United States
Environmental Protection
Agency
Office of Air Quality
Planning and Standards
Research Triangle Park, NC 27711
EPA-450/4-92-011b
June 1992
Air
& EPA
GUIDELINE FOR REGULATORY
APPLICATION OF THE
URBAN AIRSHED MODEL
FOR AREAWIDE CARBON
MONOXIDE
VOLUME II: APPENDICES
-------
EPA-450/4-92-011b
GUIDELINE FOR REGULATORY
APPLICATION OF THE
URBAN AIRSHED MODEL
FOR AREAWIDE CARBON
MONOXIDE
VOLUME II: APPENDICES
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Technical Support Division
Research Triangle Park, NC 27711
y S. Environment^'. '•^•~i
June 1992 Refijon 5, Library (r >- •'-') , . , ,
c JacKSon bom- ara, •.t.ut ..-
»».- -'- ""-'
Chica: !L
-------
This report has been funded by the United States Environmental Protection Agency (EPA) under
contract 68D00124 to Systems Applications International (SAI). Thomas N. Braverman served
as the EPA work assignment manager. Any mention of trade names or commercial products is
not intended to constitute endorsement or recommendation for use.
11
-------
Contents
Appendix A: Recommended Modeling Protocol Contents
Appendix B: Performance Measure Formulations
Appendix C: Technical Discussion of UAM Model Inputs with Example Application
111
-------
-------
Appendix A
RECOMMENDED MODELING PROTOCOL CONTENTS
-------
APPENDIX A
RECOMMENDED MODELING PROTOCOL CONTENTS
Table 2-1 of Chapter 2 lists recommended contents for a modeling protocol. This
appendix gives a general description of each component to aid in the development of the
protocol.
UAM MODELING STUDY DESIGN
Background and Objectives
The Protocol Document should describe the policy and technical objectives of the study
and pertinent background information such as the legislative mandate under which the study is
being done.
Schedule
A complete schedule for all phases of the project is needed. The critical paths and
deadlines should be identified and discussed, as should a schedule for addressing critical issues
that require special attention, such as air quality and meteorological data preparation and quality
assurance, episode selection, and emission inventory preparation and quality assurance.
Deliverables
A list of the interim and final deliverables for the modeling study should be specified.
A-l
-------
Modeling Policy Oversight/Technical Committees
The composition and responsibilities of the modeling policy oversight and technical
committees should be specified to the extent possible. Meeting frequency and circumstances
for convening a meeting should be identified. Because technical conflicts may arise, a resolution
process for handling them should be included.
Participating Organizations
The organizations that are sponsoring the modeling study and those that may contribute
to it should be identified.
Relationship to Planning/Strategy Groups
Key planning agencies and others responsible for emission projections or other model
inputs should be identified, and the means by which these groups interact to obtain realistic
growth projections and control strategies should be discussed.
DOMAIN AND DATA BASE ISSUES
Data Bases
The proposed air quality and meteorological data bases should be described. The
completeness of the data base, techniques for filling in missing data, and quality assurance
procedures should be discussed.
Base Meteorological Episode Selection
The episode selection criteria should be detailed, including the methodology to group
candidate episodes into meteorological regimes. How the episodes will be used in the modeling
study should also be described.
A-2
-------
Modeling Domain
The protocol should describe the criteria for selecting the size and location of the
modeling domain. This would include a description of the MSA/CMSA area size, locations of
major sources outside the MSA/CMSA that may affect it, sensitivity analyses that may be
conducted to assess boundary effects on domain predictions, relationship of domain size to the
episodes selected for use in the modeling study, etc.
Horizontal Grid Resolution
The protocol should describe the horizontal grid resolution to be applied to the modeling
domain. If a resolution coarser than 2 x 2 km is chosen, justification for this choice should be
provided.
Number of Vertical Layers
The protocol should specify the number of vertical layers to be used in the UAM
simulations. If a layering scheme other than the one recommended in Chapter 3 is chosen,
justification for using the alternative layering should be given.
Emission Inventory
The assumptions, methodologies, and appropriate guidance references to be used in
constructing the modeling emission inventory should be described. Quality assurance procedures
should also be described.
Specification of Initial and Boundary Conditions
The techniques to be used to specify the initial and boundary conditions for the base
meteorological episodes and the attainment year should be described. The assumptions to be
used in forecasting attainment-year conditions should be documented.
A-3
-------
Wind Field Specification
The proposed techniques for specifying the wind fields should be described. The
procedures to be used to determine the representativeness of the simulated wind fields should
be technically justified and documented (see Chapter 3).
Inversion Depth
The techniques to be used for deriving the inversion depth and strength for the modeling
domain should be described.
Sources of Other Input Data
The Protocol Document should describe the data and techniques to be used to specify
other input data, such as surface temperature, terrain, and land use and surface characteristics.
QUALITY ASSURANCE AND DIAGNOSTIC ANALYSES
Quality Assurance Tests of Input Components
The specific quality assurance tests to be used on the data input fields should be described
(see Chapter 4).
Diagnostic Tests of Base Case Simulation
The specific diagnostic tests to be used for the base-case meteorological episode
simulations should be described. As discussed in Chapter 4, these should include, at a
minimum, time-series plots, observed and predicted carbon monoxide maps, zero emissions and
zero boundary conditions tests, and tests on the mixing height variations and wind fields.
Additional diagnostic tests are encouraged and should be described in the protocol.
A-4
-------
MODEL PERFORMANCE EVALUATION
Performance Evaluation Tests
The graphical, statistical, and other measures to be used in the model performance
evaluation should be specified. At a minimum, the tests recommended in Chapter 5 should be
included. Additional measures may also be considered and should be described if they are to
be used.
ROADWAY INTERSECTION MODELING
Selection Methodology for Intersection Modeling
A description of the methodology for the selection of roadway intersections for modeling
should be provided.
Modeling Methodology
The procedures for developing traffic inputs to the roadway intersection model to meet
the study objectives should be described. Procedures should also discuss future-year traffic input
methodologies.
ATTAINMENT DEMONSTRATIONS
Identification of Attainment-Year Mandated Control Measures
The protocol document should include a description of the 1990 CAAA control measures
and other measures mandated to be implemented by the attainment year.
A-5
-------
Methodologies for Generating Control Strategy Emission Inventories
The procedures for deriving altemative-control-strategy emission scenarios to meet the
study objectives should be described.
emonst ration
Procedures for using the model simulation results in demonstrating attainment of the
carbon monoxide NAAQS should be included.
SUBMTTTAL PROCEDURES
The documentation and analyses that will be submitted for EPA Regional Office review
should be described. Also, any documentation other than the modeling protocol requiring EPA
regional office approval should be described.
A-6
-------
Appendix B
PERFORMANCE MEASURE FORMULATIONS
-------
APPENDIX B
PERFORMANCE MEASURE FORMULATIONS
RECOMMENDED PERFORMANCE MEASURES
Unpaired Highest-Prediction Accuracy (Au) (1)
where
Ay = unpaired highest-prediction accuracy (quantifies the difference between
the magnitude of the highest1 8-hour observed value and the highest 8-
hour predicted value)
co(-»-) = maximum 8-hour observed concentration over all hours and monitoring
stations
cp(.,.) = maximum 8-hour; ^dieted concentration over all hours and surface grid
squares
1 "Highest" refers to the maximum 8-hour concentration across all hours and monitoring
stations.
B-l
-------
Average Absolute Error in 8-Hour Peak Prediction Accuracy for
Paris >5.0ppm (J) (2)
where
3 -
W ^
A = mean paired peak2 prediction accuracies averaged over all monitoring
stations with observed values > 5.0 ppm
N = number of monitoring stations
jtj) = peak observed value > 5.0 ppm at monitoring station i for the period tz
tj)3 = predicted value at monitoring station i for the period tj
t = starting hour of the peak observed value at monitoring station i
2 "Peak" refers to the maximum 8-hour concentration at a particular monitoring station.
3 Predicted value derived from bilinear interpolation of the predicted values at the four
grid cells nearest to station i for each given hour averaged over the eight hour period.
B-2
-------
Average Absolute Error in the Predicted Time of the 8-Hour
Peak Concentration for Station Pairs >5.0 ppm (Ac) (3)
where
A\ = mean absolute error in the predicted time of the peak4 concentration,
paired by station (for all monitoring stations > 5.0'ppm).
N = number of monitoring stations
t0(i) = peak time of observed value > 5.0 ppm at monitoring station i
tp(i) = predicted peak time of value at monitoring station i
OTHER SUGGESTED PERFORMANCE MEASURES
For details on the formulation of other performance measures, the reader is referred to Appendix
C of the "Guideline for Regulatory Application of the Urban Airshed Model" (EPA, 1991).5
"Peak refers to the maximum 8-hour concentration at a particular monitoring station.
5 U.S. Environmental Protection Agency, 1991. Guideline for Regulatory Application
of the Urban Airshed Model. Office of Air Quality Planning and Standards, Research
Triangle Park, NC.
B-3
-------
APPENDIX C
TECHNICAL DISCUSSION OF UAM MODEL INPUTS
WITH EXAMPLE APPLICATION
-------
TECHNICAL DISCUSSION OF UAM MODEL INPUTS
WITH EXAMPLE APPLICATION
INTRODUCTION
This appendix presents a technical discussion concerning the development of UAM model
inputs. This guidance supplements the general discussion given in the UAM User's Guide
(EPA, 1990). As an aid to understanding these developments, an example application is used
throughout the appendix for the Phoenix metropolitan area for area-wide CO modeling conducted
in'support of a CO attainment demonstration. This work was performed in support of a Federal
Implementation Plan (FIP) for the Phoenix area (Causley et al., 1991). The grid cell size used
in UAM modeling was set to 1 mile to correspond to the layout of the city in square-mile
blocks. The horizontal east-west dimension was 31 miles, and the north-south dimension was
18 miles (Figure C-l). This area covers all of the current and future emission sources. An 18-
hour simulation was performed with a starting time of 1500 on December 8, 1989 and an ending
time of 1000 on the following day.
MODEL INPUTS
Terrain (TERRAIN)
If a spatially varying surface roughness and deposition field is considered to be important,
then a general procedure for determining the surface roughness lengths and the vegetation
(deposition) factors is:
1. Determine the land-use distribution of the modeling domain by consulting USGS
maps. Data on land use are available at a variety of resolutions and levels of
differentiation. Resolution of the source data has ranged from approximately 10
meters for an area where aerial photographs are available to about 1000 meters
when 1:250000 scale USGS maps are used. USGS Land Use and Land Cover
c-i
-------
357
J373- . „ 389. ..- 405, w , .
3735
3719
r 3703
- 3687
30
40
50
3671
FIGURE C-l. Phoenix DWM modeling domain with smaller UAM domain super-
imposed (isopleths of terrain heights in meters).
C-2
-------
series maps provide detailed land use information which can be used to identify
more specific categories (e.g., residential, light industrial, deciduous forest,
coniferous forest, forested wetlands, etc.). These maps are available in both
hardcopy and digitized format; for further information, contact the Earth Science
Information Center.
2. Assign roughness and vegetation factors to a grid cell based on the predominant
land-use category in that cell. The values currently recommended for use are
based on recent work by Argonne National Laboratory and are listed in Table
C-l. The choice of data source will affect the correspondence between its land
use categories and the categories listed in Table C-l. Generally, categories are
assigned so that the roughness and vegetation factors are well represented.
The TERRAIN file can be created for simple cases using the UAM preprocessor
program, TERAIN (EPA, 1990). The SUBREGION packet is used to define regions which
correspond to the major land-use types (such as urban, land, and water), while the
CONSTANTS packet is used to set the values for each of the areas.
More commonly, it is desirable to write a computer program, referring to the UAM
System's manual (EPA, 1990) for the binary output format, to create the TERRAIN file based
on the specific land-use data that the user has assembled. The FORTRAN program "CRETER"
(EPA, 1990) is one such program which was developed to read a gridded set of land use codes
and create a TERRAIN file based on the data shown in Table C-l.
For Phoenix, default surface roughness and deposition factors were used corresponding
to the average factors appropriate for the terrain characteristics of Phoenix.
Surface Temperatures (TEMPERATUR)
Specification of the 2-dimensional surface temperature fields is relatively straightforward
given the typical number of temperature monitoring stations usually available in an urban area.
The UAM TEMPERATUR preprocessor is used to generate the temperature fields from surface
C-3
-------
TABLE C-l. Surface roughness and deposition factors used by the
program CRETER. Information is based on studies by Argonne
National Laboratory.
Category
Number
1
2
3
4
5
6
7
8
9
10
' 11
Land-Use Category
Urban
Agricultural
Range
Deciduous forest
Coniferous forest
including wetland
Mixed Forest
Water
Barren land
Nonforest wetlands
Mixed agricultural and
range
Rocky (low shrubs)
Surface
Roughness
(meters)
3.00
0.25
0.05
1.00
1.00
1.00
0.0001
0.002
0.15
0.10
0.10
Deposition
Factor
0.2
0.5
0.4
0.4
0.3
0.3
0.03
0.2
0.3
0.5
0.3
C-4
-------
measurements. In general, the STAHNTERP option is used to interpolate the observations
between grid cells. The user should insure that proper minimum and maximum temperatures
are used so that data entered in error do not result in unreasonable temperature values.
For Phoenix, gridded temperature fields were derived from observed data collected at the
NWS and other local air quality monitoring network sites. This data was then interpolated using
the procedure discussed above.
Meteorological Scalars (METSCALARS)
Meteorological scalars required by the UAM include atmospheric pressure
(ATMOSPRESS), concentration of water vapor (CONCWATER), the NO2 photolysis rate
(RADFACTOR), the exposure class (EXPCLASS), and the vertical temperature gradients above
and below the height of the DIFFBREAK (TGRADABOVE and TGRADBELOW)
(DIFFBREAK values will be discussed later). Of these scalars, only ATMOSPRESS,
EXPCLASS, TGRADABOVE, and TGRADBELOW are relevant to UAM applications for CO.
The remaining parameters should be set to "dummy" values.
1. ATMOSPRESS is required for the UAM to compute mass emissions into volume
(ppm) units. It should be specified (in atmospheres) based on local pressure
observations; if such observations are unavailable, a default value of 1.0
atmospheres should be prescribed.
2. The exposure class classification (EXPCLASS) is used in computing UAM
vertical mixing coefficients. Assignment of EXPCLASS for a given one-hour
time interval requires calculation of the solar zenith angle and an estimate of
cloud cover. A FORTRAN program called "SUNFUNC" (Schere and
Demerjian, 1977) yields hourly values of solar zenith angle for a specific date and
location. Cloud cover should be estimated from hourly observations at a nearby
National Weather Service surface site, if possible.
C-5
-------
Table C-2 gives exposure class as a function of solar zenith angle and cloud
cover. Note that solar zenith angles must be specified for the end of each one-
hour time interval, while cloud cover estimates apply for the entire span of the
time interval.
3. The temperature gradients above and below the height of the OUTBREAK
(TGRADABOVE and TGRADBELOW) are used to characterize the vertical
temperature structure of the atmosphere in the modeling region. TGRADABOVE
and TGRADBELOW are used along with EXPCLASS and wind speeds to
compute vertical mixing coefficients. In addition, the point source emissions
preprocessor may use the temperature gradients in determining the effective stack
height of point source emissions.
It is frequently difficult to compute the vertical temperature gradients from available
temperature data. Vertical mixing between UAM layers is very sensitive to the sign of these
gradients, but considerably less sensitive to variations in their absolute value. It is recommended
that for each hour the following domain-wide parameters be estimated from available surface and
upper-air temperature data:
A characteristic DEFFBREAK value (DB)
A characteristic temperature at REGIONTOP (Ttop)
A characteristic temperature at DB (TDB)
A characteristic surface temperature
The vertical temperature gradients can thus be computed:
TGRADBELOW = (TDB - T^) / DB
TGRADABOVE = (T^ - TDB) / (REGIONTOP - DB)
C-6
-------
TABLE C-2. Exposure class (CE) classification
based on cloud cover and solar zenith angle.
Solar Zenith Angle Cloud Cover
(degrees) (percent) CE
> 85 85 > 50 -1
£30 £ 50 3
£ 30 > 50 2
30 < 9 ^ 55 £ 50 2
30 < 9 <; 55 > 50 1
55 < 9 <. 85 <; 50 1
55 < 9 ^ 85 > 50 0
C-7
-------
If temperature data are insufficient, it is recommended that following default values be used:
DAY NIGHT
TGRADBELOW -0.01 K/m +0.1 K/m
TGRADABOVE 0.0 K/m 0.0 K/m
Table C-3 illustrates the values used as input for the Phoenix application. The
temperature gradients were based on data collected during a previous CO study (Haney and
Ireson, 1988). All other values were based on data collected from the Phoenix NWS station for
the 8-9 December 1989 episode.
Inversion Depth (OUTBREAK)
The DIFFBREAK is the height at which the model will switch from using the
TGRADBELOW value as the temperature lapse rate to using the TGRADABOVE value (see
METSCALARS). This affects the calculation of the rate of turbulent diffusion and therefore the
rate of vertical diffusion of material across the boundaries between cells.
For CO applications the most important consideration in setting DIFFBREAK is minimization
of "artificial" vertical mixing, especially under conditions when actual vertical mixing is very
limited. Note that the model surface CO emissions mix instantaneously throughout the depth
of the lowest UAM layer. In the Phoenix application where only two vertical layers were used,
the DIFFBREAK was assigned as specified in Table C-4. It is recommended that for two
vertical layers these domain-wide daytime and nighttime values be used as a "first guess" at
DIFFBREAK. The nighttime value can be considered representative of mechanical mixed layer
depth.
The periods of maximum CO emissions during winter episodes coincide with the morning
and evening transition periods. The use of tethersondes to obtain hourly temperature and wind
profiles in the lowest 100-200 m at one or more sites during transition periods should be
C-8
-------
TABLE C-3. METSCALARS inputs for the 8-9 December 1989 base case simulation.
Hours (PST)
1500-1600
1600-1700
1700-1800
1800-1900
1900-2000
2000-2100
2100-2200
2200-2300
2300-2400
0000-0100
0100-0200
0200-0300
0300-0400
0400-0500
0500-0600
0600-0700
0700-0800
0800-0900
0900-1000
T-GRAD
OC/m)
Above Below
-.05
-.04
.0-
.01
.01
.01
.01
.01
.01
.01
.01
.01
.01
.01
.01
.01
.01
.01
.01
-.03
-.02
.029
.10
.12
.133
.128
.142
.17
.175
.18
.185
.19
.19
.19
.19
.19
.19
.19
Exposure
Class
1
1
1
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
-2
_2
-2
1
1
Radiation
Factor
.234
.118
.013
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.00
.042
.154
Water
Concentration
(ppm)
12687
12687
12687
12687
12687
12687
12687
12687
12687
12687
12687
12687
12687
12687
12687
12687
12687
12687
12687
C-9
-------
TABLE C-4. Height of the nighttime surface-based mechanical
mixed layer (DIFFBREAK) specified for the base case
UAM simulation of 8-9 December 1989.
Hour Height (m)
1500-1600 60
1600-1700 50
1700-1800 40
1800-1900 30
1900-0700 20
0700-0800 30
0800-0900 40
0900-1000 50
C-10
-------
considered if more refined estimates of inversion mixing depth are desired. Very strong and
very shallow ground-based inversions have been observed using tethersondes.
Winds (WINDS)
The Urban Airshed Model (UAM) requires hourly definition of the horizontal wind
vector at the center of each grid cell to allow airflow to be continuously represented throughout
the model domain. UAM wind inputs must be defined through a combination of measurements,
numerical calculations, and assumptions based on an understanding of the physical processes
generating the airflow. The optimal procedure to be used for wind definition depends on (1) the
significance and distribution of topographic features within and adjacent to the UAM modeling
domain; and (2) the availability and representativeness of wind measurements within the domain.
"Noncomplicated" vs. "Complicated" Airflow Patterns
Two types of airflow patterns, "noncomplicated" and "complicated," are discussed in this
subsection. The most restrictive definition of a noncomplicated airflow would be one whose
behavior in space and time is satisfactorily represented by continuous measurements at one
location within a modeling domain. In the remainder of this discussion, airflow patterns that
do not meet this criterion are deemed to be complicated.
Noncomplicated airflows are usually represented in the UAM by horizontally uniform
hourly gridded wind fields at each model level. Note that these winds are expected to vary
vertically and temporally.
Complicated airflows must be represented in the UAM by hourly gridded wind fields that
vary horizontally, vertically, and temporally. These fields are obtained through analysis of
observational data and/or through numerical modeling techniques.
Topographic Features Associated with Complicated Airflow Pattern
Topographic contour maps covering the area of potential UAM application give some
insight into the degree of complication to be expected in local airflow patterns. We expect a
complicated airflow to occur if the map indicates the following features:
C-ll
-------
A coastline
A lake or body of water contained within the modeling domain whose area
exceeds about 25 km2
A concentrated urban area surrounded by a rural, relatively
unpopulated area.
Terrain features of a horizontal scale greater than 5 km and a vertical
scale greater than 100 m, including mountains or hills, ridges, and
valleys.
Local air circulations are generated when the temperature of adjacent land and water
surfaces differs significantly. Differences in surface temperature are communicated to the
overlying air, resulting in local pressure gradients. Sea and lake breezes (directed onshore)
occur when land temperature exceeds water temperature; land breezes (directed offshore) occur
when water is warmer than adjacent land. Curvature of the coastline further complicates
sea/lake/land breeze patterns.
Urban-rural temperature contrasts may generate local airflow patterns similar to
sea/land breezes, in which the air flows from a relatively cool rural area toward a relatively
warm urban area. Urban "heat island" effects are expected to be strongest during the night and
early morning hours under relatively stagnant synoptic-scale conditions.
Terrain features affect the overlying airflow in several ways. Gravity wave
disturbances—vertically propagating waves and trapped lee waves-can occur when air is forced
over a mountain ridge. Diurnal heating and cooling of sloping ground surfaces generates
upslope- and downslope-directed flows, respectively. Slope flow development along the sides
of a valley can generate airflow over the valley which parallels the valley axis. Topographic
barriers may act to deflect and channel impinging airflow.
C-12
-------
Additionally, local airflow may be complicated by the presence of a "front" that
separates two different large-scale air masses. National Weather Service surface weather maps
should indicate the location of fronts.
For CO applications over relatively small modeling domains, it should be stressed
that the airflow within the domain may be complicated by geographic features outside the
domain.
Wind Field Development m "Noncomplicated'' Situations
In a noncomplicated situation, we assume uniformity of surface winds throughout
the domain at a given time. Ideally, hourly surface observations at a single location would be
sufficient to define these winds. However, routine observational procedures are frequently
inconsistent with the requirements for UAM representation of pollutant transport. Thus,
available surface wind observations may need to be utilized with great caution.
"Instantaneous" vs. "hour-averaged" winds. Almost all potential modeling domains contain at
least one National Weather Service or Federal Aviation Administration surface meteorological
station. These stations report hourly wind observations (direction and speed) at a standard
anemometer/vane height of 10 m AGL. The observations are meant to be instantaneous: hi
practice, an "instantaneous" wind observation represents a subjective average over a period of
approximately one minute. The UAM requires hourly vector-averaged winds. When wind
speeds are low ( < 2-3 m s ) instantaneously observed winds may differ significantly from
hour-averaged winds. In general, vector-averaging of winds which vary in direction tends to
produce averaged speeds which are lower than instantaneously observed speeds. This difference
in speed may have a significant effect on UAM CO predictions under the typical stagnant
conditions.
No clear prescription is available for deducing hour-averaged winds from
instantaneous observations. If stations reporting hour-averaged winds are sufficiently
representative to describe surface airflow within the modeling domain, it is recommended that
instantaneous wind measurements not be used in generating UAM wind inputs. If it is necessary
C-13
-------
to rely on instantaneous wind measurements to generate the wind field, it is recommended that
a UAM sensitivity experiment be conducted in which wind speeds are reduced by 50 percent.
Non-standard wind observations. Surface wind observations may be available from sources
other than the NWS and FAA. Use of data from these sources requires consideration of the
height of monitors, exposure of monitors, and averaging time of observations.
If data from more than one monitoring site are available, the data should be
examined to ascertain whether observed winds differ systematically among the sites. If
systematic differences appear ascribable to physical processes (rather than to monitor
height/exposure or to averaging time), the flow pattern must be considered to be complicated.
If the differences among site observations at specific times appear to be random, one might
consider averaging (rather than spatially interpolating) the observations to obtain a single
representative domain-wide wind at a given time.
Minimum number of upper-air wind observations. In a noncomplicated situation we can assume
horizontal uniformity of upper-air winds. However, upper-air winds can vary with height and
time. For CO applications accurate representation of upper-air winds may be considerably less
important than representation of surface winds, as most CO emissions are confined to the
surface. For a 2-layer UAM domain with a REGIONTOP of 200 m, it is probably sufficient
in noncomplicated situations to assume that winds are uniform with height.
Wind Field Development in Complicated Situations
The diagnostic wind model (DWM) (EPA, 1990) is used to develop the wind input
file for the UAM. The DWM is designed to generate gridded fields of the horizontal wind
components u and v for several user-specified vertical layers at a given time. The DWM is
designed to use local surface and upper-air observations where available, while providing a
limited representation of terrain-generated airflow features in subregions where local
observations are insufficient to account for terrain effects.
C-14
-------
A "first-guess" (u,v) field is generated by adjusting a domain-mean wind for terrain
kinematic effects, thermodynamically generated slope flows, and blocking effects.
The first-guess field is then combined with observations via an objective analysis scheme to
produce a new gridded (u,v) field. The first-guess (u,v) values are weighted relatively heavily
in- subregions within which observations are either unrepresentative or unavailable; weighting
of the first-guess field relative to observations is controlled by the DWM input parameter R,
which has units of distance and represents an effective radius of influence.
The DWM is highly "data-driven." Several important features of mesoscale airflow cannot be
represented by the model in the absence of representative observations. These features include
sea/land breezes, valley-axis flows, and low-level jets.
In some regions of complex airflow, routine NWS/FAA wind observations may be
insufficient to represent the horizontal, vertical, and temporal variability of the airflow patterns.
Intensive field measurement programs may be necessary to obtain sufficient surface and upper-
air wind data for the DWM to generate representative UAM gridded wind fields.
When exercising the DWM for CO applications, the DWM domain should
encompass terrain features outside the UAM domain which may affect the DWM representation
of the airflow within the UAM domain. Figure C-l illustrates the DWM and UAM modeling
domains for the Phoenix CO UAM application.
Figure C-2 shows the observed surface winds within the Phoenix UAM domain at
2100 MST on 8 December 1989, the time at which the maximum CO concentration was
observed during the Phoenix episode. Figure C-3 shows the DWM surface gridded wind field
derived from these observations. Note that observed wind speeds during CO episodes are
frequently very light (< 1 m s"1). Under such conditions the variability of wind direction on
small scales may be much greater than implied by a DWM field generated from a limited set of
observations; thus DWM wind flow should be interpreted with caution.
C-15
-------
w
cue
«fcl
tfcr
pfl.
J7M
(KM)
SOUTH
WIND SPEED (M/S)
FIGURE C-2 . 0&Mrv«d Surfoc* Winds - Uoocopo County. Pho«nix
2100 - 2200 LST 8 DEC 88
(KM)
IP IS
WIND SPEED (M/S)
FIGURE 03.
OWU WINDS AT UEVB. 1 (10U)
2100 UST 80EC89
C-16
-------
Use of "Pseudo-Data"
A meteorologist/modeler may wish to draw upon his knowledge and experience to
improve a DWM-generated wind field, especially when he believes that available observations
coupled with the DWM complex-terrain parameterizations are inadequate for realistic definition
of the airflow pattern. An efficient means of modifying a wind field is to input to the DWM
pseudo-data, i.e., fabricated observations at specific locations and times.
Caution is advised with the use of pseudo-data as it introduces a significant degree
of subjectivity into a modeling procedure which ideally would be objective. The development
of the DWM was motivated in part to reduce the need for pseudo-data in regions of complex
terrain. However, there may be situations when subjective fabrication of artificial data is clearly
more desirable than objective extrapolation of real data into a subregion of space and time where
it is unrepresentative. The use of any pseudo-data should be discussed in the modeling report.
We emphasize that the meteorologist/modeler must be able to justify the fabrication
of pseudo-data based on an understanding of meteorological processes. Results of previous
intensive monitoring within the domain of interest, if available, may be exploited to improve
wind field representation for an episode under current study, if comparative meteorological
analyses can demonstrate that the previous day is an analogue of the current day.
Air Quality
The UAM preprocessor has available an interpolation scheme that performs an
inverse distance-weighted interpolation. If there are only a few monitors, most of the region
will probably be a considerable distance from a monitor. In these areas, the interpolation
scheme will tend to average all of the station values. This is probably not realistic since the
monitors are typically located in urban areas and much of the region may be rural. It is
therefore often necessary to define pseudo-stations scattered through the rural portions of the
region which should be assigned some background value. The background values can be
determined from an existing rural station, if available. Otherwise, a set of default values can
be selected from Table C-5. This table (Killus et al., 1982) gives three different levels of
pollutant concentrations which can be selected based on how pristine the surroundings are
C-17
-------
TABLE. C-5. Recommended background
concentration for carbon monoxide
(Killus et al., 1982).
Value
Low
Mid
High
Area
Classification
Rural
Suburban
Urban
Concentration
(ppm)
0.1
0.2
0.5
C-18
-------
considered to be. Certain parameters can be used to adjust the interpolation method. Usually,
it is best to begin with a small radius of influence (a grid cell or so) with a large increment (10
to 25 grid cells) to be used if no valid stations are found. This method will ensure that the
station value is used in the immediate vicinity of the station, but that the value will quickly begin
to approach the background value as the distance from the station increases. Entire subregions
can be made to be dependent on one station.
The vertical structure of the concentrations should be set using a vertical relative
profile method. This method can be used to set the cell below the DIFFBREAK to the surface
value and the cell above the DIFFBREAK to the top of the modeling domain value. The top
of the modeling domain value will normally be the clean background value. The VERTICAL
PROFILE packet for this method should include four heights: (1) a height of zero with a factor
of zero; (2) a height equal to just less than the DIFFBREAK at the location of the VERTICAL
PROFILE with a factor of zero; (3) a height equal to the DIFFBREAK at the location of the
VERTICAL PROFILE with a factor of one; and (4) a height equal to the REGIONTOP at the
location of the VERTICAL PROFILE with a factor of one.
Since there is considerable uncertainty involved in the above estimation of the initial
conditions, the concentration fields should be checked with the observed data for any unrealistic
values. With only a few stations, a single inappropriate value can easily lead to an unrealistic
initial field. As a further check on the calculated values, the entire 3-dimensional field can be
integrated spatially to derive the total mass of carbon monoxide in the region. The total mass
should be smaller than one day's emissions of carbon monoxide. (This mass calculation is
performed by the UAM and printed in the output file.)
Because of the uncertainties involved, the usual practice is to reduce the influence
of the initial conditions as much as possible by starting the simulation prior to the buildup of
high carbon monoxide concentrations.
For the Phoenix modeling domain, the initial concentration field was based on
carbon monoxide data collected at monitors in the modeling domain. The values corresponding
to the specific initial hour of the simulation are shown in Table C-6.
G-19
-------
TABLE C-6. Initial surface CO conditions (ppm) for the AIRQUALITY file for the hour
1500-1600 for 8-9 December 1989, Phoenix.
Station
X-UTM
(km)
Y-UTM
(km)
Concentration
(ppm)
South Phoenix 400.1 3696.7 0.6
South Scottsdale 414.0 3704.6 0.4
North Scottsdale 415.9 3719.5 0.8
North Phoenix 400.9 3714.4 0.0
Glendale 389.3 3714.6 0.7
West Indian School Road ' 394.8 3706.2 1.6
West Phoenix 393.2 3705.9 0.5
Central Phoenix 403.2 3702.5 1.3
2039 W. Lewis Avenue 397.0 3704.3 0.5
020
-------
TOPCONC
The top of the model concentration (TOPCONC) file defines concentrations of each
species at the height of the top of the modeling region for each grid cell. Only vertical winds
(either due to divergence or due to motion of the top boundary) can transport pollutants from
the top boundary into the modeled region. Material cannot diffuse into the region from the top
boundary. Since CO applications use a fixed REGIONTOP height, the only mechanism for top
boundary concentrations to enter the region is vertical winds due to divergence in the horizontal
wind field. Since a divergence minimization technique is usually applied to the wind fields to
be input to the UAM, the vertical winds are expected to be small and hence the importance of
the top boundary concentrations should be relatively small. However, the use of the TOPCONC
file is recommended in the preparation of both the AIRQUALTTY (initial conditions) file and
the BOUNDARY (vertical profile for the horizontal boundary) file and therefore realistic top
boundary concentrations are needed.
The recommended method for setting the concentrations in cells above the
DIFFBREAK, for either AIRQUALITY or BOUNDARY, is to set them equal to the
concentration at the top of the region.
The usual procedure is to use the observed top concentrations in cells above the
DIFFBREAK; however, when these values are not known, the concentration should be set to
the values recommended in Table C-5.
For the Phoenix study, measurements of the vertical distribution of carbon
monoxide were conducted during a number of evenings at a site in central Phoenix. These
observed values formed the basis for using 1.0 ppm of carbon monoxide for all hours of the
simulation. While this value is twice the default background value of 0.5 ppm for urban areas,
it is based on observed CO vertical distribution values for Phoenix.
BOUNDARY
The recommended starting point for mixed layer boundary concentrations is the
mid-level concentration from Table C-5. If a comparison of model results with observed carbon
C-21
-------
monoxide concentrations indicates lower predicted values in an area removed from emissions
sources, concentrations along an appropriate inflow boundary can be increased. There should,
however, be some corroborating evidence in the proximity of another urban area or a shift in
the wind to cause recirculation of the pollutants to justify this change. (Examination of the wind
field through forward particle-paths will show if this recirculation is present.) Ideally, in the
latter case, the size of the region would be increased so that material is retained within the
region rather than being lost through the boundary. This may not be practical, though, due to
the increased costs associated with the simulation. Whenever the boundary concentrations are
increased, a calculation should be made to determine the effective amount of carbon monoxide
added to the region. That is, convert the concentration to kg/m3 (or other convenient unit), and
multiply by the wind component perpendicular to the boundary and the area of the affected
boundary (i.e., height of the affected layer times the cell width times the number of boundary
cells affected) to get the effective mass flux per unit tune. Compare this to the total emission
rate within the modeled region and the total emission rate in upwind areas to evaluate the realism
of the selected concentrations. When the UAM is rerun with the revised boundary
concentrations, the desired effect on the results may or may not be realized. If it is not, a return
to the original boundary conditions will be necessary and the change reevaluated.
In some cases, monitoring sites will be located near enough to region boundaries
so that the values measured there are characteristic of boundary concentrations nearby. The user
should be cautious of setting boundary concentrations higher than 1.0 ppm based on a limited
set of stations. A check of the mass-flux through the boundary is in order whenever setting the
concentrations to an elevated value: The user must, however, always be cautious of using
station values (particularly "hot-spot" monitors) that may be affected by local emission sources
to set any major length of the boundary. High concentrations should be limited to a grid cell
or two to either side of the station. The user should also remember that the larger the grid cell
is, the higher the mass-flux per cell is.
The recommended method for allocating pollutants vertically along the boundaries
is the same as for the initial conditions. The RELPROFRAT vertical method should be used
with the factors set such that the surface value is used up to the DIFFBREAK and the
TOPCONC value is used above the DIFFBREAK (see section on ArRQUALITY). If constant
C-22
-------
boundary conditions are used all around, the CONSTANT vertical method can be used to
simplify the preparation of the boundary inputs. To use any elevated concentrations above the
DEFFBREAK, the user must have very strong evidence in the form of upper-air observations to
justify the values used.
For the Phoenix analysis forward particle paths starting at 1600, 1800, and 2000
MST were performed to determine if recirculation of air took place within the boundaries of the
model domain. An example forward particle trajectory (Figure C-4) shows that some
recirculation of air took place; later particle paths showed the air mass eventually being advected
southwest of central Phoenix. Therefore, the domain was concluded to be large enough to
encompass the recirculation of carbon monoxide. Boundary conditions were set somewhat high
to 1.0 ppm for all boundaries both above and below the DIFFBREAK based on limited air
quality monitoring stations located near and outside the boundaries. The sensitivity simulation
(zero boundary conditions) results showed little change from the base case concentrations,
indicating that the boundary conditions are of little importance in producing elevated carbon
monoxide concentrations, especially near the center of the city.
Mobile Emissions
Preparation of Vehicle Emission Factors with MQBILE4.1
The emission factors used to estimate carbon monoxide emissions from on-road
motor vehicles are provided by EPA's MOBILE emission factor model. As of the dated
preparation of this document, the current version of this model is MOBILE4.1 (EPA, 1991).
Mobile emission factors vary greatly with several parameters, including vehicle speed, fuel
oxygen content, vehicle fleet characteristics, ambient temperature, and inspection/maintenance
program characteristics. Many of these parameters are required inputs to the MOBILE model,
as listed in Table C-7.
In addition to these required inputs, a variety of optional inputs may be used with
MOBILE as shown in Table C-8. These allow the user to replace default national average data
on fleet characteristics with data that may be more suitable to the region being modeled. When
C-23
-------
378
394
411
427
- 3708
3692
FIGURE C-4. Surface air parcel trajectories (forward) for Phoenix,
8 December 1989: released at 1600 1ST.
C-24
-------
TABLE C-7. Required input parameters for EPA's MOBILE models.
Calendar year
Ambient daily or hourly temperature
Minimum and maximum daily temperature
Base RVP
In-use RVP*
In-use RVP start year*
Region altitude
Speeds
Operating modes
Not always used by MOBILE, but input record is required.
C-25
-------
TABLE C-8. Optional input parameters for EPA's MOBILE models.
Alternate basic emissions rates'"
Alternate vehicle tampering rates*
Fleet Characterization Data:
Fleet registration distribution**
Fleet mileage accumulation**
Diesel penetration rate**
Vehicle class distribution**
Inspection & Maintenance Programs:
start year
model
compliance rate
frequency of inspection
test type
special mechanic training
Anti-Tampering Programs:
start year
vehicle classes inspected
frequency of inspection
list of equipment inspected
Refueling Programs (Stage II):
start year
LDGV % system efficiency
Oxygenated Fuel Use:
Average fuel oxygen content
Market penetration
RVP allowance
stringency %
waiver rates
program type
vehicle classes inspected
alternate credits
model years inspected
program type
compliance rate
alternate credits
phase-in period
HDGV % system
efficiency
C-26
-------
region-specific data are available, it is recommended that these data be used in place of the
default data in the model. The MOBILE User's Guide provides further detail on optional inputs
to the model.
Two of the required inputs to the MOBILE model, speed and temperature, vary too
much across a region to be properly represented by a single value. Therefore, regardless of
whether county-wide VMT estimates are being used, or a combination of VMT and speeds
obtained from transportation modeling, a series of MOBILE runs is required in order to obtain
emission factors representative of the range of speeds and temperatures which occur in a region.
A third input to the model, operating mode fraction, varies by time of day and vehicle class.
The following sections discuss methods for properly representing these parameters in a regional
mobile source inventory.
Treatment of Speed Variation
Average speeds differ according to roadway type, such as freeways and residential
streets. The average speeds will vary by time of day since roadway congestion (such as is
experienced during peak a.m. and p.m. commute hours) may result in significantly reduced
speeds. The simplest way to represent average speeds in a region is to assign regional VMT
estimates to some type of roadway classification scheme. Table C-9 gives one such classification
scheme, used in the NAPAP inventories, and the corresponding Federal Highway Administration
(FHWA) road classes. The FHWA roadway types shown in Table C-9 are actually a subset of
the entire FHWA classification system since the FHWA also classifies roads by area type (TRB,
1985). For each roadway type represented in the chosen classification scheme, an average speed
is assigned. Vehicle emission factors must be developed for the average speed represented by
each roadway type.
It is recommended that a further refinement be introduced through splitting the day
into peak and off-peak periods, and assigning speeds representative of congested and uncongested
roadway conditions. Estimates would then need to be given for the amount of VMT occurring
during each of these periods. Allotment of VMT by time of day is discussed later in this
section.
C-27
-------
TABLE C-9. Roadway categories in emission inventories.
NAPAP Road Type
Corresponding FHWA Road Types
Limited Access
Rural
Urban
Rural and Urban Interstate
Rural and Urban Other Principal
Arterials
Other Principal
Other Freeways and Expressways
Rural and Urban Minor Arterials
Rural Major Collector
Rural Minor Collector
Rural Local
Urban Major Collector
Urban Minor Collector
Urban Local
C-28
-------
If outputs from a validated transportation model are available for the region to be
modeled, the speed associated with each section of roadway, or link, explicitly modeled will be
provided by the model. In addition, average speeds associated with intrazonal travel, i.e., travel
which is not assigned to a specific link but instead is assumed to occur within a specific TAZ,
can be provided by the transportation model. From this information, an emissions modeler must
determine the range of speeds represented in the region to be modeled, and then develop
emission factors which represent speed conditions throughout this range. The speed increments
used in modeling should be from 2 to 5 mph. In Phoenix, the speeds used ranged from 4 to 56
mph in increments of 4 mph (Causley et al., 1991).
When some mobile emission models (e.g., TRFCONV) are used for the mobile
source CO inventory, speed curve functions are required by the model to generate the data
necessary, to obtain vehicle speed as a function of the volume-to-capacity ratio for a given link.
Either default curves from the Bureau of Public Roads (BPR) or locale-specific speed curves can
be used. In the Phoenix CO inventories, locale-specific curves prepared by the Arizona
Department of Environmental Quality for the Maricopa Association of Governments (MAG)
were used. A comparison of the BPR and MAG speed curves is shown in Figure C-5 (Causley
et al., 1991).
Temperature Variation and Diurnal Activity
The representation of daily temperature range in CO emission inventories depends
upon the number of time periods one is explicitly modeling. MOBILE has the capability to take
user-provided information on daily minimum and maximum temperature, which should be
representative of the actual CO episode being modeled, and provide an emission factor which
represents the daily average. This can then be combiiu.d with daily VMT totals for a region to
estimate total emissions, which are then allocated over each hour of the day based upon assumed
patterns of diurnal vehicle activity.
If possible, diurnal vehicle activity should be based upon regional traffic survey
information. If the episode to be modeled includes a weekend, caution should be exercised in
using data from such surveys since they are often only representative of weekday vehicle
activity. Weekend vehicle activity exhibits smaller a.m. and p.m. peaks. Lacking specific
C-29
-------
co
•o
u
o
u
V
a:
a.
vw
o
«8
V
3
BO
>» (3 m
*- 5? 2
u
CD
O.
a
u
4)
3
rH
O
c.
ca
V '
at
0)
00
flj
Id
01
vn
w
C-30
-------
survey information on weekend travel patterns, engineering judgment must be relied upon to
determine diurnal variation in vehicle activity on weekends.
Another, more realistic, way to treat temperature effects is to run MOBILE for
specific periods of the day, or even for each hour, and obtain emission factors representative of
the temperatures of these time periods. In this case, VMT is assigned to specific time periods
before being combined with the vehicle emission factors calculated for the temperatures
associated with that time period to produce total emissions. Some software, specifically the
model TRFCONV (Causley and Duvall, 1992), has been developed which uses link-based VMT
estimates and regional average hourly temperatures to arrive at hourly, gridded emission
estimates. This model requires input of a matrix of carbon monoxide emission factors by
temperature, where the temperatures represented should cover the range experienced over the
region during the day. In the Phoenix work, temperatures used ranged from 30 to 90°F in
increments of 10°F. Figure C-6 shows the hourly mobile-source CO emissions for Phoenix
(Causley, 1991).
Engine Thermal States
Vehicle exhaust emissions can be classified as occurring during three different
engine thermal states: The cold start state, representative of vehicles which have just been
started after completely cooling off (the time required for an engine to completely cool depends
upon whether the vehicle is catalyst or noncatalyst); the hot start state, representative of vehicles
which have been restarted before the engines had completely cooled; and hot stabilized,
representative of the exhaust emissions of a fully warmed up vehicle. The distribution of VMT
accumulated by vehicles among the three engine thermal states varies by time of day and also
by road type. For example, there are more cold starts in the morning, and vehicles are more
likely to be in the hot stabilized mode on limited access roadways. The distribution of VMT by
engine thermal state varies by vehicle class, although such differences are not significant for CO
modeling.
One may input the fraction of VMT accumulated in the cold and hot start modes
into MOBILE, which subsequently corrects the exhaust CO emission factor. One should take
advantage of this feature when developing emission factors specific to roadway classes or time
C-31
-------
o
o
CM
to
O)
O)
I *
•H
T 5
T o
e-
w>
CM -2
O) o,
O) c
T- o
1 ^
i a
I o
•» -^
T B
I «
O
u
O
o>
01
u
u
3
O
03
t!
CO
O)
I
u
u
C-32
-------
of day. Further information on the variation of engine thermal states distributions can be
obtained from United States Department of Transportation (DOT) publications (DOT, 1978).
Additional Features
When using MOBILE to develop a regional inventory, some features of a region
may require more innovative use of the model. It is recommended that these features be
accounted for in developing the emission factors. These features are discussed in the following
sections.
Inspection and Maintenance Programs
Some regions have a significant amount of total VMT accumulated by vehicles not
subject to the local inspection and maintenance (I/M) program. In the Phoenix work, 12 percent
of the vehicles were assumed to be "non I/M" (not subject to the Phoenix I/M and anti-
tampering programs). This value was based upon a survey of out-of-state license plates in the
Phoenix region. Non-I/M vehicles were modeled separately from the I/M vehicles, and resulting
emission factors weighted together to provide composite I/M and non I/M regional emission
rates for the TRFCONV model (Causley et al., 1991).
Oxygenated Fuel Use
It is sometimes more difficult to develop emission factors for regions which are
subject to oxygenated fuels programs since a mixture of fuel types may be in use in the region.
MOBILE will develop emission factors for vehicles using oxygenated fuel, and will factor in the
assumed market penetration of the oxygenated fuel. Thus it is possible to obtain from the model
emission factors which reflect the combined use of oxygenated and conventional gasolines in the
fleet. More information on this feature, which is new in MOBILE, is given in the User's Guide
to MOBILE (EPA, 1991).
MODEL PERFORMANCE AND ATTAINMENT DEMONSTRATION
To illustrate the application of the UAM, we briefly discuss in this section the
model performance and attainment demonstration results for the Phoenix modeling study.
C-33
-------
Figures C-7 and C-8 are examples of time-series and ground-level isopleths used in evaluating
model performance. The isopleths show the observed hour concentration value at each
monitoring station along with the UAM predicted concentrations; they indicate the UAM's
ability to replicate both the temporal profile and magnitudes of the observed concentrations and
the spatial concentration gradients.
Statistical performance results show that the unpaired highest 8-hour prediction had
an overprediction of 30.5 percent. The average station 8-hour peak prediction accuracy for
paired (time and space) values greater than 5.0 ppm showed an overprediction of 19.4 percent.
The final recommended statistical measure showed the average absolute error in the predicted
time of the 8-hour peak concentration, paired by station, to be within 2 hours of the maximum
8-hour observed value. All of these statistical parameters are within the performance ranges
discussed in Chapter 5.
An important task in the attainment demonstration is in developing future-year
vehicle emission factors for CO. The MOBILE model should be adequate for most uses since
it has incorporated CAAA regulations for vehicle CO emission standards. However, care should
be taken to ensure that the regulations in force in future years in the region modeled have
actually been incorporated into the version of the MOBILE model used.
Transportation modeling results are not always available for the years required for
emission inventory estimates. For example, the Phoenix CO modeling work required a 1991
inventory, but transportation model results were available only for 1990 and 1995. In this case,
the options were either to grow the 1990 data or to backcast from the 1995 data. Growing the
1990 data seemed to provide the more conservative emissions estimate since it created more
congestion (traffic volume) on the existing network than backcasting from the 1995 data. In the
Phoenix modeling, growth in VMT was estimated through linear interpolation between total
VMT in 1990 and 1995 (SAI, 1991).
Future-year emission inventories were developed for 1991, 1992, 1995, and 2001.
The TRFCONV program was the primary tool used in creating the future-year inventory. The
Link attribute value (LAV) data and link coordinate data for that year were substituted for the
base year data, and emission factors were changed to reflect fleet changes for each year.
C-34
-------
^P W V€ **
«NNNN
-
'" O
•*n(Vi-'OO»r»eifl-*cocM-«o
.N N CM CM N
affl^oio^nN — O-
N
i F i i r IT i i i i i i i i i
i i i i
B -
B
•a
2
2
Cd
I I J t J I II I I I
r* I I
E
P
oo
(Hdd) 03
-------
Time : 20OO - 2100 UST
378
NORTH
.598
Moximum Volue • 17.58 ppm
Minimum Volu« « 0.83 ppm
418
10
NSCT I
I I I I t < I t
tit I' I
3712
3692
10
20
SOUTH
FIGURE C-8. UAM-calculated CO concentration isopleths (ppm) for
8 December 1989 (2000-2100 LSI).
C-36
-------
Five different control strategies were selected for UAM simulations for each year
(1991, 1992, 1995, and 2001). The five control strategies consisted of an oxygenated fuels
program(27o2), a Reid vapor pressure (RVP) reduction program (RVP10), a combined
oxygenated fuels and RVP program (OXYRVP), a voluntary no-drive day program (VNDD),
and a trip reduction ordinance (TRO).
These UAM control strategy results were then combined with roadway intersection
modeling results. Table C-10 presents results from one of these roadway intersections. The
values in the table are 8-hour running averages. In this application, the roadway intersection
model used was CALQ3HC (EPA, 1992). Note that the combined 8-hour UAM and CALQ3HC
concentrations do not necessarily coincide with the maximum 8-hour UAM plus CALQ3HC
8-hour concentrations because of temporal differences in the time of each model's predicted
maximum CO concentration. These results show that by 1995 and into 2001, the 8-hour CO
average will remain below the NAAQS of 9.0 ppm at the Thomas and Grand Avenue hot spot.
However, elsewhere, the UAM modeling results alone showed that the 8-hour CO average
would continue to exceed the 8-hour CO NAAQS.
C-37
-------
It
&
e
o
u.
—
§•
o
X
u
venu
•a
e
2
O
3
o
S
"3
01
•3
1
o
E
e
U
e
U
u.
5
H
S X
1 a
•a 3
« <
S u
Is
•a <
M ~3
s
i
§ e
I1
= ^
<••> -b
cjS
7j
E o
1 +
•g s
1 D
e
en
Emission ;
u
•« P-l OO «— OS p< P4 »- s§ o OS -« —
00 P- [5 •"*• O NO NO NO * •<*• g!J I/I V>
— o'ojo'o'dd dgjd^dcs
P"^ ^M ^^ ^N ^^ •— I *•• ^-* ^* ^H
So ON — • *o — • *ooor*»^»nvo
«fnoocsr^
^^odoo o^oddo
OO "^ ^^ ^N f^ ^f ^f f^ ^^ &* if (^ 00
-— oo^vo^cor*- ow^t*-»ooooo
O -*»^) cu ^
2 « 2 b
oo 'S o .O 'S^o*
? sir s i t
1 « r? 8 1 i ® S 8 S- s
§ >^"2Qa >^IQi
Q at! @ ^ S '"3 oJ © ^ S '-5
* *" J8 1 S 0 * J! 1 S <5
i s { * 1 1 s H n
Q s@l|e--i ! ® 1 §• & -3
- 3sc.?3S * e e ^ s «
e u 5 5 > 5 oi Uuu>eQi
s 1 S & 1 1 * S ? S 1 1 •&
Q taOOas>H cqOOos>p-
.° .° -2 -2
i 'i a. i a. |
I 8 * 2Q S p 2' a 8
c/j ««« o^-isEuQo50 o > ™ a. Q Q <*
os 1— 3XI2>z5Zos'«
g£ng^caOc40:>HoN'i:aOrMflfi>H§;
-• •* r~ os o o
p^ O -; p PJ PJ
OO OO P" P^ O P~
NO P* ^ ON NO vn
00 p-' od r~ oo' cc
oo *o ^^ lo oo •••
-; ON P-) r; O —
m' d d o o o
^ ". <*. ~~. ™. ^
P4 NO 00 \O •* —
ON. ON NO »^ OO 00
oo" r-^ oo' oo oo co
__
«
o. J:
2 «
'3 o J4
a. - "E
o | E
§T ri "f O g
as @ 1 § •?
g 1 1 1 1
U g S > 1 2
S »S >J ."2 ^ g.
3 x x S o c
03 O O a! > t-
flu
3 x ^ > i S
CQ o pj as ;> H
C-38
-------
1
1
g
9
U
d
U
OQ
<
_*>
E Z
2 CX
13
« <
S u
E
3 ^a
E •§
'Si
s
i
H
-
y r.
?r •
»'l
<•» a
S"
*J
I u
l|
5- ?
« 3
9
Emission Scenari
1
U
8 3 £
— do
r- o o
•* •fl; rn
oo ^^ oo
5 2 -
in — d
M (M CM
odd
p» ft oo
>O in ^
oo r~ oo
1
0
(ft MB
— ^*
CU "
0£ ©
* -s
°* -9
9 ® 1
a g g
S 1 ?
CD O O
1 >
s p
50 s >• z
8
00
00*
—
d
cs
d
8
00
^3
1
I
X
O
1
<0
g
s
Trip Reduction Oi
o
e
OQ
C
Q
•o
o
D.
•a
00
S
o
'§
u
a
u
00
"S.
1
1
borhood
•a
i
w
^
«
u
•a
3
J
cf
o
•a
2
1
i
J
u
15
e
3
s
o
o
u
u
so
1
1
00
E
|
S
-
_j
u
V
<
i
ribution to maximu
I
o
X
a
u
r.
0
1
.§
s
u
1
b
u
X
I
"{•
i
03
=9
g
1
i
s
ire-year 1
a
•g
"2
e
o
V
o
u
O
U
u
&
o
maximum 8-hour a
design day.
•!^
g 3
S u
1 '1
M
£ .a
.
1
§
'§
r average concentr
|
OO
g
1
a
<*
g
fr
*•
o
JS
u
«
J
*S
g
t>
y
S
2
u
•g
1
o*
m
y
1
c
a
§
'«
r average concentr
|
ao
E
a
1
«
C-39
-------
Appendix C
References
Causley, M. C., and L. L. Duvall. 1992. "TRFCONV User's Guide." Systems
Applications International, San Rafael, CA (SYSAPP-92/021).
Causley, M. C., J. L. Haney, R. G. Ireson, J. G. Heiken, H. Tunggal, A. B.
Hudischewskyj, and S. B. Shepard, 1991. "Carbon Monoxide Air Quality
Modeling of the Phoenix Metropolitan Area in Support of the Federal
Implementation Plan, Volume I: • Main Report." Systems Applications
International, San Rafael, CA (SYSAPP-91/105a).
Haney, J. L., and R. G. Ireson, 1988. "Application of the Urban Airshed Model for Carbon
Monoxide (CO) in Phoenix, Arizona." Systems Applications International, San
Rafael, CA (SYSAPP-88/083). Presented at the 81st Annual Meeting of the Air
Pollution Control Association, Dallas, Texas, June 20-24, 1988.
Killus, J. P., G. Z. Whitten, and R. G. Johnson, 1982. "Modeling of Simulated
Photochemical Smog with Kinetic Mechanisms." Systems Applications International,
San Rafael, CA (Publication no. 82172).
Schere, K. L., and K. L. Demerjian, 1977. Calculation of Selected Photolytic Rate
Constants Over a Diurnal Range. U.S. Environmental Protection Agency (EPA-
600/4-77-015).
TRB. 1985. "Highway Capacity Manual. Special Report No. 209." Transportation
Research Board, National Research Council, National Academy of Sciences.
U.S. Department of Transportation. 1978. The Determination of Vehicular Cold and Hot
Operating Fractions for Estimating Highway Emissions. DOT-FH-11-9207, prepared
by State of Alabama Highway Department, Montgomery, Alabama for U.S.
Department of Transportation, Washington, DC.
U.S. Environmental Protection Agency. 1990. User's Guide for the Urban Airshed Model.
Volumes I-V. EPA-450/4-90-007, U.S. Environmental Protection Agency (NTTS
No: PB91-131243).
U.S. Environmental Protection Agency. 1991. User's Guide to MOBILE4.1. U.S.
Environmental Protection Agency, Ann Arbor, MI (July 1991).
U.S. Environmental Protection Agency. 1992. User's Guide to CAL3QHC. Research
Triangle Park, NC (to be issued in August 1992).
C-40
-------
TECHNICAL HEPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO 2.
EPA-450/4-92-011b
4. TITLE AND SUBTITLE
Guideline for Ragulatory Application of the Urban
Airshed Model for Areawide Carbon Monoxide
Volume II: Appendices
7. AUTHOR(S)
Edward L. Carr, Julie L. Fieber and Robert C. Kessler
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Systems Applications International
San Rafael, California 94903
12. SPONSORING AGENCY NAME AND ADDRESS
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
.Tiine 1992
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO. 5
11. CONTRACT/GRANT NO.
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
IS. SUPPLEMENTARY NOTES
16. ABSTRACT
State Implementation Plan (SIP) revisions demonstrating attainment of the Carbon
Monoxide (CO) National Ambient Air Quality Standards (NAAQS) are required under
the Clean Air Act Amendments of 1990. Urban areawide modeling and intersection
modeling are recommended to address attainment of the CO NAAQS. The Urban
Airshed Model (UAM) has been identified as an effective urban areawide model for
evaluating emission control requirements needed to attain the CO NAAQS. The
purpose of this document is to provide guidance in the procedures used to apply
the UAM for CO SIP attainment demonstrations.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
Atmospheric Dispersion Models
Carbon Monoxide (CO)
State Implementation Plan (SIP)
Clean Air Act Amendments
Urban Areawide Modeling
18. DISTRIBUTION STATEMENT
b. IDENTIFIERS/OPEN ENDED TERMS
* •
19. SECURITY CLASS (This Report)
2O. SECURITY CLASS (This page)
c. COSATl Field/Group
21. NO. OF PAGES
55
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
------- |