, based either transient driving cycle (c=1) or
steady cruise driving cycle (c=2)

SCF1 = 0.17930 + (0.038561 * SPEEDv ) - (0.00041067 * SPEEDv2) (4-72)

SCF2 = 0.26929 + (0.054607 * SPEEDv) - (0.00069818 * SPEEDv2) (4-73)

SPEEDv is the average speed for vehicle class v (mph) - user input

Motorcycles do not have catalytic emission controls, therefore organic emission factors are not
calculated and sulfate emission factors are deemed negligible. PART5 emission factors for
motorcycles are almost entirely for lead particulate matter. The lead particulate emission factor
for 2-stroke engines is 0.33 g/mi and for 4-stoke engines 0.046 grams/mile. For model years
before 1978 there were 51 percent 4-stroke engines and 49 percent 2-stroke engines, computed as
follows:

LEADm, motorcycle = [(0.49 * 0.33) + (0.51 * 0.046)] * PSL (4-74)

For model years 1978 and later motorcycles are assumed to consist entirely of 4-stroke engines:

LEADm,motorcycle = 0.046 * PSL (4-75)

where: PSL is the fraction of all particles that are emitted based on a specified upper particle
size cutoff (10 micrometers running in MEPPS)

The specific derivations of the lead particulate emission formulae are given in Appendix 1 of the
PART 5 user’s guide (U.S. EPA, 1995b).

Sulfate Emission Factors for Gasoline-Fueled Vehicles

Particulate sulfate emission factors consist of direct and indirect sulfate material. The direct
sulfate is exhausted as sulfuric acid, and the indirect sulfate is formed later in the atmosphere
from exhausted SO2. The indirect sulfate in the model is calculated based on the assumption that
it consists entirely of ammonium sulfate and ammonium bisulfate. The direct sulfate, indirect
sulfate, and gaseous sulfate emission factors are computed in PART5, and the emission factors
reported as grams/mile traveled.

4-66
EPA/600/R-99/030

The direct sulfate from non-catalyst vehicles using leaded fuel (includes catalyst-equipped
vehicles which are misfueled, making the catalyst ineffective) is calculated as:

DSULFN = .002, for speeds equal to or less than 19.6 mph (4-76)

DSULFN = .001, for speeds equal to or greater than 34.8 mph (4-77)

The direct sulfate from catalyst vehicles is calculated as :

DSULFCm,v = [FRACcat/no air (.005) + FRACcat/air (.016)] (4-78)

for speeds equal to or less than 19.6 mph

DSULFCm,v = [FRACox/noair (.005) + FRAC3w/noair (.001) + FRACox/air (.020)

+FRAC3w/air (.025)] (4-79)
for speeds equal to or greater than 34.8 mph

For speeds between 19.6 and 34.8 mph, DSULFN and DSULFC are interpolated between
Equations 4-76 and 4-77 and 4-78 and 4-79, respectively.

where: m denotes a specified model year

v denotes a specified vehicle class
FRACcat/noair is the fraction of vehicles which are catalyst equipped with no air pump
FRACcat/air is the fraction of vehicles which are catalyst equipped with an air pump
FRACox/noair is the fraction of vehicles which are oxygen catalyst equipped with no air
pump
FRAC3w/noair is the fraction of vehicles which are 3-way catalyst equipped with no air
pump
FRACox/air is the fraction of vehicles which are oxidation catalyst equipped with an
air pump
FRAC3w/air is the fraction of vehicles which are 3-way catalyst equipped with air
pump

The direct sulfate from all gasoline-fueled vehicles is computed as:

DSULFm,v = CTLFRCm,v * DSULFCm,v + (1.0 -CTLFRCm,v) * DSULFN (4-80)

where: CTLFRCm,v = CATFCTm,v (1-RMISm,v) (4-81)

the fraction of the vehicle class that has an effective catalyst

The PART5 assumes that all sulfur in fuel is exhausted as either sulfate or gaseous sulfur dioxide
(SO2). Therefore when the direct sulfate emission factor is calculated, the remaining sulfur in the
fuel is considered to be exhausted as SO2. The amount of sulfur remaining in the fuel after the

4-67
 EPA/600/R-99/030

direct sulfate emission factor has been determined must be calculated to find the amount of sulfur
exhausted as SO2 (grams/mile).

The following equation is used to determine the fraction of sulfur in the fuel that has been
directly converted to sulfate (DSULFm,v calculated in Equation 4-80 above). The equation
calculates direct sulfate as a function of the fuel sulfur content, DCNVRT (the fraction of sulfur
in the fuel that is converted to direct sulfate), and the fuel economy.

DSULFm,v = 13.6078 * (1.0 + WATER) * FDNSTY * SWGHT * DCNVRT

/ FEm,v (4-82)

where: m denotes a given model year

v denotes a given vehicle class
DSULFm,v is the direct sulfate emission factor (grams/mile)
WATER is the weight ratio of seven water molecules to sulfate, 1.2857, based the
estimate that at 50 percent humidity, seven water molecules bond with each
sulfuric acid molecule
FDNSTY is the fuel density in lb/gal (6.09 pounds/gallon)
FEm,v is the fuel economy
SWGHT is the weight percent of sulfur content in fuel (.034, except for reformulated
fuel phase II, for year 2000 and later .0138)
DNVRT is the percent of sulfur in the fuel that is directly converted into sulfate (2
percent)
13.6078 is a unit conversion factor equal to (453.592 * 3.0)/100, where 453. 592 is
equal to the number of gram in a pound, 3.0 is the weight ratio of sulfate to
sulfur, and division by 100 corrects for the weight percent of sulfur, SWGHT

If DSULFCm,v and DSULFN (from Equations 4-76 through 4-79) are substituted in Equation 4-
82, one can solve for the fractions of sulfur in the fuel that are converted to sulfate separately for
catalyst and noncatalyst vehicles:

FCNVRCm,.v = DSULFCm,v * FEm,v / (13.6078 * (1.0 + WATER) *

FDNSTY* SWGHT) (4-83)

FCNVRNm,v = DSULFN * FEm,v / (13.6078 * (1.0 + WATER) *

FDNSTY * SWGHT) (4-84)

where: FCNVRCm,v is the fraction of the percent of fuel that is directly converted into sulfate
for catalyst equipped vehicles
FCNVRNm,v is the fraction of the percent of fuel that is directly converted into
sulfate for non-catalyst vehicles

4-68
EPA/600/R-99/030

The gaseous sulfur emission factors, which are dependent on the above fractions, are calculated
from the following equation:

SO2m,v = 9.072 * FDNSTY * SWGHT * (1.0 -DCNVRT) / FEm,v (4-85)

where: SO2m,v is the gaseous sulfur emission factor of a vehicle

9.072 is a unit conversion factor equal to (453.592 *2)/100, where 453.592 is the
number of grams in a pound, 2 is the weight ratio of SO2 to sulfur, and the
division by 100 corrects for the weight percent of sulfur, SWGHT

Additional details concerning the calculation of gaseous sulfur emission factors are given in the
PART5 user’s guide (U.S. EPA, 1995b). In addition to direct sulfate and gaseous sulfate
emission factors, PART5 estimates an indirect sulfate emission factor by assuming that a fraction
of the gaseous sulfur dioxide emissions is later converted in the atmosphere to sulfate material.
Based on ambient sulfur and sulfate measurements in 11 cities, it is estimated that 12 percent of
all gaseous sulfur is converted to sulfate. Additional information on the calculation of indirect
sulfate is given in the PART5 user’s guide.

Sulfate Emission Factors for Diesel-Fueled Vehicles

For diesel-fueled vehicles, PART5 calculates sulfate emission factors again assuming that all
sulfur in the fuel is exhausted as either sulfuric acid or gaseous sulfur dioxide. The direct sulfate
emission factor (grams/mile) is calculated using the following equation:

DSULFm,v = 13.6078 * (1.0 + WATER) * FDNSTY * SWGHTD *

DCNVRT / FEm,v (4-86)

where: m is a specified model year

y is a specified vehicle class
DSULFm,v is the direct sulfate emission factor for a vehicle (grams/mile)
DCNVRT is the fraction of sulfur in the fuel that is converted directly to sulfate (2.0
percent

FDNSTY is the density of diesel fuel (7.11 pounds/gallon)

FEm,v is the fuel economy for a vehicle (miles/gallon)
SWGHTD is the weight percent of sulfur in diesel fuel (0.25 for high sulfur fuel,
0.05 for low sulfur fuel used in 1993 and later)
WATER is the weight ratio of seven water molecules to sulfate (1.2857)
13.6078 is a unit conversion factor equal to 493.592 * 3/100, where 493.592 is the
number of grams in a pound, 3 is the weight ratio of sulfate to sulfur, and the
division by 100 corrects for the weight percent of sulfur SWGHTD

The gaseous sulfur emission factor is calculated by the following equation:

4-69
 EPA/600/R-99/030

SO2m,v = 9.072 * FDNSTY * SWGHTD * (1.0 - DCNVRT) / FEm,v (Eq.4-

87)

where:
SO2m,v is the sulfur emission factor (grams/mile) of a vehicle
9.072 is a unit conversion factor equal to (453.592 * 2/100) where 453.592 is the
number of grams in a pound, 2 is the weight ratio of SO2 to sulfur, and the
division by 100 corrects for the weight percent of sulfur, SWGHTD

The indirect sulfate emission factor, consisting mainly of ammonium sulfate and ammonium
bisulfate is calculated using the following equation:

ISULFm,v = ICNVRT * SO2m,v * (3/2) * AMNWGT (4-88)

where:
ISULFm,v is the indirect sulfate emission factor of a vehicle (grams/mile)
ICNVRT is the fraction of SO2 converted to sulfate in the atmosphere (12 percent)
3/2 is the weight ratio of SO2 to sulfate
AMNWGT is the estimated weight ratio of the combination of ammonium bisulfate
and ammonium sulfate in the atmosphere to sulfate (1.6)

Total Exhaust Particulate Emission Factors for Gasoline-Fueled Vehicles

The total exhaust particulate emission factors for light-duty gasoline fueled vehicles are
calculated by summing lead, direct sulfate, and a carbon emission factor which includes soluble
organic material and other remaining carbon. Table 4-5 presents a summary of the carbon
emission factors by vehicle model year and type of technology.

Table 4-5. Carbon Emission Factors for Gasoline-Fueled Vehicles (grams/mile)

Vehicle Type/ Leaded Fuel Unleaded/ Catalyst Unleaded/ Catalyst Unleaded/ Non-
Model Year Equipped (No air Equipped (with air catalyst equipped
pump) pump)

Light-Duty
Gasoline Vehicles

pre-1970 .193 NA NA .030

1970-1974 .068 .0060 .0250 .030

1975-1980 .030 .0060 .0250 .030

1981 + .017 .0043 .0043 .017

4-70
EPA/600/R-99/030

Table 4-5. Carbon Emission Factors for Gasoline-Fueled Vehicles (grams/mile)

Vehicle Type/ Leaded Fuel Unleaded/ Catalyst Unleaded/ Catalyst Unleaded/ Non-
Model Year Equipped (No air Equipped (with air catalyst equipped
pump) pump)

Light-Duty
Gasoline Trucks I

pre-1970 .193 NA NA .030

1970-1974 .068 .0060 .0250 .030

1975-1986 .030 .0060 .0250 .030

1987 + .017 .0043 .0043 .017

Light-Duty
Gasoline Trucks II

pre-1979 .370 NA NA .054

1979-1986 .068 .0060 .0250 .030

1987 + .030 .0043 .0043 .017

Heavy-Duty
Gasoline Vehicles

pre-1987 .370 .054 .054 .054

1987 + .163 .054 .054 .054

Total Exhaust Particulate Emission Factors for Diesel-Fueled Vehicles

The total exhaust particulate emission factors are for diesel-fueled vehicle categories and model
years (EFDPMm,v). The emission factors for heavy-duty vehicles are in units of grams/brake-
horsepower hour (g/BHP-hr), which are converted to grams/mile by PART5. The conversion
factors and emission factors vary by model year. The emission factors for light-duty diesel
vehicles are in units of grams/mile. The total exhaust emission factors given in Table 4-6 are
based on high-sulfur fuel. The sulfur content in diesel fuel was reduced in 1993 by EPA
regulatory requirements. Consequently, when a the specified calendar year is 1993 or later,
PART5 will adjust the exhaust emission factors for lower sulfur fuel. Particulate emission
factors for diesel-fueled vehicles are not adjusted for speed.

4-71
 EPA/600/R-99/030

Table 4-6. Exhaust Particulate Emission Factors for Diesel-Fueled Vehicles

Vehicle Type/ Model Year Group Exhaust Particulate Emission Factor

Light-Duty Diesel Vehicles (grams/mile)

pre-1981 .700

1981 .259

1982-1984 .256

1985-1986 .255

1987 .134

1988-1990 .132

1991-1993 .131

1994-1995 .128

1996 + .100

Light-Duty Diesel Trucks (grams/mile)

pre-1981 .700

1981 .309

1982-1984 .354

1985-1986 .358

1987 .334

1988-1990 .291

1991-1993 .294

1994-1996 .130

1997 + .109

Class 2B Heavy-Duty Diesel Vehicles (grams/brake

horsepower-hour)

pre-1988 .5156

1988-1990 .5140

1991-1993 .2873

1994 + .1011

Light Heavy-Duty Diesel Vehicles (grams/brake

horsepower-hour)

4-72
EPA/600/R-99/030

Table 4-6. Exhaust Particulate Emission Factors for Diesel-Fueled Vehicles

Vehicle Type/ Model Year Group Exhaust Particulate Emission Factor

pre-1988 .5156

1988-1990 .5140

1991-1993 .2873

1994 + .1011

Medium Heavy-Duty Diesel Vehicles (grams/brake

horsepower-hour)

pre-1987 .6946

1988-1990 .4790

1991-1993 .2747

1994 + .0948

Heavy Heavy-Duty Diesel Vehicles (grams/brake

horsepower-hour)

pre-1987 .6444

1988-1990 .4360

1991-1993 .2709

1994 + .0836

Buses (grams/brake horsepower-hour)

pre-1987 .6931

1988-1990 .4790

1991 .2772

1992 without traps .1716

1992 with traps .0257

1993 without traps .1457

1998 with traps .0240

1994 + .0591

Soluble Organic Fraction and Remaining Carbon Portion Emission Factors for Diesel-Fueled
Vehicles

4-73
 EPA/600/R-99/030

The PART5 model calculates the Soluble Organic Fraction (SOF) emission factor as a fraction of
the remaining mass, using the following equation:

SOFm,v, = [EFm,v- DSULFm,v ] * (fractionSOF,v) (4-89)

The Remaining Carbon Portion (RCP) is defined the remainder (everything else):

RCPm,v = Efm,v - DSULFm,v - SOFm,v (4-90)

where: m is the model year of a selected vehicle

v is the class of a selected vehicle
SOFm,v is the Soluble Organic Fraction of the exhaust particulate emission factor
(grams/mile)
RCPm,v is the remaining carbon portion (elemental carbon) of the exhaust particulate
emission factor (grams/mile)
EFm,v is EFDPMm,v * CFm,v , the exhaust particulate emission factor for a vehicle
(grams/mile)
EFDPMm,v is the exhaust particulate emission factor for a vehicle (grams/Brake
Horsepower-hour)
CFm,v is the conversion factor from grams/Brake Horsepower-hour to grams/mile
(Brake Horsepower-hour/mile)
fractionSOF,v is the fraction of the non-sulfate portion (ie. the carbon portion) of the
diesel exhaust particulate emission factor which is organic carbon for a
vehicle (Brake Horsepower-hours/mile)

The Soluble Organic Fractions (fractionSOF,v) for different vehicle classes are as follows (U.S.
EPA, 1990):

C 0.18 for LDDVs (Light-Duty Diesel Vehicles)

C 0.50 for LDDT1s (Class 1 Light-Duty Diesel Trucks)
C 0.48 for LDDT2s (Class 2 Light-Duty Diesel Trucks)
C 0.51 for LHDDVs (Light Heavy-Duty Diesel Vehicles) and 2BHDDVs (Class 2B Heavy-
Duty Diesel Vehicles)
C 0.44 for MHDDVs (Medium Heavy-Duty Diesel Vehicles) and BUSES
C 0.24 for HHDDVs (Heavy Heavy-Duty Diesel Vehicles)

Idle Emission Factors for Heavy Diesel-Fueled Vehicles

Idle emission factor data in grams/hour were collected from manufacturers for heavy-duty diesel
vehicle classes only. Consequently, the idle emission factors are not included into the “all
vehicles” emission category. The vehicle class emission factors are calculated by averaging
together model-year-specific emission data, where the model-year-specific emission data are
weighted by the estimated travel fraction of that model year within the vehicle class.

4-74
EPA/600/R-99/030

The idle emission factors are model-year-specific but the model year emission rates do not vary
by vehicle class. Consequently, the same model year emission factors are used for all the heavy-
duty diesel classes, and the differences between idle emission factors between classes reflects
only the differing travel fractions between model years for a class. As a result, the emission
factors reported for the smaller of the heavy-duty vehicle classes, such as 2BHDDV and
LHDDV, may be over estimated. The base idle emission factors in PART5 for all heavy-duty
diesel vehicles are as follows:

C 5.370 grams/hour for models prior to 1988

C 3.174 grams/hour for model years 1988-1990
C 1.860 grams/hour for model years 1991-1993
C 1.004 grams/hour for model year 1994 +

Re-entrained Dust from Unpaved Roads

Re-entrained road dust emission factors for PM10 (particulate matter less than 10 micrometers in
size) in PART5 are estimated using an equation (4-91) based on ambient measurements.
Because these measurements include particulate matter from brake wear, tailpipe exhaust, tire
wear, and ambient background particulate matter concentrations, these factors must be subtracted
from the unpaved road particulate emission factors before the latter factors can be applied. It is
necessary to obtain a data base of unpaved and paved road types by county for regional modeling
purposes. The MEPPS contains procedures and estimates for 1995 developed for EPA by
calculating re-entrained emissions by the month at the state road-type level for the average
vehicle fleet, and then allocated to the county road type by population for unpaved roads and by
total VMT for paved roads. These procedures should be consulted for more details, such as the
means of estimating unpaved and paved VMT data. The equation used to calculate PM10 from
unpaved roads is as follows:

UNPVD = PSDUNP * 5.9 * (UNSILT/12.0) * (SPD/30.0) * (WEIGHT/3.0)0.7 *

(VWHEEL/4.0)0.5 * (365-IPDAYS)/365 * 453.592 (4-91)

where: UNPVD is the fleet average unpaved road dust emission factor (grams/mile)
PSDUNP is the fraction of particles less than or equal to the particle size cutoff (the
cutoff is 10 micrometers in MEPPS)
UNSILT is the percent silt content of the surface material (user input)
SPD is the average vehicle speed in miles/hour (user input)
WEIGHT is the fleet average vehicle weight (user input in pounds)
VWHEEL is the fleet average number of wheels (user input, default is 4)
IPDAYS is the average number of precipitation days per year with greater than 0.01
inches of rain (user input - MEPPS contains climatological default data for
1995)
453.592 is the number of grams in a pound

4-75
 EPA/600/R-99/030

Emission factors for brake and tire wear (in addition to exhaust emissions) must be calculated so
that they may be subtracted from the unpaved road emission factors. The brake wear emission
factor is assumed to be the same for all vehicle classes. It is set equal to:

BRAKE = 0.0128 * PSBRK (4-92)

where: PSBRK is the fraction of particles less than or equal to the particle size cutoff. The
emission factor 0.0128 grams/mile is taken from U.S. EPA, 1985c.

The tire wear emission factor is calculated using the following equation:

EFTIREv = 0.002 * PSTIRE * IVEHWLv (4-93)

where: v is a selected vehicle class

EFTIREv is the tire wear emission factor
0.002 is the emission rate of airborne particulate matter from tire wear for light-duty
vehicles (U.S. EPA, 1985)
PSTIRE is the fraction of particles less than or equal to the particle size cutoff
IVEHWLv is the average number of wheels on a vehicle of a given class, where
LDGV=4, LDGT1,2=4, HDGV=6, MC=2, LDDV=4, LDDT=4, 2BHDDV=4,
LHDDV=6, MHDDV=6,HHDDV=18, BUSES=4

Re-entrained Dust from Paved Roads

The PM10 emission factor for paved roads is estimated similarly to that for unpaved roads. The
VMT data for paved roads may be developed following the procedures described in U. S. EPA,
1998. The equation used to estimated paved road re-entrained dust emission factors again is
based on ambient measurements, and tailpipe, brake wear, and tire wear emission factors must be
subtracted prior to use of the paved road dust emission factor. The paved road emission factors
are calculated by:

PAVED = PSDPVD * (PVSILT/2.0)0.65 * (WEIGHT/3.0)1.5 (4-94)

where: PAVED is the fleet average paved road dust emission factor (grams/mile)
PSDPVD is the base emission factor for the particle size cutoff (10 micrometers in
MEPPS)
PVSILT is the road surface silt loading (grams/square meter) (user input)
WEIGHT is the fleet average vehicle weight (input by the user in pounds)

Application in MEPPS

The PART5 model is a companion of Mobile5a in MEPPS. When running MEPPS interactively
through the Tools Manager, the user is prompted to specify input data and options under

4-76
EPA/600/R-99/030

EMPRO, Mobile Source Model. Standard lookup data files can be found in the Models-3
directory structure at /home/models3/datasets/nostudies/part5/. Default/example input data sets
are included for the 1995 calendar year. When a study is created, the study input data are located
at $EMS_HOME/project/$EMS_PROJECT/raw_data/$EMS_DOMAIN/common/.
If the user is processing mobile emission data using the Study Planner, it is necessary to ensure
that the appropriate PART5 input data sets are in place first. This may be accomplished when
using MEPPS interactively to establish MEPPS directory structure prior to running through
Study Planner. If the data are in place, Study Planner will automatically process them.

When processing mobile particulate emission data, the input and output data of PART5 are
assigned to and summed within each grid cell in the same fashion given in detail under the
Gridding VMT discussion of the implementation of the Mobile 5a model (above).

4.2.5 Chemical Speciation of Emission Data

Chemical transport models, such as CMAQ, require that emission data be provided for either
individual species or specific species groups or “lumped” species. This is necessary so that the
atmospheric chemistry of pollutants may be more accurately modeled. However, an initial
processing step is required because emission data are often reported for pollutants that are
aggregates of many species, such as VOC. These aggregate pollutants must be split into their
component species, or “speciated”.

The speciation takes two forms, discrete and lumped-model.

In discrete speciation, a pollutant is split into the individual components which comprise the
pollutant. For the organic pollutant TOG (total organic gas), the individual components which
comprise the pollutant are dependent on a variety of factors including the process, fuel type, and
device from which the emissions occurred. For example, TOG from the exhaust of automobile
may contain approximately fifty discrete organic compounds (benzene, methane, toluene,
hexane, etc.), while TOG from degreasers may contain approximately seventy discrete organic
compounds. NOX (nitrogen oxides), which is related to combustion processes, is speciated
(discrete) into NO and NO2 (and sometimes HONO).

The discrete components in an emission stream are determined by a number of methods

including source testing, surrogate application, and engineering knowledge of the process. Many
sources have been inventoried and a compendium of species profiles has been assembled by the
US EPA in the Air Emissions Species Manual (US EPA, 1988). The Air Emissions Species
Manual contains a list of TOG and particulate matter (PM) species profiles to which a substantial
number of emission sources have been assigned. A compendium of species profiles currently
available is compiled in the US EPA Speciate database. The database is available on the Air
Chief CD ROM, which is updated annually (US EPA, 1997). Each of the species profiles
identifies the mass percent of the discrete compounds that comprise TOG and PM. Note that
NOX, SOX (sulfur oxides), and CO (carbon monoxide) do not have source-specific speciation

4-77
 EPA/600/R-99/030

profiles. For all sources, NOX is discretely speciated into NO and NO2 (and sometimes HONO),
and CO is treated explicitly. The speciation processor currently assumes that NOx is speciated to
95 percent NO and 5 percent NO2, based on average observed values.

The MEPPS EMPRO speciation processor provides chemical speciation of hydrocarbons,

oxides of nitrogen, and sulfur oxides. For example, aggregations of hydrocarbon species such as
TOG are disaggregated to their component individual chemical species. The processor speciates
the spatially and temporally allocated emission estimates that are prepared by the emission
estimation processors and models (e.g., point source processor, mobile emission model). The
speciation is accomplished using look-up tables of profiles containing source-category specific
and specie-specific chemical split factors (percent of mass of source emissions attributable to
each specie for a given source category). The general components of emission speciation
processing are illustrated in Figure 4-6.

It is computationally prohibitive to model the chemistry of all discrete VOC compounds in the
emissions stream in photochemical grid models. Therefore, individual organic species
comprising Total Organic Gas are assigned to one or more model species (groups of species)
according to the chemical mechanism that is being used. Thus, instead of modeling with a larger
number of discrete compounds, the discrete compounds are lumped into a much smaller number
of mechanism species. The rules for assigning the discrete compounds to the mechanism species
are mechanism dependent and typically involve lumping in one of two ways: 1) lumping
compounds with similar reactivity characteristics into a single mechanism species (the lumped
molecule approach) or 2) assigning molecular fragments of an individual compound to one or
more mechanism species on the basis of molecular structure (lumped structure approach). In
Models-3, the lumping of discrete compounds to form mechanism species is carried out using
tables of split factors that assign the discrete compounds to mechanism species. Currently, two
mechanisms, the Regional Acid Deposition Model 2.0 (RADM2) and Carbon Bond 4 (CB-4) are
available in Models-3. The user may also define lumping procedures for an alternative
mechanism if the assignment methods for that mechanism are known, or the user may create a
modified version of the lumping procedure for one of the two mechanisms in Models-3.

In the CB-4 mechanism, discrete compounds are assigned to mechanism species on the basis of
the compound’s carbon bond structure (Gery et al.,1989). For instance, single carbon-carbon
hydrocarbon bonds are assigned to a paraffin group (PAR), and carbon-carbon double bonds are
assigned to an olefinic group (OLE). Thus, an individual discrete VOC could be dis-aggregated
into more than one mechanism species depending on its structure. Descriptions of the CB-4
mechanism species are given in Table 4-8.

The RADM2 mechanism lumps discrete compounds on the basis of their prevalence in the
atmosphere, common reactivity, and/or molecular weight (Stockwell et al., 1990). For RADM2,
individual VOC compounds are first assigned to one of 32 lumped groups. The 32 lumped
groups are then further reduced to 15 groups for increased computational efficiency. The
relationship between the 32 and 15 lumped groups is shown in Table 4-9. The EMPRO does not

4-78
EPA/600/R-99/030

include chemical or size fractionation of particulate matter, although size fractionation of

particulate matter is planned in the future.

Figure 4-6 General Components of Emission Speciation Processing

4-79
 EPA/600/R-99/030

Table 4-8. Carbon Bond-4 Lumped Species

CB4 Lumped Species CB4 Lumped Species Lumped Species Description

Number Code

1 OLE Olefinic carbon bond (C=C)

2 PAR Paraffin carbon bond (C-C)

3 TOL Toluene (C6H5-CH3)

4 XYL Xylene (C6H4-(CH3)2)

5 FORM Formaldehyde (CH2=O)

6 ALD2 High MW aldehydes (RCHO, R>H)

7 ETH Ethene (CH2=CH2)

8 MEOH Methanol (CH3OH)

9 ETOH Ethanol (C2H5OH)

10 ISOP Isoprene

11 NR Non Reactives as methane

12 NO Nitric Oxide

13 NO2 Nitrogen Dioxide

14 CO Carbon Monoxide

15 SO2 Sulfur Dioxide

16 AERO Aerosols (PM)

4-80
EPA/600/R-99/030

Table 4-9. RADM 2.0 Lumped Species Descriptions

RADM 2.0 32 RADM 2.0 16 Allocation Reactivity Factor Description of the 32 RADM
Lumped Lumped Species Factor (ratio) Lumped Species Before Grouping to
Species (fraction of 16 Lumped Species
one)

1 CH4 1 1 Methane

2 ETH 1 1 Ethane

3 HC3 1 0.519 Propane

4 HC3 1 0.964 Alkanes (0.25-0.50)

5 HC5 1 0.956 Alkanes (0.50-1.00)

6 HC8 1 0.945 Alkanes (1.00-2.00)

7 HC8 1 1.141 Alkanes (>2.00)

8 HC8 0.91 1.101 Alkane/Aromatic Mix

8 XYL 0.09 1 Alkane/Aromatic Mix

9 OL2 1 1 Ethene

10 OLT 1 1 Propene

11 OLT 1 1 Alkenes (Primary)

12 OLI 1 1 Alkenes (Internal)

13 OLT 0.5 1 Alkenes (Prim/Int Mix)

13 OLI 0.5 1 Alkenes (Prim/Int Mix)

14 TOL 1 0.293 Benzene/Halobenzenes

15 TOL 1 1 Aromatics (<2 react)

16 XYL 1 1 Aromatics (>2 react)

17 CSL 1 1 Phenols and Cresols

18 OLT 0.5 1 Styrenes

18 TOL 0.5 1 Styrenes

19 HCHO 1 1 Formaldehyde

20 ALD 1 1 Higher Aldehydes

21 KET 1 0.253 Acetone

22 KET 1 1 Ketones

4-81
 EPA/600/R-99/030

Table 4-9. RADM 2.0 Lumped Species Descriptions

RADM 2.0 32 RADM 2.0 16 Allocation Reactivity Factor Description of the 32 RADM
Lumped Lumped Species Factor (ratio) Lumped Species Before Grouping to
Species (fraction of 16 Lumped Species
one)

23 ORA2 1 1 Organic Acids

24 HC3 1 0.343 Acetylene

25 HC3 1 0.078 Haloalkenes

26 . 1 1 Unreactive

27 HC3 1 0.404 Others (<0.25 react)

28 HC3 1 1.215 Others (0.25-0.5 react)

29 HC5 1 1.075 Others (0.5-1.0 react)

30 HC8 1 1.011 Others (>1.00 react)

31 . 1 1 Unidentified

32 . 1 1 Unassigned

In Models-3, the user initiates speciation by selecting the chemical mechanism to be used for a
study in the Study Manager under the Models-3 framework, and within the MEPPS main
window under Tools Manager. The procedure is explained in detail in Chapter 6 of Models-3
Volume 9B: User Manual. The speciation processor is written primarily in SAS®; however, it
includes some FORTRAN programs. ARC/INFO® is not used in the speciation processor.

Speciation Processing in MEPPS

In MEPPS, speciated gaseous emissions can be calculated in either of two units – moles/hour or
Kg/hour. (Particle emissions are always calculated as Kg/hr.) Emissions in moles are required by
the Models-3 CMAQ, but some analyses require that emissions be expressed in mass per unit
time. Emissions in mole and mass units are calculated using the following two generic
equations:

chemestij ' 'p 'm hrimp ( 1000 ( rogtotogmp ( factorjmp / divisor jmp (4-95)

where: chemestij is the gridded emissions estimate for hour i and mechanism species j
(moles/hour)

4-82
EPA/600/R-99/030

hrimp is the gridded hourly emission estimate of pollutant p (NOx, SOx, CO, PM,
NH3, or ROG) for hour i and source category m (Kg /hr). The factor of 1000
is used to convert kilograms to grams.
rogtotogmp is a ROG to TOG conversion factor (discussed below)
factorjmp is a mole-based split factor to allocate total emissions of pollutant p from
source category m to mechanism species j

divisorjmp is a second conversion factor to allocate total emissions of pollutant p from

source category m to mechanism species j

and
estKgij ' 'p 'm hrimp ( rogtotogmp ( xmassjmp (4-96)

where: estKgij is the gridded emissions estimate for hour i and mechanism species j
(Kg/hour)
hrimp is the gridded hourly emission estimate of pollutant p (NOx, Sox, CO, PM,
NH3, or ROG) for hour i and source category m (Kg /hr)
rogtotogmp is a ROG to TOG conversion factor (discussed below)
xmassjmp is a mass-based split factor to allocate total emissions of pollutant p from
source category m to mechanism species j

These equations show that emissions for mechanism species are computed by summing the
contributions of pollutants from the different source categories that emit that pollutant. As
indicated previously, the assignment of a pollutant to a specific mechanism species usually
mechanism dependent, although some assignments may be handled the same way in different
mechanisms. In Equations 4-95 and 4-96, the terms factorjmp, divisorjmp, and xmassjmp are
mechanism specific apportioning factors that will be defined further below. Note that, in some
cases a pollutant will contribute to only one mechanism species, but in other cases a pollutant
will contribute to more than on species. For example, emissions of ammonia are always assigned
to a model species named NH3, but NOx is allocated between two mechanism species NO and
NO2. The details of the allocation procedure for both the RADM2 and CB4 mechanisms will be
described further below. First, however, the derivation and use of the rogtotogmp conversion
factor is described.

ROG-to-TOG Adjustment

The ROG (Reactive organic gas) to TOG conversion portion of the speciation process is
selected when the emission inventory is loaded by INPRO (see section 4.2.2). VOC substances
deemed non-reactive (e.g., methane and ethane) are often excluded from inventories, and some
emission measurement techniques do not capture all discrete compounds in an emission stream
(e.g, formaldehyde). The ROG-to-TOG adjustment factor is used to account for those missing
components. It is calculated using Equations 4-97 and 4-98:

4-83
 EPA/600/R-99/030

mistogm ' 'k ntsrdk,m (4-97)

rogtotogm ' 1.0 / mistogm (4-98)

where: mistogi is the sum of mass fractions of the discrete VOC compounds deemed non-
reactive or not included in an emission inventory for source category m (grams
of missing compound per gram of TOG)
ntsrdk,m is the mass fraction of the missing or non-reactive compound k in the
emission stream for source category m
rogtotogm is the ROG to TOG conversion factor

Note that the rogtotogm,p adjustment factor applies only to anthropogenic emissions of VOC, and
thus the p subscript used in Equation 4-95 and 4-96 has been dropped here. In equations 4-95
and 4-96, rogtotogm,p is set to 1.0 for all pollutants other than ROG and also set to 1.0 when
VOC emissions of biogenic origin are being computed.

Carbon Bond 4 (CB4) Speciation Factors

The CB-4 chemical mechanism is widely used in air quality model simulations of ozone
concentrations at urban and regional spatial scales. As described in the introductory part of this
section, the basis of the CB-4 mechanism is that reactivity of organic compounds in the
atmosphere can reasonably be simulated by mechanism species that represent different carbon
bond types. A detailed description of the CB-4 mechanism is contained in Section 8.2.1. The
focus of the discussion here is on how the CB-4 apportioning factors used in Equations 4-95 and
4-96 are calculated. Since speciation of anthropogenic VOC emissions is done differently than
for other pollutants, the discussion is divided into two subsections.

Anthropogenic VOC speciation. For the CB4 mechanism, the apportioning factor divisorjmp in
equation 4-95 is set to 1.0 for all anthropogenic VOC emission calculations(i.e., it is essentially
not used). The factorjmp term for anthropogenic VOC is set equal to a split factor term sjl that is
computed for a specified VOC emission profile l, which in turn is assigned to a source category
m (i.e., an SCC or ASC). The split factor for each profile and mechanism species is computed as
follows:

sfj,l ' 'k xmfk,l / mw k ( xnumj,k (4-99)

where: sfj,l is a molar split factor for CB4 species j and VOC profile l (moles of CB4
species/gram of TOG)
xmfk,l is a mass fraction of discrete VOC compound k in VOC profile l (grams of
discrete VOC/grams of TOG)
mwk is the molecular weight of discrete VOC compound k (grams of discrete
VOC/mole of discrete VOC)

4-84
EPA/600/R-99/030

xnumj,k is the moles of discrete VOC compound k assigned to CB4 species j (moles
of CB4 species/mole of discrete VOC)

In Equation 4-96, the xmassjmp term apportions mass emissions to CB4 species for a given
pollutant and source category. Analogous to the CB4 molar split factor defined above, the
xmassjmp term is set equal to a mass split factor that apportions the mass of a discrete VOC
compound according to the fraction of carbon atoms that is assigned to a mechanism species:

xmassj,l ' 'k ( xmfk,l ( cnj ) / cnk (4-100)

where: xmassj,l is the mass split factor for CB4 species j and VOC profile l (moles of CB4
species/gram of TOG)
xmfk,l is the mass fraction of discrete VOC compound k in VOC profile l (grams of
discrete VOC/grams of TOG)
cnj is the number of carbon atoms in CB4 species j
cnk is the number of carbon atoms in discrete VOC compound k

Again, one VOC profile is assigned to each source category, and thus split factors are generated
for each source category/mechanism species combination.

Other speciation. For pollutants and source categories other than anthropogenic VOC, the terms
factorjmp, divisorjmp, and xmassjmp are simply assigned numeric values that determine the
allocation of pollutant emissions. These are summarized in Table 4-10.

For all pollutants (except particles) in Table 4-10, the divisor corresponds to the molecular
weight of the pollutant and is used in equation 4-76 to convert mass emissions of the pollutant to
moles of pollutant. The two split factors factorjmp, and xmassjmp are used to apportion the
pollutant emissions to the mechanism species. A factor of 1.0 indicates a 1:1 correspondence.
Factors different from one indicate a disaggregation or lumping of pollutant emissions into
individual species.

CO and NH3 are treated explicitly in many air quality models, and therefore emissions of these
compounds are not split into other components. Similarly, particle emissions are treated
explicitly so no lumping or dis-aggregation is necessary for them either.

When SOx emissions are speciated, 97% of the SOx mass is treated as SO2. The remaining 3%
of the SOx mass corresponds to SO4 (or SULF), but is dropped from further consideration. SO2
and SO4 emissions are treated as explicit species.

4-85
 EPA/600/R-99/030

Table 4-10. CB4 Split Factors for Pollutants Other Than Anthropogenic VOC

Source Category Pollutant CB4 factorjmp, divisorjmp, xmassjmp

Species

All CO CO 1.0 28.0 1.0

All NH3 NH3 1.0 17.0 1.0

All AERO AERO 1.0 1.0 1.0

All PM10 PM10 1.0 1.0 1.0

All PM2_5 PM2_5 1.0 1.0 1.0

All except BIO NOX NO 0.62 30.0 0.62

All except BIO NOX NO2 0.05 46.0 0.05

All SO2 SO2 1.0 64.0 1.0

All SOX SO2 0.97 64.0 0.97

All SO4 SULF 1.0 96.0 1.0

BIO NO NO 1.0 30.0 1.0

BIO Isoprene ISOP 1.0 68.12 1.0

BIO OVOC NR 0.5 148.0 0.05

BIO OVOC OLE 0.5 148.0 0.10

BIO OVOC PAR 8.0 148.0 0.85

BIO TERP ALD2 1.5 136.23 0.3

BIO TERP OLE 0.5 136.23 0.1

BIO TERP PAR 6.0 136.23 0.6

BIO TERP TERPB 1.0 136.23 1.0

In MEPPS, it is assumed that NOx is composed of 95% NO and 5% NO2 (by mass). However,
actual NOX composition can vary from 89%/11% to 95%/5%. In a few cases, a small percentage
(<2%) of NOX emissions is assumed to be nitrous acid (HONO). Since NOx emissions are
typically reported as NO2 mass, it is necessary to normalize the NOx emissions by the molecular
weight of NO2. Hence, the molar split factors (factorjmp) for NO and NO2 are calculated as
follows:

sfNO ' xmfNO ( mwNO / mwNO2 (4-101)

4-86
EPA/600/R-99/030

sfNO2 ' xmfNO2 ( mwNO2 / mwNO2 (4-102)

With a NOx composition of 95% NO (as NO2) and 5% NO2 by mass, the split factors for NO
and NO2 become 0.62 and 0.05, respectively. The mass split factors are calculated in the same
manner, and thus have the same values as for the molar split factors.

The speciation of biogenic emissions in CB4 is the same as used in the Regional Oxidant Model
(EPA, 1989). Isoprene is treated as an explicit species and terpene emissions are speciated into
1.0 mole of OLE, 6.0 moles of PAR, and 3.0 moles of ALD2. Terpenes are also assigned to the
model species TERPB for special processing in the aerosol module of the CMAQ. The apparent
double counting of terpene emissions is accounted for in the CMAQ processing however.
Finally, the category OVOC (other VOCs from biogenic sources) is apportioned to 1.0 mole of
OLE, 8.5 moles of PAR, and 0.5 mole of NR. Here it assumed that number of carbon atoms in a
OVOC molecule is 10.

RADM2 Speciation Factors

The RADM2 chemical mechanism was developed for and has been used principally in regional
air quality simulations of sulfur dioxide (SO2) and oxides of nitrogen for acid rain assessment
(Walters and Saegar, 1990). A detailed description of the RADM2 mechanism is given in
section 8.2.2. The speciation procedure used for RADM2 is fully described in Walters and
Saeger (1990), and will only be summarized here. This section focuses on how the RADM2
apportioning factors used in Equations 4-95 and 4-96 are calculated. Again, the discussion is
divided into two subsections, one dealing with the speciation of anthropogenic VOC emissions
and one with all other speciation.

Anthropogenic VOC speciation. The RADM2 mechanism requires that discrete organic VOC
compounds be lumped into 15 mechanism species based on common reactivity and reaction
products. As with the CB4 mechanism, the apportioning factor divisorjmp in Equation 4-95 is set
to 1.0 for all anthropogenic VOC emission calculations, and thus is not used. Also as with the
CB4 mechanism, The factorjmp term for anthropogenic VOC is set equal to a split factor term sjl
that is computed for a specified VOC emission profile l, which in turn is assigned to a source
category m (i.e., an SCC or ASC). However, the RADM2 split factors are calculated using the
following equation:

sfj,l ' 'k xmfk,l ( afacj,k ( rfacj,k / mw k (4-103)

where: sfj,l is the molar split factor for RADM2 species j and VOC profile l (moles of
RADM2 species/gram of TOG)
xmfk,l is the mass fraction of discrete VOC compound k in VOC profile l (grams of
discrete VOC/grams of TOG)

4-87
 EPA/600/R-99/030

afacj,k is the allocation factor listed in Table 4-6 for RADM2 species j and discrete
VOC compound k

rfacj,k is the reactivity factor listed in Table 4-6 for RADM2 species j and discrete
VOC compound k
mwk is the molecular weight of discrete VOC compound k (grams of discrete
VOC/mole of discrete VOC)

In the RADM2 speciation process, each discrete VOC compound is assigned to one of the 32
lumped species categories, and that assignment is used to further lumped to one of the 15 species
in the condensed group. See Middleton et al. (1990) for information on how the allocation and
reactivity factors are derived.

In Equation 4-96, the term xmassjmp apportions mass emissions to RADM2 species for a given
pollutant and source category. Analogous to the RADM2 molar split factor defined above, the
term xmassjmp is set equal to a mass split factor that apportions the mass of a discrete VOC
compound according to the mass fraction of the discrete VOC compound in the emission stream
and the mechanism specific allocation factor:

xmassj,l ' 'k xmfk,l ( afack,l (4-104)

where: xmassj,l is the mass split factor for RADM2 species j and and VOC profile l (moles
of CB4 species/gram of TOG)
xmfk,l is the mass fraction of discrete VOC compound k in VOC profile l (grams of
discrete VOC/grams of TOG)
afacj,k is the allocation factor listed in Table 4-6 for RADM2 species j and discrete
VOC compound k

Other speciation. For pollutants and source categories other than anthropogenic VOC, the terms
factorjmp, divisorjmp, and xmassjmp are simply assigned numeric variables that control the
allocation of emissions, similar to what is done for the CB4 mechanism. The factors for
RADM2 are summarized in Table 4-11.

Except for the biogenic categories, the contents of Table 4-11 are identical to those in Table 4-10
for the CB4 mechanism. Thus, the reader is referred to the corresponding CB4 section for
information on those source category/pollutant combinations. The biogenic portions of the two
tables are similar in that isoprene is treated as an explicit species in RADM2, and terpenes and
other VOCs (OVOC) of biogenic origin are apportioned to mechanism species. In RADM2,
terpenes are apportioned entirely to OLI (Middleton et al., 1990). As with the CB4 mechanism
,terpenes are also assigned to the model species TERPB for special processing in the aerosol
module of the CMAQ, but the apparent double counting of terpene emissions is accounted for in
the CMAQ processing however. Finally, the assignment of other VOC species to RADM2

4-88
EPA/600/R-99/030

species is carried out in a manner analogous to that for CB4: 85% is assumed to correspond to
slow reacting alkanes, 15% to fast reacting olefins, and 5% is assumed to be non-reactive. Thus,
in RADM2 85% of the OVOC mass is assigned to the category containing the slowest reacting
alkanes (category 27 in Table 4-6). The split factor is obtained by multiplying the mass fraction
(.85) by the reactivity factor (0.404) to give a molar split factor of 0.343. The 10% of the mass of
OVOC assumed to be reactive olefins is assigned to the OLI category. Both the mass and molar
split factors for this compound are simply 0.1.

Table 4-11. RADM2 Split Factors for Pollutants Other Than Anthropogenic VOC
Source Category Pollutant CB4 Species factorjmp, divisorjmp, xmassjmp

All CO CO 1.0 28.0 1.0

All NH3 NH3 1.0 17.0 1.0

All AERO AERO 1.0 1.0 1.0

All PM10 PM10 1.0 1.0 1.0

All PM2_5 PM2_5 1.0 1.0 1.0

All except BIO NOX NO 0.62 30.0 0.62

All except BIO NOX NO2 0.05 46.0 0.05

All SO2 SO2 1.0 64.0 1.0

All SOX SO2 0.97 64.0 0.97

All SO4 SULF 1.0 96.0 1.0

BIO NO NO 1.0 30.0 1.0

BIO Isoprene ISO 1.0 68.12 1.0

BIO OVOC HC3 0.343 148.0 0.85

BIO OVOC OLI 0.1 148.0 0.1

BIO TERP OLI 1.0 136.23 1.0

BIO TERP TERPB 1.0 136.23 1.0

4.2.6 Output Processor (OUTPRO)

The results of MEPPS processing must be in a form that is useful to the Models-3 framework and
readily evaluated for the substantive content and quality of the data. The OUTPRO processes
spatially and temporally allocated, speciated emission data files to prepare them for use by the
Models-3 framework and its components, including CMAQ. In addition, OUTPRO prepares
many standard and user-defined emission summary reports. The processing includes the
following items:

4-89
 EPA/600/R-99/030

C Merging of temporally allocated point, area, biogenic, and mobile source emission data
files.

C Merging of speciated point, area, biogenic, and mobile source emission data files .

C Merging of point, area, biogenic, and mobile source files into a consolidated two-
dimensional emission file. Merging is optional; and data type files may be maintained
and output separately.

C Preparation of three-dimensional point source emission files for use by the plume rise and
Plume-in-Grid models available in CMAQ. The user may define groups of similar stacks
by specifying percentage tolerance differences in different physical stack properties. In
addition, the user may categorize minor, major, and major elevated point source emission
(MEPSE) stacks by their emission rates and/or stack properties . The classification
criteria are not pre-specified, but are set by the user. Specifically, any combination of
emissions of specified pollutants in tons/day and/or physical stack parameters including
height, diameter, flow rate, or exhaust velocity may be used to define a major or MEPSE
point source. Generally, MEPSE stacks are the largest of point sources, such as electric
utility stacks. Major point sources are usually somewhat smaller, but significant point
sources. Source-specific information is necessary to classify MEPSE or major point
sources. Smaller sources not classified as MEPSE or major point sources or without
source-specific information are referred to as minor point sources, and are usually
included with the surface area sources by OUTPRO, unless specified otherwise by the
user.

C Conversion of output file format to NetCDF I/O API format. This allows other
components of the Models-3 framework, including the CMAQ and visualization tools, to
use the emission files. The CMAQ accepts the emission files and processes them for
input to CMAQ using the Emission Chemical Input Processor (ECIP). The ECIP is
described in Section 7.

C Preparation of summary reports of the processed emission files, including summaries by

primary emission type (point, area, biogenic, mobile), by geographic area (grid area, grid
cell, state, county, etc.), by source category code and groups of codes (tiers, or
combinations of these items. In addition, the reports rank emission values by amount,
type, geographic area, etc.

4.3 Models-3 Emission Projection Processor (MEPRO)

4-90
EPA/600/R-99/030

It is often necessary to iteratively project emissions and emission control combinations to future
years in order to model future year ambient pollutant concentrations relative to regulatory
standards. Consequently the Models-3 system includes a projection and control system, the
Models-3 Emission Projection (MEPRO) processor to perform the necessary work. The MEPRO
was developed from the Multiple Projection System (MPS). The MPS was originally designed
as a stand-alone software tool to assist the EPA in projecting and tracking the “reasonable further
progress” of regulatory emission reduction programs for criteria pollutants (Monroe et al., 1994).
It may be invoked from Strategy Manager from the Models-3 framework or run separately.
Because MPS was designed as a PC application using the Superbase® programming language,
MEPRO must be run under an emulator (SoftWindows®) if it used in an UNIX operating system.
Currently, MEPRO is more efficiently run on a PC workstation with a Windows NT® operating
system. Figure 4-7 illustrates the relationship of MEPRO with the rest of MEPPS.

Using MEPRO, the user may edit regulatory control factors, control efficiency, rule penetration,
rule effectiveness, as well as "across-the-board" emission adjustments and source category code-
specific emission projections . Projected emission inventory data, and/or inventory data with
revised emission controls and efficiencies are passed to EMPRO (in the
/raw_data/$EMS_DOMAIN/common/ directory) for processing to the NetCDF I/O API format
required for air quality modeling by CMAQ. Mobile source emissions are not projected in
MEPRO. Instead, VMT is projected, and the projected mobile sources emissions are computed in
the mobile source model of EMPRO.

The MEPRO will project emissions of NOx, VOC, and CO for each year from 1991 through
2010, using the 1990 EPA inventory for the base year. The capability to project SO2 and PM
emission data may be provided in the future. Projections are by source category code of
emissions for any area in the United States. The base year and the annual projection factors will
be updated as new data become available. The annual growth factors for VOC, NOx, and CO are
taken from look-up tables containing economic growth factors by county for the United States.
The growth factors are based on economic forecasts applied to specific source category codes
by the Economic Growth Analysis System (EGAS) (US EPA, 1995a).

The EGAS system is a PC-based tool which uses a hierarchical three-tiered approach to generate
growth factors. Tier 1 is the National Economic Tier. It includes an economic model by
Regional Economic Models, Inc (REMI) (Treyz et al., 1994) which is primarily based on the
Bureau of Labor Statistics (BLS) American Workforce 1992-2005 projections. After 2005, the
BLS moderate growth labor force participation rates and the Census Bureau’s middle population
projections for the United States are used to forecast the labor force. The second tier is the
Regional Economic Tier, the results of which are overlain on the Tier 1 results. The Regional
Economic Tier contains separate economic models developed by REMI for each non-attainment
area and attainment area of each state. The largest area addressed by one model is a state. The
third tier is the Growth Factor Tier, which contains commercial, residential, industrial, and
electric utility models, and a VMT growth module. The commercial, residential, and industrial
energy models were developed by the Argonne National Laboratory and were used in the

4-91
 EPA/600/R-99/030

National Acid Precipitation Assessment Program (Boyd et al., 1990). Electrical Utilities
projection was accomplished using the Neural Network Electric Utility Model (EUMOD). The
default economic projection tables from EGAS are based on U.S. Bureau of Labor Statistics
forecasts. Projection factor tables based on forecasts by the Wharton School of Economics are
provided as an option.

Figure 4-7 MEPRO Relationship to Other MEPPS Components

The method of applying regulatory and growth projection factors to point and area source
emission data to obtain future year daily controlled emission data is summarized by Equation 4-
105. Because MEPRO projects only the VMT data for mobile source emissions, the projection
method is slightly different (Equation 4-106). The projected VMT data are passed to the
EMPRO mobile-source emission model to be converted into projection mobile source emission
data.

PCONE = DCONE * PGF * (1 + AF/100) *

[1-(PCE * PRE * PRP)]/[1-(CE * RE * RP)] (4-105)

where: PCONE is the point or area source future-year daily controlled emissions
DCONE is the base year daily-controlled emissions

4-92
EPA/600/R-99/030

PGF is the projected year growth factor (percent)

AF is the projected year emission adjustment factor (percent)
PCE is the projected year control efficiency
PRE is the projected year rule effectiveness
PRP is the projected year rule penetration
CE is the (adjusted) base year control efficiency
RE is the (adjusted) base year rule effectiveness
RP is the (adjusted) base year rule penetration

For mobile source data:

FPE = BVMT * EF * PGF * (1 + AF/100) (4-106)

where: FPE is the future projected emissions (on-road mobile)

BVMT is the base VMT by vehicle class by facility class by year
EF is the emission factor
PGF is the projected year growth factor
AF is the projected year emission adjustment factor

Additional details for EGAS and MPS are given the references cited above.

4.4 Emission Processing Interface

An accurate characterization of the spatial and temporal variability of emissions at the surface
and aloft is vital for realistic air quality grid modeling. The MEPPS creates separate emission
files for surface area and elevated point sources for a particular domain and time period to be
modeled. Consequently, an interface processor program was needed to efficiently consolidate
these various emissions types into a single, hourly gridded data file for use in grid model
simulations.

4.4.1 Overview of Key Features of ECIP

The Emission-Chemistry Interface Processor (ECIP) serves as the key link between the MEPPS
system and the CMAQ Chemistry Transport Model (CCTM). The primary function of ECIP is
to generate an hourly 3-dimensional (3-D) emission data file for the CCTM from the individual
emission file types produced by the MEPPS.

The schematic diagram in Figure 4-8 shows the principal input files used to drive ECIP.
The notable elevated point sources were likely separated into major and MEPSE (Major Elevated
Point Source Emissions) source groups, based on a user-specified emission rate criterion during
MEPPS processing. The MEPSE group contains the largest point source emissions and is
intended to be specially simulated by the CCTM Plume-in-Grid (PING) treatment. However, the
PING treatment is an optional capability of the CCTM and when it is not exercised, the MEPSE
emission file must be processed by ECIP for the modeling scenario so that the significant

4-93
 EPA/600/R-99/030

emissions from these point sources are included in the 3-D emission data file with the other
major point source emissions. Thus, this capability to include or omit the MEPSE emissions in
ECIP allows for CCTM runs to be performed without PING or with the PING treatment,
respectively, while not requiring separate runs of the MEPPS emissions system. A companion
stack parameter file for each point source emission file is needed by ECIP for plume rise
calculations. In addition, meteorological data files generated by MCIP are also required to run
ECIP in order to simulate point source plume processes. The ECIP 3-D emission file displayed
in Figure 4.6 contains the surface area anthropogenic and biogenic emissions, and the elevated
emissions from major point sources and the MEPSEs, if appropriate. The 3-D emissions output
file from ECIP is ready for direct input into the CCTM.

AREA MEPSE MAJOR PS MCIP

STACK
PS EMIS MET
EMIS (opt) EMIS PARMS
DATA

ECIP

3-D EMIS

CCTM

Figure 4-8 Schematic Flow Diagram of the Input Files and 3-D Emission Data File for
CCTM generated by ECIP

A specific set of tasks is performed by ECIP for preparation of the 3-D emission file. The 2-D
gridded area emissions are incorporated into the first model layer since they represent near-
surface releases. In contrast, each elevated point source plume must be subjected to plume rise
and initial vertical spread processes prior to the allocation of the plume emissions into the proper
grid cells aloft. Another capability of ECIP can be applied in case the CCTM domain is smaller
than the emission domain. ECIP can perform "spatial windowing" of the emissions needed for a
particular CCTM domain from a larger MEPPS emissions domain.

4-94
EPA/600/R-99/030

The methods used in ECIP to treat the processes impacting point source plumes are described in
subsequent sections. While these approaches employ existing scientific techniques, it is
recognized that other formulations exist. However, the modular design of the Models-3 coding
structure allows for the implementation and application of alternative algorithms to treat a
particular process.

4.4.2 Plume Rise of Point Source Emissions

The rise of a buoyant plume above stack height is strongly dependent upon the initial stack
parameters and atmospheric vertical structure at the time of release. A realistic determination of
the height of final plume rise is important to incorporating the plume emissions into the proper
vertical layer(s) of the model. The initial buoyancy flux (Fb), which is a key parameter in plume
rise formulas, is given by

g(Ts&Ta) (Vsd 2)
Fb ' (4-107)
4 Ts

where Ts and Ta are the stack exit temperature and ambient temperature at stack top,
respectively. Other notable stack parameters in Equation 4-107 include the plume exit velocity
(Vs) and stack diameter (d), while g is gravity. Clearly, the magnitude of Fb is greatly influenced
by these stack exit parameters. For buoyant plumes, which exist for the vast majority of point
sources, Fb is greater than zero since Ts > Ta. However, if Ts < Ta , then Fb is set to zero.

The vertical profile of wind and temperature also greatly impact plume rise. In particular,
advances in the accuracy of plume rise estimates have resulted from taking into consideration the
vertical variations in the thermal stability and wind structure which frequently display a strong
height dependency in the atmosphere. A key atmospheric stability parameter (s) used to
distinguish the different stability regimes aloft is defined by

s ' (g/Ta)(d2/dz) (4-108)

where d2/dz is the vertical potential temperature gradient. The value of s is employed as a
criterion to apply the appropriate stability-dependent plume rise formula. However, for the
initial calculation of plume rise, the convective velocity scale (H*) is used as an indicator variable
to identify the particular stability regime and it is defined by
g
H( ' wT (4-109)
Ts

4-95
 EPA/600/R-99/030

where Ts is the 1.5m air temperature, and wT is the surface heat flux covariance. Unstable
conditions are defined for H* > 0.03H*min , while stable conditions exit when H* < -0.03H*min .
Neutral conditions occur for the range of values between these criteria. The default value of
H*min has been set to 10-4 m2/s3.

The layer-by-layer approach described by Turner (1985), as originally suggested by Briggs

(1975), has been applied in ECIP for the determination of final plume rise ()h) based on the
stability of each vertical layer. This practical scheme takes advantage of the vertical resolution
of the hourly, temperature and wind profiles and other 2-D meteorological parameters provided
by the MM5 dynamic mesoscale meteorological model outputs as postprocessed through MCIP.
The method, as outlined by Turner (1985), is an iterative approach which computes plume rise
through each layer. An initial plume rise calculation is performed using meteorological variables
derived at the stack top height with a stability-dependent plume rise formula at this level. Two
methods, linear interpolation or a surface similarity scheme by Byun (1990), are available for
deriving temperature and wind at stack height from the modeled profile values. For tall stacks,
negligible differences were found in the derived values between these methods. If the projected
effective plume rise height (he = hs + )h) exceeds the top of the layer containing the stack top,
the amount of rise is limited to the height of the current layer top. Then residual buoyancy flux
(FR) is determined with an inverted form of the plume rise equation just applied. Using FR, the
procedure is repeated to determine the plume rise using the profile parameters for the next higher
layer. This method is applied over successive layers until the buoyancy flux is completely
exhausted. The plume rise ceases at the level where Fb = 0.

A set of analytical plume rise equations presented by Briggs (1984) for different atmospheric
stabilities have been utilized in ECIP for all point sources. The various plume rise equations are
provided below. The final effective plume centerline height (he) is found by adding the
computed plume rise ()h) to hs.

4.4.2.1 Plume Rise Treatment for Stable Conditions

For stable atmospheric conditions, plume rise is taken from Briggs (1984) equation for bent-over
plumes.
1/3
Fb
)h ' 2.6 (4-110)
us

where u is the wind speed for the layer. Equation 4-88 is applied when H* < -0.03H* for the
initial plume rise computation at the stack, and when s > 10-5 for subsequent layers in the current
approach.

4-96
EPA/600/R-99/030

4.4.2.2 Plume Rise Treatment for Unstable Conditions

The plume rise formula proposed by Briggs (1984) during unstable conditions is given by
&2/5
)h ' 3[Fb/u]3/5H( (4-111)

However, Briggs (1983) suggested a reasonable approximation for H* which permits Equation 4-
111 to be applied in the following form.
)h ' 30 [Fb/u]3/5 (4-112)

The rationale for the simplification is due to the lack of data for evaluation to justify a more
complicated form. Equation 4-112 is applied with H* > 0.03Hmin for the first plume rise
computation at the stack and for higher layers for s < 10-5 .

4.4.2.3 Plume Rise Treatment for Neutral Conditions

The neutral formula developed by Briggs (1984) plume rise equation has been modified into the
following expression.

2 3/5 2 2/5
)h ' 1.2[Fb/(u u( )] [hs % 1.3Fb/(u u( )] (4-113)

where u* is the surface friction velocity. Equation 4-113 introduces minor differences from the
other form. Equation 4-113 is a more computationally efficient form suggested by Briggs
(1983, communication) since his original neutral formula requires iteration to solve. In the
plume rise algorithm of ECIP, the neutral plume rise equation is also solved during other
stability conditions in the layers aloft. A comparison between the two plume rise estimates is
made and the lower value is selected before proceeding.

4.4.2.4 Special Conditions

For the cases when Fb = 0, plume rise can occur due to momentum provided by the exit velocity
of the plume out of the stack. Therefore, a momentum rise formula (Turner, 1985) has also been
implemented to consider these situations and is given by
)hm ' 3d v s/u (4-114)

If Equation 4-114 is selected, no further plume rise computations are performed.

4-97
 EPA/600/R-99/030

Another situation occurs when the stack is below the PBL height (zi) under unstable conditions
and the condition (hs - zi) < 200 m applies. A treatment for limited plume penetration above the
PBL is determined when this situation is triggered. This condition most often occurs during the
morning period. Since zi is generally growing rapidly, this period is generally brief. To
consider plume penetration of the overlying stable layer, a practical algorithm employed
previously by Byun and Binkowski (1991) has been implemented based on Briggs (1984). If hs
is less than 200 m below zi, then the following equation is solved.

zb ' 3.9[Fb/us]1/3 (4-115)

If zi is greater than zb, , then the plume top height (zt) is set to zi and he is defined to be 2/3zt.
However, plume penetration is permitted when zi < zb in Equation 4-115. For this case, the
plume top height is defined to be the height of the top of the next higher layer and the effective
plume height is again computed as 2/3ht .

4.4.3 Method for the Treatment of Initial Vertical Plume Spread

Buoyancy-induced turbulence promotes plume expansion during the rise phase. A widely-used
method from Briggs (1975) designates the vertical thickness of a plume to be equivalent to the
amount of plume rise. With this method, the heights of the top and bottom of the plume are
determined by
ht ' h s % 1.5 )h
hb ' h s & 0.5)h (4-116)

where ht and hb are the heights of the plume top and bottom, respectively. Since the plume
thickness is directly related to the amount of plume rise, this approach leads to rather thick
plumes during the nocturnal period. Experimental plume dimension data suggest more limited
vertical thickness for plumes during the nighttime hours. As an alternative, an empirical form
has also been included. It is based on analyses of observed plume dimensions and vertical
temperature gradients (Gillani, 1996 communication). He found the best-fit empirical result is
given by
Fz ' A e (&B dT/dz) (4-117)

where the standard deviation of plume depth (Fz ) is a function of the vertical temperature
gradient (dT/dz) at he. Values for A and B are given by 10 and 117, respectively. A minimum
value specified for Fz is 3 m. With this method, ht and hb are determined to be ±2.15Fz above

4-98
EPA/600/R-99/030

and below hs , respectively. This approach provides for smaller plume thicknesses during the
nocturnal period.

4.4.4 Vertical Allocation of Plume Emissions

Rather than dumping the entire emissions of a plume into a single layer, an approach has been
developed to allow for the allocation of plume emissions into multiple layers since a plume can
often span more than one layer. This situation occurs often as more vertical layers are used in the
model since model layers are thinner.

Once ht and hb have been computed for each plume, these values, along with the heights of the
model layer interfaces (zz ) are employed to determine the fractional amount of plume overlap
across each layer. The method uses the fractional amount of the plume depth residing within a
layer in order to weight the amount of plume emissions incorporated into a particular layer. If
both ht and hb are contained within a particular layer, all the plume's emissions are allocated into
one layer. As noted above, the number of layers receiving plume emissions is also dependent on
the number of vertical layers in the model. A model configuration with fewer vertical layers
generally implies greater layer thicknesses.

4.4.5 Generation of 3-D Emissions

Once the plume rise and plume partitioning functions have been performed, the emissions from
each point source plume are transferred to the 3-D emission array which also contains the surface
area emissions in the first layer. The 3-D emission array is written at an hourly interval for the
entire simulation period to a data file in a format compatible for use in CCTM simulations.

4.5 Data Requirements

The following items are the data input requirements for operation of MEPPS. Those items that
must be supplied by the user are marked (U). Those items that are fixed internal lookup tables,
provided with Models-3, or that may be generated in MEPPS are marked (I).

C Complete annual (point, area, and mobile-source data by source category code) Regional,
National, or international emission inventories are necessary for regional modeling. The
inventories preferably should be in the ASCII format of the EPA National Emission
Trends (NET) inventories. However, IDA allows import and conversion of any inventory
with known fields formats in ASCII, SAS, or NetCDF format. Currently, the 1985
NAPAP, 1988 National Inventory, 1990 National Interim Inventory, and 1990 National
Emission Trends (NET) inventories for criteria pollutants are included. Limited data for
southern Canada are available as a part of these inventories (I). Emission inventories for
other areas, years, or pollutants must be supplied by the user (U).

C Existing hourly emission data, such as CEM data, may be used directly or substituted for
hourly emission data derived by temporal disaggregation from annual emission

4-99
 EPA/600/R-99/030

inventories. The hourly data may be imported through the File Converter and IDA. The
Models-3 system is includes 1995 CEM data in the SAS data set format provided by the
US EPA Office of Acid Rain (I). Other hourly data must be supplied by the user if
desired (U).

C Hourly meteorological data for surface temperature and solar radiation, in NetCDF I/O
API format, (from MM5 processed via MCIP) must be available for modeling biogenic
or mobile source emission data. If data files for the appropriate case area available, the
user may select them in MEPPS or when running Study Planner. Otherwise, it is
necessary to first run MM5 and output the meteorology files through MCIP (I).

C Chemical speciation profiles matched to source category codes are necessary for the
speciation processor (I). These may be updated with new information.

C Temporal allocation profiles, by source category code, must be available to accomplish

temporal allocation of emission data to hourly emission data (I). These may be updated
with new information

C Geographic coverages for surrogate spatial allocation (gridding) of emission data. The
user may supply additional coverages: Those supplied with MEPPS include (I):

Political boundaries at the county level for North America

Land-water boundaries and features for North America
County-level population information for North
America (with a gridded surrogate for Canada)
Federal Highway Administration major highway
coverage for the United States
TIGER-LINE detail road coverage for the United
States
Land cover for North America. Currently is at
county-scale for the United States and gridded
at coarser resolution for Canada and Mexico.
Land cover for the United States at one
kilometer resolution is anticipated in 1999

C Species emission factors for biogenic emission modeling (I)

C Standard Mobile 5a model input information by political area (usually state and/or
county), including vehicle fleet composition data, fuel type use by geographic area,
inspection and maintenance program information, etc, (U). A template with examples is
provided for the user to edit.

4-100
EPA/600/R-99/030

C Road silt loading, geographic distribution of paved and unpaved roads, vehicle fleet
composition data, fuel use data, and inspection and maintenance data are necessary as
input to the PART5 mobile particulate model.

C Geographic, source category code-specific growth projection factors. These are provided
for use with MEPRO (I).

C Source category and/or geographically specific emission control data, and regulatory
factors including control efficiency, rule effectiveness, and rule penetration for use in
MEPRO. These must be supplied by the user because controls are not standard, and in
fact are a key variable in examining different emission scenarios (U).

4.6 Plans for Improvement

Plans for improvement of MEPPS may be divided into long and short-term improvements. The
short-term improvements are those that are anticipated to be in the Models-3 release scheduled
for the summer of 1999.

Short-term Improvements

C The IDA will be improved to consolidate and further automate much of the quality
control processing, format and unit conversion, and data file manipulation. In particular,
format templates of internal formats will be added to assist importing of emission data
files into the system, and quality control for the CEM data will be enhanced.

C Although SAS® has proved to be a useful tool to date, increasing data handling
requirements will likely overwhelm the data handling capabilities of SAS®. Therefore, it
is necessary to convert MEPPS to a fully Models-3 framework compliant system in order
to take full advantage of the or object-oriented data base architecture of the system (eg.,
Orbix®) to more efficiently manage very large amounts of data. The use of the Sparse
Matrix Operator Kernel Emission (SMOKE) system, in conjunction with the Models-3
object-oriented architecture and expansion of existing functionality should substantially
improve performance because relatively inefficient processing sequences will not be
necessary, and it will not be necessary to manipulate all elements of large files for each
operation. More information about SMOKE can be found in Coats et al. (1995). An
initial (but not complete) version of SMOKE is planned for installation by summer 1999.

Long-term Improvements

C New mobile source emission estimation models are being developed by the U.S. EPA
Office of Mobile Sources. They are scheduled for completion late summer of 1999.
Mobile 6 will replace Mobile 5a and Mobile 5b, and an Off-road Mobile Source Model

4-101
 EPA/600/R-99/030

will be introduced. These models will be installed in MEPPS when they are available and
resources allow.

C The MEPPS contains the split factor assignments for two common chemical speciation
mechanisms, CB-4 and RADM 2, that may be selected by users. These speciation
mechanisms will also be in SMOKE. There are plans to also install the split factor
information and computational mechanisms necessary to use for the Statewide Air
Pollution Research Center (SAPRC) (Carter, 1988), and eventually, code necessary to
support the more complex Morphecule mechanism being developed at the University of
North Carolina.

C Additional quality control and reporting capabilities will be added to SMOKE, equaling
or surpassing those capabilities in currently in MEPPS.

C The emission data processing system could be enhanced to support nested grid structures.
Currently, there is only limited support for nested grid structures and no support for
multiscale grid structures. Under the current formulation, EMPRO (including the
gridding processor) must be run consecutively for each grid structure that exists within
the nested grid structure. Repetitive runs are inefficient, and computer resources are
poorly utilized because certain areas in the modeling domain will be processed more than
once. In addition, the gridding processor cannot generate rotated grids, which may be a
limitation for some applications. If an air quality modeling study requires a rotated
emission modeling grid, a knowledgeable ARC/INFO® user must prepare the grid
independently from menu options provided in Model-3 and MEPPS.

C Tools such as NetCDF and the I/O API will evolve to directly accommodate geographic
data references in their structure. This will allow manipulation of geographic data
(gridding, for example) to be accomplished without the use of non-conforming
commercial software tools now in Models-3 system.

4.7 References

Alados, I., I. Foyo-Moreno, and L. Alados-Arboledas, 1996: Photosynthetically Active

Radiation: Measurements and Modeling, Agri. and Forest Meteor., 78, pp. 121-131.

Beck, L., L.A.. Bravo, B.L. Peer, M.L. Saeger, and Y. Yan, 1994: A Data Attributes Rating
System, Presented at the International Conference on the Emission Inventory: Applications and
Improvement, Raleigh, NC., November 1-3, 1994, 12 pp.

Boyd, G.A., E.C. Kokkelenberg, and M.H. Ross, 1990: Sectoral Electricity and Fossil Fuel
Demand in U.S. Manufacturing: Development of the Industrial Regional Activity and Energy
Demand (INRAD) Model, Argonne National Laboratory, Argonne, IL.

4-102
EPA/600/R-99/030

Briggs, G.A., 1975: Plume Rise Predictions. In: Lectures on Air Pollution and Environmental
Impact Analyses, Workshop, Proceedings, Boston, MA, 1975, pp 59-111.

Briggs, G.A., 1983 communication: Plume rise equations used in EPA models. Meteorology
Division, Environmental Sciences Research Lab., Research Triangle Park, NC.

Briggs, G.A., 1984: Plume Rise and Buoyancy Effects. In: Atmospheric Science and Power
Production, D. R. Anderson, Ed., DOE/TIC-27601 (DE84005177), Technical Information
Center, U.S. DOE, Oak Ridge, TN, 850 pp.

Byun, D.W., 1990: On the Analytical Solutions of Flux-profile Relationships for the
Atmospheric Surface Layer. J. of Appl. Meteorol., 29, No. 7, 652-657

Byun, D.W. and F.S. Binkowski, 1991: Sensitivity of RADM to Point Source Emissions
Processing. Seventh AMS/AWMA Joint Conf. on Applications of Air Poll. Meteorol., New
Orleans, LA, Jan. 14-18, 1991, Preprints, Amer. Meteorol. Soc., Boston, MA., 1991, pp. 70-73.

Carter, W.P.L., 1988: Documentation for the SAPRC Atmospheric Photochemical Mechanism
Preparation and Emissions Processing Programs for Implementation in Airshed Models, Prepared
for the California Air Resources Board, Report A5-122-32.

Coats, C.J., Jr., 1995: High Performance Algorithms in the Sparse Matrix Operator Kernel
Emissions (SMOKE) Modeling System, Microelectronics Center of North Carolina,
Environmental Systems Division, Research Triangle Park, NC, 6 pp.

E.H. Pechan and Associates, 1994: Industrial SO2 and NOx Tracking System, EPA Contract No.
68-D1-0146, 38 pp.

Emission Inventory Improvement Program (EIIP), 1996: Biogenic Sources Preferred Methods,
Volume V, prepared for the EIIP Area Sources Committee by the Radian Corporation, Research
Triangle Park, NC, 104 pp.

Fratt, D.B., D.F. Mudgett, and R.A. Walters, 1990: The 1985 NAPAP Emissions Inventory:
Development of Temporal Allocation Factors, EPA-600/7-89-010d, U.S. Environmental
Protection Agency, Office of Research and Development, Washington, D.C., 209 pp.

Geron, C.D., A.B. Guenther, and T.E. Pierce, 1994: An Improved Model for Estimating
Emissions of Volatile Organic Compounds from Forests in the Eastern United States, J.
Geophys. Res., 99, D6, pp. 12773-12791.

Gery, M.W., G.Z. Whitten, J.P. Killus, and M.C. Dodge, M.C., 1989: A Photochemical Kinetics
Mechanism for Urban and Regional Scale Computer Models, J. Geophys. Res., 94, D10, 12295-
12956.

4-103
 EPA/600/R-99/030

Gillani, N.V., 1996: personal communication: Analyses results of plume data.

Guenther, A.B., P. Zimmermann, and M. Wildemuth, 1994: Natural Volatile Organic Compound
Emission Rate Estimates for U.S. Woodland Landscapes, Atmos. Environ., 28, 1197-1210.

Kinnee, E., C. Geron, and T. Pierce, 1997: United States Land Use Inventory for Estimating
Ozone Precursor Emissions, Ecol. Appl., 7, 46-58.

Middleton, P., W.R. Stockwell, and W.P.L. Carter, 1990: Aggregation and Analysis of Volatile
Organic Compound Emissions for Regional Modeling, Atmos. Environ., 24A, pp. 1107-1133.

Modica, L.G., D.R. Dulleba, R.A. Walters, and J.E. Langstaff, 1989: Flexible Regional
Emissions Data System (FREDS) Documentation for the 1985 NAPAP Emissions Inventory,
EPA-600/9-89-047, U.S. Environmental Protection Agency, Research Triangle Park, NC, 574 pp.

Monroe, C.C., T.A. Dean, and W.R. Barnard, 1994: Multiple Projections System: Version 1.0
User's Manual, EPA-600/R-94-085, 70 pp.

Moody, T., J.D. Winkler, T. Wilson, and S. Kersteter, 1995: The Development and
Improvement of Temporal Allocation Factor Files, EPA-600/R-95-004, 457 pp.

Moran, M.D., M.T. Scholtz, C.F. Slama, A. Dorkalam, A. Taylor, N.S. Ting, D. Davies, C.
Sobkowicz, P.A. Makar, and S. Venkatesh, 1998: An Overview of CEPS1.0: Version 1.0 of the
Canadian Emissions Processing System for Regional-Scale Air Quality Models, in the
Proceedings of a Specialty Conference on Emission Inventory: Planning for the Future, Air and
Waste Management Association, Pittsburgh, VIP-77, Volume I, 95-106.

Morris R.E., M.A. Yocke, T.C. Myers and V. Mirabella, 1992: Overview of the Variable-Grid
Urban Airshed Model (UAM-V), Paper presented to the Air and Waste Mangement Association,
85th Annual Meeting, Kansas City, MO, June 21-26, 1992, 13 pp.

Ozone Transport Assessment Group, 1997: Emission Inventory Development Report (Draft), 3
Volumes.

Pierce, T.E., B.K. Lamb, and A.R. VanMeter, 1990: Development of a Biogenic Emissions
Inventory System for Regional Scale Air Pollution Models, Paper presented to the Air and
Waste Management Association, 83rd Annual Meeting, Pittsburgh, PA, June 24-29, 1990, 16 pp.

Pierce, T., C. Heron, L. Bender, R. Dennis, G. Tennyson, and A. Guenther, 1998: The Influence
of Increased Isoprene Emissions on Regional Ozone Modeling, J. Geophys. Res., in press.

Saeger, M., J. Langstaff, R. Walters, L. Modica, D. Zimmerman, D. Fratt, D. Dulleba, R. Ryan, J.

Demmy, W. Tax., D. Sprague, D. Mudgett, and A.S. Werner, 1989: The 1985 NAPAP

4-104
EPA/600/R-99/030

Emissions Inventory (Version 2): Development of the Annual Data and Modelers' Tapes, U.S.
Environmental Protection Agency, Office of Research and Development, EPA-600/7-89-012a,
692 pp.

Stockwell, W.R., P. Middleton, and J.S. Chang, 1990: The Second Generation Regional Acid
Deposition Model Chemical Mechanism for Regional Air Quality Modeling, J. Geophys. Res.,
95 (D10), pp. 16,343-16367.

Treyz, G. et al.: 1994: Model Documentation for the REMI EDFS-14 Forecasting and Simulation
Model, Regional Economic Models, Inc., Amherst, MA.

Turner, D.B., 1985: Proposed Pragmatic Methods for Estimating Plume Rise and Plume
Penetration through Atmospheric Layers. Atmos. Environ., 19, 1215-1218.

U.S. EPA, Office of Mobile Sources, 1985: Size Specific Total Particulate Emission Factors for
Mobile Sources, EPA-460/3-85-005, 294 pp.

U.S. EPA, Office of Air Quality Planning and Standards, 1988: Air Species Manual: Volume 1,
Volatile Organic Compound Species Profiles, EPA-450/2-88-003a, 492 pp.

U.S. EPA, Office of Research and Development, 1989: Development of the Regional Oxidant
Model Version 2.1, U.S. EPA Technical Report.

U.S. EPA, Office of Mobile Sources, 1991: User's Guide to MOBILE 4.1 (Mobile Source
Emission Factor Model), EPA-AA-TEB-91-01, 354 pp.

U.S. EPA, Office of Air Quality Planning and Standards, 1992: User's Guide for the Urban
Airshed Model, Volume IV: User's Manual for the Emissions Preprocessor System 2.0, EPA-
450/4-90-007D(R), 497 pp.

U.S. EPA, Office of Air Quality Planning and Standards, 1993: Regional Interim Emission
Inventories 1987-1991), EPA-454/R-93-0212, 54 pp.

U.S. EPA, Office of Air Qaulity Planning and Standards, 1994: Emissions Inventory for the
National Particulate Matter Study, Final Draft, Prepared by E.H. Pechan, Inc., EPA Contract No.
68-D3005, WA No. 0-10, 83 pp.

U.S. EPA, Office of Air Quality Planning and Standards, 1995: Compilation of Air Pollutant
Emission Factors: Volume II: Mobile Sources, AP-42, Fifth Edition, 293 pp.

U.S. EPA, Office of Research and Development, 1995a: Economic Growth Analysis System:
User's Guide Version 3.0, EPA-600/R-95-132b, 89 pp.

4-105
 EPA/600/R-99/030

U.S. EPA, Office of Mobile Sources, 1995b: Draft User’s Guide to PART5: A Program for
Calculating Particle Emissions from Motor Vehicles, EPA-AA-AQAB-94-2, 67 pp.

U.S. EPA, Office of Mobile Sources, 1996: User's Guide for MOBILE 5a (Mobile Source
Emission Factor Model), EPA-AA-TEB-92-01, 176 pp.

U.S. EPA, Office of Air Quality Planning and Standards, 1997: Air CHIEF, Version 5 (Compact
Disk containing air quality speciation profiles, emission factors, and other emission-related data),
Research Triangle Park, NC.

U.S. EPA, Office of Air Quality Planning and Standards, 1998: National Air Pollutant Emission
Trends Procedures Document, 1900-1996 Projections 1999-2010, EPA-454/R-98-008, 712 pp.

U.S. EPA, Office of Research and Development, 1995: Models-3: Volume 7, System
Requirements (Draft) .

U.S. EPA, Office of Research and Development, 1995: Models-3: System Design (Draft).

Walters, R.A. and M.L. Saeger, 1990: The 1985 NAPAP Emissions Inventory: Development of
Species Allocation Factors, EPA-600/7-89-010f, 470 pp.

Wilkinson, J.G., C.F. Loomis, D.E. McNally, R.A. Emigh, and T.W. Tesche, 1994: Technical
Formulation Document: SARMAP/LMOS Emissions Modeling System (EMS-95), Final
Report, Lake Michigan Air Directors Consortium and the California Air Resources Board, AG-
90/TS26 and AG-90/TS27, 120 pp.

Williams, E., A. Guenther, and F. Fehsenfeld, 1992: An Inventory of Nitric Oxide Emissions
from Soils in the United States, J. Gaffes. Res., 97, 7511-7519.

Yienger, J.J. and H. Levy II, 1995: Empirical Model of Global Soil-Biogenic Nox Emissions, J.
Gaffes. Res., 100, No. D6, 11447-11464.

This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.

4-106