United States Office of Air Quality EPA-450/4-81-005a
Environmental Protection Planning and Standards January 1981
Agency Research Triangle Park NC 27711
Air
&ER& Application Of The
Empirical Kinetic
Modeling Approach
To Urban Areas
Volume I: San Francisco/
Sacramento
-------
EPA-450/4-81-005a
Application Of The Empirical Kinetic
Modeling Approach To Urban Areas
Volume I: San Francisco/Sacramento
by
G.Z. Whitten, H. Hogo and R.G. Johnson
Systems Applications, Incorporated
950 Northgate Drive
San Rafael, California 94903
Contract No. 68-02-3376
EPA Project Officer: Gerald L. Gipson
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Air, Noise and Radiation
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
January 1981
-------
DISCLAIMER
This document is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are
available free of charge to Federal employees, current EPA contractors and
grantees, and nonprofit organizations - in limited quantities - from the
Library Services Office (MD-35), U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina 27711; or, for a fee, from the
National Technical Information Service, 5285 Port Royal Road, Springfield,
Virginia 22161.
This report was furnished to the Environmental Protection Agency by
Systems Applications, Incorporated, 950 Northgate Drive, San Rafael,
California 94903, in fulfillment of Contract No. 68-02-3376. The contents
of this report are reproduced herein as received from Systems
Applications, Incorporated. The opinions, findings and conclusions
expressed are those of the author and not necessarily those of the
Environmental Protection Agency.
Publication No. EPA-450/4-81-005a
32R/1
ii
-------
CONTENTS
DISCLAIMER i i
LIST OF ILLUSTRATIONS v
LIST OF TABLES vii
LIST OF EXHIBITS ix
1 INTRODUCTION .- 1
1.1 Background and Objectives 2
2 SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS 4
2.1 Summary 4
2.2 Conclusions 4
2.3 Recommendations 6
3 TRAJECTORY MODEL APPLICATIONS 9
3.1 San Francisco Applications 9
3.1.1 Model Definition 11
3.1.2 San Francisco Case Studies 18
3.1.3 Summary of Results for San Francisco
Case Studies 31
3.2 Sacramento Applications 33
3.2.1 Model Definition 35
3.2.2 Sacramento Case Studies 43
3.2.3 Summary of Results for Sacramento
Case Studies 51
4 SENSITIVITY STUDIES 57
4.1 The San Francisco Region 58
4.1.1 Initial Conditions 58
4.1.2 Mixing Heights 58
4.1.3 Emissions Vector 60
4.1.4 Model Size 60
4.2 The Sacramento Region 63
32R/1
m
-------
5 EKMA APPLICATIONS 66
5.1 Considerations Related to the Use of
Isopleth Diagrams 67
5.1.1 Fixed Monitoring Sites 67
5.1.2 Uneven Changes in Emissions 67
5.1.3 Transported Pollutants 68
5.2 San Francisco Applications 68
5.2.1 Basic Comparisons between LIRAQ and EKMA 68
5.2.2 Further Investigations... 74
5.3 Sacramento Application 86
REFERENCES 103
32^/1
iv
-------
ILLUSTRATIONS
1 Trajectory 1 for 26 July 1973 19
2 Trajectory 2 for 26 July 1973 21
3 Trajectory 1 for 20 August 1973 23
4 Trajectory 2 for 20 August 1973 25
5 Trajectory 1 for 24 July 1974 27
6 Trajectory 2 for 24 July 1974 29
7 Locations of Monitoring Sites in the Sacramento Area 34
8 Trajectory Used for 24 August 1976 44
9 Comparison of Ozone between Station Observations
and Model Predictions for City-Specific EKMA
for 24 August 1976—Sacramento 49
10 Trajectory Used for 28 June 1976 50
11 Comparison of Ozone between Station Observations
and Model Predictions for 28 June 1976—Sacramento 54
12 Effects of Inversion Heights on Maximum Ozone
Produced: 26 June 1973, Trajectory 1 61
13 Effects of Wind Speed across San Francisco on
Ozone Production: 26 June 1973 62
14 Effects of Trajectory Width on Maximum Ozone:
26 July 1973 64
15 Isopleth for 26 July 1973: Trajectory 1—HC/NOX
Ratio = 10/1 71
32R/1
-------
16 Level III Trajectory for San Francisco--26 July 1973 72
17 Isopleth for 26 July 1973: Trajectory 1—HC/NOX
Ratio = 1/1 77
18 Isopleth for 26 July 1973: Trajectory 1--HC/NOX
Ratio = 100/1 78
19 Isopleth for 20 August 1973: Trajectory 1 80
20 Isopleth for 24 July 1974: Trajectory 2 81
21 New-Style Isopleth (1,1) for 26 July 1973:
Trajectory 1 84
22 New-Style Isopleth (1,1) for 26 July 1973:
Trajectory 1 Using the Carbon-Bond Mechanism 87
23 Map of the Hypothetical Urban Area Used to Study
Transportation Control Strategies 88
24 Trajectory Used for the Control-Strategy Runs 90
25 City-Specific EKMA Isopleth for the Control-Strategy
Calculations 95
26 Level III Trajectory for Hypothetical Region:
24 August 1976 100
32R/1
vi
-------
TABLES
1 Input Data for the San Francisco Trajectories 14
2 Summary of Results for the San Francisco Case Studies 32
3 OZIPM/CBM Emissions and Mixing-Height Inputs for
24 August 1976 45
4 Initial Conditions for Calculations of 24 August 1976 46
5 Comparison of the SAI Trajectory Model and the OZIPM/CBM
Calculations for 24 August 1976 for NOX, PAR, ARO,
and Ozone 47
6 Emission Rates and Mixing Heights Used in the
OZIPM/CBM Calculations for 28 June 1976 52
7 Initial Conditions Used in the City-Specific EKMA
Calculations of 28 June 1976 53
8 Comparison of Ozone Predictions for Four Different
Models on 28 June 1976 and 24 August 1976 in
Sacramento 55
9 Values for Maximum Ozone Versus Values for Initial
Condi tions 59
10 Control-Strategy Comparisons for San Francisco
between LIRAQ and EKMA Applications 70
11 Initial Conditions Used in the Level III EKMA
Calculations for 26 July 1973 in the San Francisco
Region 75
32R/1
vii
-------
12 Sensitivity of Control-Strategy Evaluation to HC/NOX
Ratio for Trajectory 1--26 July 1973 79
13 Sensitivity of Control-Strategy Evaluation to
Traj ectory Path 83
14 Control-Strategy Comparisons Using a New-Style Isopleth.... 85
15 SAI Airshed Model Predicted Peak Ozone Concentrations
Computed for Different Control Strategies in the
Hypothetical Region 89
16 Emission Rates and Mixing Heights Used in EKMA
Applications to Hypothetical Region Control Strategies 91
17 Trajectory Model Outputs Used as Inputs for the
EKMA Simul ations 92
18 Comparison of Maximum Predicted Ozone for the SAI
Airshed, SAI Trajectory, OZIPM/CBM, and City-Specific
EKMA Models: Base-Case Conditions Used for the
Control-Strategies Study of the Hypothetical Region 93
19 Control-Strategy Study Total Emissions Resulting
from a 30 Percent Reduction in Mobile HC and NOX
Emissions 96
20 Comparison of Maximum Ozone Predictions between the
Trajectory Model, OZIPM/CBM, and City-Specific
EKMA with Different Control Scenarios 97
21 Initial Conditions Used in the Level III EKMA
Calculations for Hypothetical Region for 24 August 1976
Meteorol ogy 98
32&/1
viii
-------
EXHIBITS
1 Q-tapes and Q-files Provided to the EPA by the BAAQMD 10
32^/1
ix
-------
1 INTRODUCTION
The Environmental Protection Agency (EPA) has developed a method for
estimating the emission controls that will be needed to meet the National
Ambient Air Quality Standard (NAAQS) for ozone (03) concentrations in
urban areas. This method, known as the Empirical Kinetic Modeling
Approach (EKMA), uses an isopleth diagram for ozone concentrations that is
related to hydrocarbon (HC) and nitrogen oxide (NOX) precursor levels.
Each point on the isopleth diagram represents the maximum one-hour ozone
level reached as a result of the specific HC and NOX combination described
by the abscissa and ordinate values of the isopleth diagram. The ozone
value generated from each HC and NOX combination is obtained by a com-
puter-based trajectory model—OZIPP (Whitten and Hogo, 1978a). OZIPP con-
tains many simplifying assumptions that are designed to strike a balance
between factors such as the state of knowledge, data availability, com-
puter size, predictive accuracy, and overall cost.
This report presents the results of one of several studies that
evaluate the EKMA. The approach taken in this particular evaluation of
the EKMA uses gridded input files that were previously prepared for use
with large, sophisticated airshed models of the grid variety. One of
these models—LIRAQ--was used to prepare an Air Quality Maintenance Plan
(AQMP) for the San Francisco urban area (BAAQMD, 1979). The other—the
SAI Urban Airshed Model—had been used in a mobile-source-control evalua-
tion study conducted for the California Department of Transportation (DOT)
(Reynolds et al., 1979) in the Sacramento region. When the gridded data
from these more sophisticated models are input to the OZIPP trajectory
model, the results can be compared with those of the other models to form
one basis for evaluating the EKMA.
This study also compared the OZIPP trajectory simulations either with
observed data or with predictions made by the other models on the basis of
change in emissions values. In addition, parts of this study involved
alternative input assumptions and changes to the model definitions so that
both sensitive parameters and improvements to the EKMA could be evaluated.
32R/2
-------
1.1 Background and Objectives
For several years the Office of Research and Development of the EPA
has sponsored a coordinated program consisting of fundamental chemical
research, smog chamber experimentation, and model development. The
objective of this program is to develop mathematical models capable of
simulating the dynamics of the chemical reactions and dispersion of the
gaseous pollutants found in the lower troposphere. These atmospheric
models are to be used both to gain an understanding of air pollution and
to predict the probable outcome of control scenarios. To this end, the
EPA Office of Air Quality Planning and Standards (OAQPS) has embarked on
an extensive program of model comparison and evaluation.
The specific purpose of the OAQPS program is to assess the suita-
bility of available photochemical oxidant modeling approaches for use by
states as planning tools in ozone air quality planning and management.
The program includes comparisons and evaluations of the various approaches
for typical U.S. cities. As part of the program, large, detailed data
bases are being assembled for several example cities.
The primary objective of this study is to evaluate one such approach-
—the EKMA—through the application of the photochemical trajectory model
that forms its basis. This model is used to generate isopleth diagrams of
constant urban ozone maximums as a function of the morning precursor
levels of nonmethane hydrocarbons (NMHC) and nitrogen oxides. The EKMA
method uses these results to estimate, in a relative manner, the precursor
control requirements necessary to reduce ozone levels below the National
Ambient Air Quality Standard for ozone. The method was developed for use
at several levels of sophistication, depending on the extent of input data
available. If the basic model is fundamentally sound, the simulated
results should approach the observed values as the availability of
detailed data increases.
The EPA has prepared highly detailed data concerning precursor
emissions, meteorology, and ambient air quality for Tulsa and Philadelphia
to test various air quality simulation models such as the OZIPP and the
SAI Urban Airshed Model. In addition, there are data bases already
available for St. Louis, Los Angeles, San Francisco, and Sacramento.
Because some models are grid-based and thus designed to reproduce spatial
and temporal variations within the metropolitan airshed, detailed
information files are required. The availability of detailed data and of
the results from parallel studies using sophisticated models offers the
following advantages for making an in-depth study of the EKMA:
> The simple application of the OZIPP trajectory model,
using either the actual available data inputs or the
32R/2
-------
corresponding data simulated in other models, should
generate results that can be compared both with ambient
air data and with the results of the other models. The
timing of results regarding data such as the ozone
maximums and precursor decay should also be comparable.
Tests of the sensitivity of the results to variations in
the input data can demonstrate important data requirements
for successful application of OZIPP. This information
should be useful in determining cost effectiveness for
future uses. The results of the sensitivity study are
also useful for improving the trajectory model in case the
present version does not adequately simulate observed
ambient data or the simulations of more complex models.
The application of the EKMA procedures to air quality
planning demonstrates the utility of the EKMA compared
with that of methods based either on simpler models (such
as a rollback model) or on more complex models (such as
the SAI Urban Airshed Model).
The potential effectiveness of OZIPP model improvements
can be assessed using the results of the complex models as
a standard.
32R/2
-------
SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS
2.1 SUMMARY
We have evaluated the EKMA by comparing OZIPP with other models pre-
viously applied to the San Francisco and Sacramento regions. The study
was carried out at several levels, beginning with the trajectory model
that forms the basis of the EKMA and ending with the control-strategy
predictions that result from employing the EKMA methodology. In the San
Francisco region OZIPP was compared with the LIRAQ model, which was used
for the 1979 Air Quality Maintenance Plan (AQMP). In the Sacramento
region comparisons were made with the SAI Airshed Model, which has been
applied in a study for the Department of Transportation (Reynolds et al.,
1979).
The basic OZIPP model is a simple, moving-air-parcel, or trajectory,
model that uses a detailed chemical mechanism for the surrogate propylene
and butane hydrocarbons. Time-dependent, precursor-emissions factors can
be treated, along with expansion of the air parcel, and entrainment is
treated by assuming that constant concentrations exist outside the parcel.
2.2 CONCLUSIONS
The primary emphasis of this comparison study was directed toward the
discovery of features in the trajectory model that could explain dif-
ferences in the results of OZIPP from those of some other model. To that
end, most of the OZIPP simulations utilized initial conditions derived
from the model used for comparison purposes. Thus, for the most part, the
study used the LIRAQ and Airshed Model data files to generate appropriate
OZIPP inputs. Although the utilization of these files tends to circumvent
the issue of normal-source data preparation, the comparison of models is
clarified by use of common data bases. Furthermore, costs are mini-
mized. For some of the OZIPP simulations, we used air quality data rather
than information common to the comparison model. This step allowed the
categorization of overall differences as either effects resulting from the
data utilized or those attributable to specific features of the models.
32R/2
-------
This study, and a similar study by Whitten and Hogo (1981) using Los
Angeles data, show that OZIPP can provide results that are often quite
similar to those of the more complex models. By carefully separating the
models according to specific formulation differences and through careful
analysis of intermediate results, the most important feature in formula-
tion that produced different results between two models could often be
isolated. Even though the absolute values predicted by the OZIPP trajec-
tory model may be significantly different from those of the more complex
models (one of the main differences is often a result of the different
chemical mechanisms employed), the isopleth applications of the EKMA
(which uses the trajectory model results in a relative fashion) typically
show control predictions corresponding to those of the more complex
models.
In the comparison of the EKMA results with those of the LIRAQ model
for the San Francisco region, primary attention was given to the tra-
jectory leading to the design ozone of the AQMP. Although there are
uncertainties associated with deriving any specific trajectory path, in
this study the air parcel that eventually affects the site of maximum
ozone concentration (Livermore) apparently originates at 0800 PDT over the
Pacific Ocean just south of San Francisco. This trajectory is discussed
in detail in section 3.2. It was found that the precursor loading for
this trajectory and for several others was extremely sensitive to wind
speed because of widely varying emissions density in the area.
Another sensitive input was the size of the air parcel. The emis-
sions files used in the LIRAQ studies were for 5-km grids. Using a narrow
size (<5 km) for the width of the EKMA trajectory made the results
sensitive to the exact choice of a path between, or over, the grid squares
of the LIRAQ inventory. Sometimes the emissions density changed by more
than two orders of magnitude between adjacent grid squares. For most
cases, a 10-km width produced reasonable and similar results for slightly
different trajectory paths. Some trajectories were 15 km wide, roughly
coinciding with the size of the city of San Francisco proper, the main
source of the emissions.
The SAI Airshed, or grid, Model served as the basis of the Sacramento
study. SAI trajectory model simulations were made using recreated input
files from the earlier grid model study. In some instances, important
differences between grid and trajectory model results could not be
resolved. For these cases, the study focused largely on comparing the
OZIPP results with those of the SAI trajectory model (rather than the grid
model) because the trajectory model was used to generate input data for
OZIPP. Obtaining data directly from the trajectory model eliminated many
uncertainties in the model comparisons, such as those related to model
size, wind speed, or mixing depths. Differences due to model formulation
32R/2
-------
thus became more apparent than those of the LIRAQ comparisons. The most
sensitive differences, both absolutely and relatively, were found in the
chemical mechanisms. '
The SAI models used in the Sacramento study employed an early version
of the Carbon-Bond Mechanism (CBM) (Whitten, Hogo, and Killus, 1980). In
general, the CBM chemistry generated higher absolute levels of ozone than
the standard EKMA chemistry. In the control-prediction phase, the two
chemistries predicted similar results when HC and NOX were simultaneously
reduced; but the CBM chemistry predicted more ozone reduction than the
standard EKMA chemistry when HC reductions only were involved. Although
the OZIPP model and the SAI trajectory model are much different in terms
of complexity (apart from chemistry), these factors apparently did not
play a significant role for the Sacramento area. Thus, such factors as
multilevel mixing, eddy diffusion, elevated emissions, and variable
concentrations aloft were found to be unimportant when common input data
were used.
2.3 RECOMMENDATIONS
The following considerations are suggested for future studies:
> For cases in which differences in control-strategy
predictions are found to be related to differences in the
chemistry employed in various models, it is important to
investigate the chemical reactions that produce these
differences. If appropriate smog chamber experiments that
would validate one or the other types of mechanisms can be
identified, these experiments should be performed.
> A generic concept can be associated with the EKMA that
goes beyond its present formulation and uses. Develop-
mental work on the EKMA should continue to update its
formulation and improve its utility. Specifically, the
following steps are suggested:
- This study demonstrates that significant differences in
both absolute model results and in predicted control
requirements can sometimes occur when the Carbon-Bond
Mechanism is substituted for the propylene/butane
chemistry currently used in OZIPP. Further studies
should be undertaken to investigate and resolve
potential discrepancies between the two mechanisms.
32R/2
-------
- Credibility of the EKMA could be enhanced by providing
a means for considering urban hydrocarbon reactivity.
Either the latter should be related to the propylene
reactivity used in OZIPP, or the chemical mechanism
utilized in the OZIPP should be changed to be compat-
ible with available reactivity information.
- Isopleths based on emissions inventories might be
further developed to perhaps replace the present
initial-condition-based isopleths. Urban HC and NOX
measurements are often not relevant to the initial
conditions that are appropriate to a city-specific
trajectory, as this study demonstrated for several of
the San Francisco trajectories.
- The mixing-depth algorithm should be improved so that
the changes with time correspond to observations rather
than produce a constant-dilution factor.
- The transported pollutants, both in the surface layer
and aloft, might be varied according to the level of
control of emissions. The variance could be controll-
able by the user. The current procedure necessitates
two separate isopleth diagrams. (Again, further
development would be necessary to incorporate this
refinement.)
- The regular city-specific algorithm might be modified
to allow optional starting times in place of the
present 0800 PDT starting time. Earlier starting times
would emphasize dependence on an emissions inventory,
whereas later starting times might emphasize dependence
on initial conditions.
Modifications to the current version of the EKMA have also
been suggested in other studies. Although such sugges-
tions have not been addressed in this study, their
implementation might also be considered.
- The two-stage approach (in which stationary-source
emissions are held constant in the first stage and
mobile-source emissions are held constant in the second
stage) might be further developed so that local
districts can use EKMA isopleths to make the control
predictions for stationary emissions.
32R/2
-------
- Point-source emissions might be handled so that the
interaction with all possible urban HC/NOX concentra-
tions could be evaluated as part of a new source review
procedure.
- Percent cutback diagrams (PCD) might be added to the
OZIPP code so that both the isopleth and the PCD are
generated.
- Temperature changes, deposition, variable reactivities,
nitrate chemistry, and sulfate chemistry can also be
added to OZIPP. Isopleths of species other than ozone
can also be generated by OZIPP.
- Statistical packages can be combined with the EKMA to
generate the probable number of exceedances resulting
from various control scenarios.
> The use of more complex models such as the SAI Airshed
Model should be accompanied by the use of the EKMA, both
for absolute trajectory simulations along the path to the
major ozone peaks and for control-strategy guidance from
the isopleth diagrams. The studies performed to date show
that unless the wind fields are complicated by such
factors as wind shear, the two types of models normally
give similar results. In the absence of complex meteoro-
logical factors, important differences in results would
indicate the existence of problems in one or both of the
models that should be investigated.
32R/2
-------
TRAJECTORY MODEL APPLICATIONS
This section focuses on applying an EKMA-type trajectory model to
predict absolute values of ozone. Predictions were made for several cases
in San Francisco and Sacramento. The predicted ozone levels were compared
with observed concentrations, when these were available, and with predic-
tions of more complex models. The results can be used to assess the
fundamentals of the formulation model and to identify important differen-
ces between models.
3.1 San Francisco Applications
The 1979 air quality maintenance plan for the Bay Area Air Quality
Management District (BAAQMD) was developed using the Livermore Regional
Air Quality (LIRAQ) Model (BAAQMD, 1979). During this study, sufficient
air quality, emissions, and meteorological data were collected to permit
application of the OZIPP-type trajectory model. The trajectory model
predictions could then be compared with both observed data and the LIRAQ
predictions. Furthermore, the control strategies tested using the LIRAQ
model could be tested using the EKMA technique to compare its utility with
that of the more complex LIRAQ model. (The control strategy tests are
addressed in greater detail in chapter 5.)
The LIRAQ model was used to simulate the distribution of photochemi-
cal pollutants in the BAAQMD on two days—26 July and 20 August 1973. The
data necessary for applying the LIRAQ model on these two days are avail-
able on tapes supplied by the BAAQMD and are identified in exhibit 1.
Many of these data have also been reported by Feldstein et al. (1979) and
others. Using the data for these two days, we undertook specific case
studies using the OZIPP-type trajectory model. Case studies for one
additional day (24 July 1974) were also performed, even though the results
could not be compared with those of the LIRAQ model, but only with
observed air quality data.
In the discussion that follows, the LIRAQ model and the OZIPP-type
trajectory model are first described. Then, specific case studies are
discussed. Finally, the overall results are summarized.
32R/2
-------
Attention:
John Siuuerhiys (MD-14)
SHIPPING LIST
Q-TAPES AND Q- FILES PROVIDED TO EPA, PUR BAAQMD-ABAG CONTRACT
BAAQMD Assigned File
Q-Tape Namber Number
1 1
2
3
4
S
2 1
2
3
4
3 1
2
3
4
5
4 1
2
3
4
5
S 1
2
3
4
5
6
6 1
2
3
4
S
6
7 1
2
3
4
S
6
S 1
2
3
4
9 1
2
3
4
S
10 1
2
3
4
S
Description of Q-Files, in Order on Q-T«pe
QSRUN. July 26, 1973
QSOR, 1975 Baseline, Region 1. S Kms
QICON, July 26, 1973
QGEO
QRAD, July 26, 1973
Nota Bene; CEO repeated on Q-Tape 9,
QSOR for 75 Baseline, Region 1, 5 Kms.
is repeated on Q-Tape 7.
QTRANO, July 26, 1973
QTRAN1, July 26, 1973
QTRAN2, July 26, 1973
QTRAN3, July 26, 1973
QICON, Nov S-6, 1976
QTRANO, Nov 5-6, 1976
QTRAN1, Nov 5-6, 1976
QTRAN2, Nov S-6, 1976
QTRAN3, Nov 5-6, 1976
QICON, July 24, 1974
QTRANO, July 24, 1974
QTRAN1, July 24, 1974
QTRAN2, July 24, 1974
QTRAN3, July 24, 1974
QSOR, 2000 Baseline, Region 1, 5 Kms
POP, 2000 Baseline, Region 1, 5 Kms
AIR, 2000 Baseline, Region 1 , S Kms
CAR, 2000 Baseline, Region 1, 5 Kms
PNT, 2000 Baseline, Region 1, S Kms
SKY, 2000 Baseline, Region 1, S Kms
QSOR, 1985 Baseline, Region 1, 5 Kms
POP, 1985 Baseline, Region 1, 5 Kms
AIR, 1985 Baseline, Region 1. 5 Kms
CAR, 1985 Baseline, Region 1, S Kms
PNT, 1985 Baseline, Region 1 , 5 Kms
SKY, 1985 Baseline, Region 1 , 5 Kms
QSOR, 1975 Baseline, Region 1 , 5 Kms
POP, 1975 Baseline, Region 1, 5 Kms
AIR, 1975 Baseline, Region 1 , 5 Kms
CAR, 1975 Baseline, Region 1 , 5 Kms
PNT, 1975 Baseline, Region 1 , 5 Kms
SKY, 1975 Baseline. Region 1. S Kms
QMRUN, July 24, 1974
QMET, July 24, 1974
QMRUN, Nov S-6. 1976
QMET, Nov 5-6, 1976
Nota Bene; These are all MASCON inputs.
QGEO
QMRUN, Aug 20, 1973
QMET, Aug 20, 1973
QMRUN, July 26, 1973
QMET. July 26, 1973
Nota Bene; These are all MASCON inputs.
QGEO repeated on Q-Tape 1. QGEOs identical.
QSRUN, Aug 20, 1973
QICON, Aug 20, 1973
QTRANO, Aug 20, 1973
QTRAN1, Aug 20, 1973
QTRAN2, Aug 20, 1973
EXHIBIT 1. Q-TAPES AND Q-FILES PROVIDED TO THE EPA BY THE
BAAQMD
32R 3
10
-------
3.1.1 Model Definition
The LIRAQ model is of the grid variety, providing estimates of
pollutant concentrations as a function of time throughout the modeling
region. The model mathematically simulates the complex physical and
chemical interactions taking place in the atmosphere, employing a lumped
parameter chemical mechanism to describe the photochemical reactions
taking place. Specifically, the model predicts instantaneous pollutant
concentrations in single-layer grids superimposed on the modeling
region. Surface concentrations are estimated by an analytic function.
For a more detailed description of the model, the reader is referred to
the study of MacCracken and Sauter (1975).
The intent of applying the OZIPP-type trajectory model is to follow
the evolution of photochemical pollutants, particularly ozone, along the
air parcel trajectory leading to an observed, peak ozone concentration.
This approach is consistent with that which the EPA has termed a Level II
analysis (Federal Register, 1979). The standard EKMA trajectory model
contained in the computer program OZIPP (Whitten and Hogo, 1978a) was not
applicable to such an analysis; however, the subsequent program OZIPM
(Whitten and Hogo, 1978b), contains options that facilitate such applica-
tions and, thus, was used in this study.
The OZIPP model was based on the reactions of high initial precursor
concentrations in a major urban-source region. The model follows the
original air parcel throughout the day as it travels to a suburban site
where ozone maximums have been observed. This conceptual pattern was
similar to the pattern of ozone development seen in the smog chamber
experiments used to validate the OZIPP model (Dimitriades 1977; Dodge,
1977). The OZIPM model permits the use of alternative starting times and
hourly mixing height inputs that are more representative of the conditions
considered in this study. This version of the model also provides the
option of replacing the standard EKMA chemistry with other mechanisms. As
described later, this option was used in some portions of the study.
The detailed trajectory model approach using the OZIPM model requires
the definition of numerous variables, including the air parcel trajectory
leading to a peak ozone level; initial pollutant concentrations within the
air parcel; the diurnal variation in mixing height; emissions occurring
along the trajectory path; pollutant concentrations above the mixed layer
that may later be entrained into the surface-based mixed layer; and
chemical reactivity. The procedures for defining these variables are
described in the following subsections.
32R/2 11
-------
3.1.1.1 Trajectory
To specify trajectories describing the leading of air parcels to
elevated ozone levels, first, the two highest observed ozone concentra-
tions for each of the three study days were obtained from the LIRAQ QICON
(the file containing observed air quality data). Wind-field maps were
then created using the LIRAQ QMET input files for these days. For
example, on 26 July 1973, the maximum ozone level was 0.176 ppm at
Livermore at 1700 PDT. According to the LIRAQ surface wind-field maps,
the direction and speed for the previous hour (1600 to 1700 PDT) at
Livermore was 25 km/hr at 250°. The location of the air parcel at 1600
PDT was thus determined by moving 25 km from Livermore at an angle of
250°. This procedure was repeated for each preceding hour until the
parcel reached the ocean.
In this case, the trajectory ending at Livermore had passed over
South San Francisco in the morning. According to the QMET file for this
day, wind vectors were to the north and south of the trajectory. The
peninsula to the south of San Francisco is a source of high emissions and
the composition of the air parcel is thus extremely sensitive to the speed
with which it travels across the peninsula. Because the trajectory passed
directly over the San Francisco International Airport (SFO), wind speed
data from the airport were used with the QMET data to determine the
trajectory path across the peninsula.
By definition, then, the trajectory path ended at a point of high
ozone concentration. However, additional concentration values also were
available from the closest monitoring station for the hour during which
the trajectory passed by. The contour lines presented in the LIRAQ
reports (see MacCracken and Sauter, 1975) show that, at least, the ozone
levels simulated in the LIRAQ model do not have many steep gradients.
Therefore, the nearest station value and LIRAQ model predictions could
also be compared to the OZIPM predictions along the trajectory path.
3.1.1.2 Initial Concentrations
All trajectory paths studied for the San Francisco area appeared to
originate over the Pacific Ocean. Thus, initial concentrations were
considered to be insignificant and, therefore, were assumed to be at
background levels. Sensitivity of model predictions to this assumption is
further evaluated in chapter 4.
32R/2 12
-------
3.1.1.3 Mixing Heights
Mixing heights were obtained from the LIRAQ QTRAN files. Since these
data are presented for three-hour periods only, values for the remaining
hours were obtained by linear interpolation. Thus, a mixing height for
0800 PDT would be two-thirds of the 0700 PDT value plus one-third of the
1000 PDT value. The mixing heights used in this study are given in table
1.
The mixing layer typically increases in thickness as the warm surface
air erodes the inversion layer. Under such conditions air is entrained
from aloft that, in turn, increases the volume of the conceptual air
parcel of the OZIPM trajectory model. Under some circumstances the change
in mixing height may not be a result of entrainment from aloft. For
instance, an air mass moving at high velocity would initially tend to flow
over a stagnant air mass. Also, in areas of convergent wind flow such as
in the case of air passing from open terrain into a canyon, mixing height
cari be pushed up from the sides. In this type of change in mixing
height—narrowing of the sides of an air parcel, accompanied by an
increase in the height of the air parcel—the volume of the air parcel may
not change. Since changes in the mixing height in the OZIPM model are
intended to represent a change in volume for the moving air parcel,
changes in mixing height that do not reflect a change in volume should not
be included as input data. Unfortunately, the wind flow in complex
terrain is not easily separated into mixing-height changes, either those
accompanied by entrainment of air from the sides or those accompanied by
entrainment of air from aloft. Although entrainment of air into the sides
of the EKMA model could be treated as entrainment from aloft, the basic
concepts behind the OZIPM model suggest that such air is considered to
contain concentrations identical with those of the model parcel itself.
Therefore, the data related to air entrained from the sides would
normally be ignored along with any changes in volume associated with such
phenomena.
3.1.1.4 Emissions
The version of OZIPP that is available in the OZIPM computer code
contains options for post-0800 emissions of both HC and NOX. For vir-
tually all the trajectories used in this study, initial conditions were
not significant, and the emissions occurring along the trajectory were
responsible for the ozone maximums. The OZIPM model uses post-initial
emissions input data in the form of fractions related to the initial
concentrations. If the initial HC concentration was 1 ppm, emissions
during some subsequent hour sufficient to add an additional 1 ppmC would
merely be included in OZIPM as 1.0 for that hour. If the inversion height
32R/2
13
-------
TABLE 1. INPUT DATA FOR THE SAN FRANCISCO TRAJECTORIES
(a) Initial Concentrations
Species
NOX
NMOC
Oo
Transported
Surface
0.001
0.01
0.0
Pollutants
Aloft
0.01
0.05
0.04
N02/NOX = 0.5.
Surface hydrocarbon reactivity: standard OZIPP conditions.
Aloft hydrocarbon reactivity: standard OZIPP conditions.
26 July 1973
(b) Emission Hydrocarbon Reactivities
20 August 1973
24 July 1974
Species Trajectory 1 Trajectory 2 Trajectory 1 Trajectory 2 Trajectory 1 Trajectory 2
PROP
BUT
FORM
ACET
0.221
0.779
0.008
0.012
0.203
0.797
0.007
0.011
0.223
0.777
0.007
0.011
0.275
0.725
0.008
0.012
0.226
0.774
0.008
0.012
0.225
0.773
0.008
0.013
(c) Mixing Height
(m)
Date
26 July 1973—trajectory 1
26 July 1973—trajectory 2
20 August 1973—trajectory 1
20 August 1973—trajectory 2
24 July 1974—trajectory 1
24 July 1974—trajectory 2
Morning Mixing Height
PDT
100 at 0800
60 at 0800
204 at 0800
142 at 0800
50 at 0800
50 at 0800
Afternoon Mixing Height
PDT
137 at 1300
750 at 1700
730 at 1500
410 at 1600
250 at 1700
670 at 1600
3zR/m
14
-------
TABLE 1 (Continued)
(d) Emissions
26 July 1973—Trajectory 1
Time
(PDT)
1000
1100
1200
1300
1400
1500
1600
1700
26 July
Time
(PDT)
1300
1400
1500
1600
1700
20 August
Time
(PDT)
0700
0800
0900
1000
1100
1200
1300
1400
1500
NOx
(ppm)
0.019
0.043
0.001
0.037
0.0001
0.0007
0.0015
0.0038
HC
(ppmC)
0.172
0.457
0.198
0.491
0.006
0.046
0.042
0.034
1973— Trajectory 2
NOX
(ppm)
0.0028
0.0594
0.0597
0.0241
0.0214
HC
(ppmC)
0.0481
0.8337
0.7324
0.2483
0.336
1973— Trajectory 1
NOX
(ppm)
0.0089
0.0262
0.0002
0.0080
0.0103
0.0056
0.0038
0.0038
0.0042
HC
(ppmC)
0.0546
0.2304
0.0490
0.1223
0.0569
0.0494
0.1575
0.1559
0.0628
32R/m 15
-------
TABLE 1 (Concluded)
20 August 1973—Trajectory 2
Time
(PDT)
1100
1200
1300
1400
1500
1600
1700
24 July
Time
(PDT)
1200
1300
1400
1500
1600
1700
24 July
Time
(PDT)
1000
1100
1200
1300
1400
1500
1600
1700
NOX
(ppm)
0.0159
0.0002
0.0052
0.0219
0.0001
0.0086
0.0524
HC
(ppmC)
0.1849
0.0007
0.0002
0.0041
0.0001
0.0011
0.0032
1974 — Trajectory 1
NOX
(ppm)
0.0055
0.0919
0.0712
0.0400
0.0233
0.0150
HC
(ppmC)
0.1452
0.960
1.402
0.4207
0.1145
0.1283
1974— Trajectory 2
NOX
(PPm)
0.0079
0.0828
0.0207
0.0268
0.0123
0.0183
0.0627
0.0591
HC
(ppmC)
0.139
0.859
0.180
0.200
0.252
0.372
0.694
0.479
32R/lt 16
-------
changed between the time of the initial condition and the time of subse-
quent emissions, the computer code would automatically adjust the emis-
sions factor.
The LIRAQ QSOR input file for 1975 was used to determine the source
emissions. The data in this file, expressed in units of gm/sec/grid-cell,
with 5-km grid cells, were converted to ppm using the following
information:
grams
moles _
liter sec-grid cell
value in
QSOR file
mole
gram
M W
3600 sec
fir
t
one hour
to second
conversion
inversion
height in
meters
1000
convert
to liters
ppm =
_ moles 24.4 a
10
ter mole
ideal
gas
The LIRAQ emissions inventory was prepared by dividing all emissions
of hydrocarbons into three categories (propylene, butane, and aldehydes)
for use in the LIRAQ chemical mechanism (see MacCracken and Sauter,
1975). Therefore, the alkane and alkene values were multiplied by 4
(i.e., butane) and 3 (i.e., propylene), respectively, to obtain units in
ppmC. The alkane, alkene, and aldehyde values were added for each hour to
obtain the hourly hydrocarbon emissions. The reactivities used are given
in table 1.
The width of the trajectory was determined by the emissions grad-
ient. In one case, there was as much as a factor of 100 between grid
cells, and a 15-km-wide trajectory was used.
3.1.1.5 Pollutant Concentrations Aloft
Aircraft data (see MacCracken and Sauter, 1975, pp. 871-948) were
used to determine the concentrations aloft. The values used were the
average of the measured quantities obtained when the aircraft was above
the OZIPM trajectory. Because hydrocarbon data for 1977 were not avail-
able, data for 1974 were used. The concentrations used are given in table
1.
szR/2
17
-------
3.1.1.6 Chemistry
In all cases, the standard EKMA chemistry was employed, but the
reactivity and aldehyde fractions were taken from the LIRAQ input files.
The effects of considering a different chemical mechanism for evaluating a
control strategy are addressed in chapter 5.
3.1.2 San Francisco Case Studies
The six trajectories developed for the San Francisco region are
illustrated in figures l(a) through 6(a). In each case the final destina-
tion was chosen to coincide with a monitoring station. Figures l(b)
through 6(b) illustrate the results of the six trajectory model runs for
this region. The HC emission profiles used in the OZIPM computer code are
also shown, along with observed ozone "Values and those computed by the
LIRAQ model.
The main emphasis in this study was applied to the 26 July 1973
trajectory shown in figure 1. This trajectory was chosen because it
represented the trajectory to the highest observed ozone concentration for
the prototype meteorology day modeled in the LIRAQ study. According to
our analysis of available wind data, the air parcel that eventually
affects the Livermore area originates at 0800 PDT over the ocean just
south of the city of San Francisco. Between 0900 and about 1000 hours,
the trajectory passes over the San Francisco peninsula where emissions
from four major freeways and San Francisco International Airport are added
to the emissions from the industrial area of South San Francisco. The air
parcel then passes over the bay and veers south along the east shore where
freeways and industrial emissions from the Oakland area are entrained and,
finally, moves east through a canyon into the Livermore Valley. As
documented in chapter 4, the model results are extremely sensitive to wind
speed because of the precursor emissions injected along the trajectory.
The simulation results for trajectory 2, 26 July 1973, are shown in
figure 2. There is a dip in the simulated ozone profile at 1600 PDT
because of a large and sudden increase in mixing height (163 to 700 m in
one hour). One complication in comparing LIRAQ and the OZIPM stems from
the analytical function used in the LIRAQ model to simulate vertical
gradients in concentration. Evidently, the analytic function used in the
LIRAQ model to account for vertical mixing allows the simulated ground
concentrations to respond more slowly to rapid changes in the mixing depth
than is the case in the OZIPM model, which assumes instantaneous mixing
throughout the mixed layer.
32R/2
18
-------
Rl'C
PSE
•IP.:".
570 £90
UTM Coordinate
(a) Trajectory Path
FIGURE 1. TRAJECTORY 1 FOR 26 JULY 1973
650
32R 3
'19
-------
(uidd)
32R 3
20
-------
4300
4290
42BD
4270
4260
4250
4240
4230
4220
0)
re 4210
c
"2 4200
0
" 4190
= 4160
4170
4160
4150
5140
4130
5120
4110
4100
£TC
—
-
-
-
-
C
J.riBB
" \
" /N
- ,/^v
—
-
-
-
I !
0 510
-FPL
• RTC
4PJTT
'f'SE
•IV5N -tDIi"
DFB
rRTl
+ rl^SL)
r-;:-N -tr?s
1300
i^
rue
PDT IsFH ' ^
X **.\
-KDFM
530
550
570 590 610
UTM Coordinate
(a) Trajectory Path
FIGURE 2. TRAJECTORY 2 FOR 26 JULY 1973
»DDL
1 __j L_
630 650
32R 3
21
-------
32R 3
O)
cc
c
O
a;
U
c
o
o
22
-------
FPL
••PSE
570 590 610
UTM Coordinate
630
650
(a) Trajectory Path
FIGURE 3. TRAJECTORY 1 FOR 20 AUGUST 1973
32R 3
23
-------
32R 3
24
-------
REI!
r-SE
p>rr
•S4200
54190
[90
510
530 550
570 590
UTM Coordinate
»DGL
J |_
610 630
650
(a) Trajectory Path
FIGURE 4. TRAJECTORY 2 FOR 20 AUGUST 1973
32R 3
25
-------
OJ
c
o
-------
0)
4-»
(O
s-
o
O
FPL
••PS£
109JJ),
^K'.
510 530 550 570 590 610 630 650
UTM Coordinate
(a) Trajectory Path
FIGURE 5. TRAJECTORY 1 FOR 24 JULY 1974
32R 3
27
-------
to
OJ
CCL
re
3
I— -r- (/)
T3
0)
=3
O
O
O
UJ
a:
32R 3
28
-------
FBL
RMT
'PSE
HfiCT
610 630
650
UTM Coordinate
(a) Trajectory Path
FIGURE 6. TRAJECTORY 2 FOR 24 JULY 1974
32R 3
29
-------
o
o
o
I/O
CO
o
(UJdd)
9UOZQ
O
CVJ
•
O
o
•
C\f
o
«—I
•
o
o
o
o
o
vo
o
o
in
o
o
O >i
O (O
CO O
in
O)
o:
(O
o ,±
I I
O)
•o
o
o
o
vo
o
o
CM
o
o
o
(aidd)
32R 3
30
-------
The simulation results for trajectory 1, 20 August 1973, are shown in
figure 3 and are similar to the results for the 26 July 1973 trajectory
shown in figure 1 in that the injection of precursors is strongly depen-
dent on the choice of wind speed. The sparseness of wind-speed data in
the LIRAQ files and the difficulty of accurately interpreting the LIRAQ
wind fields introduce large uncertainties as to the path and timing of the
trajectory. The sensitivity to wind speed is discussed in section 4.
The results for trajectory 2, 20 August 1973, shown in figure 4,
demonstrate a close agreement between the LIRAQ and OZIPM models and the
observed ozone profile.
The trajectories of 24 July 1974, which are shown in figures 5 and 6,
cannot be compared for the LIRAQ and OZIPM models because the results of
the former are not available. The trajectory illustrated in figure 5
shows a drop in the ozone at 1600 PDT due to a sudden increase in the
mixing height (96 to 242 m in one hour as is the case for trajectory 2 of
26^July 1973). The most important change, which would increase the
simulated ozone for the trajectory results shown in figure 6, would be a
path slightly more to the west than the one shown. The emissions density
on either side of the OZIPM model is quite different from that of the
trajectory path, because the path determined from the LIRAQ wind fields
follows the edge of the bay. This type of "path" sensitivity is also
demonstrated in section 4 for the 26 July 1973 trajectory shown in figure
1.
3.1.3 Summary of Results for San Francisco Case Studies
Table 2 summarizes the results of the case studies just described.
The series of six trajectories for the San Francisco region represents a
severe test for a simple type of model such as that employed in OZIPM.
Special care was required in using the options built into the model
because of the terrain features, wind patterns, and varied emissions
densities found and because of the lack of dependence on initial condi-
tions. As discussed later in chapter 4, several factors were involved in
preparing the input data, which also required careful analysis. Neverthe-
less, the input data selected appear to provide a reasonable level of
success when the OZIPM results are compared with either observed data or
the LIRAQ results.
Primary attention was given to trajectory 1 of 26 July 1973; thus,
the poor results obtained for trajectory 2 of 24 July 1974 might be
partially due to the failure to consider refinements to the obviously
sensitive input data. However, the lack of a LIRAQ simulation for that
day eliminated a principal standard of comparison and, therefore, our
efforts were concentrated elsewhere.
32R/2
31
-------
TABLE 2. SUMMARY OF RESULTS FOR THE SAN FRANCISCO CASE STUDIES
Date
26 July 1973
20 August 1973
24 July 1974
Trajectory
No.
1
2
1
2
1
2
Observed
0.18
0.10
0.08
0.20
0.20
Maximum 03
(ppm)
OZIPM-Predicted
0.20
0.06
0.12
0.09
0.19
0.12
LIRAQ-Predicted
0.19
0.08
0.04
0.07
NA*
NA*
NA = not applicable
32
-------
When examining the results shown in table 1, it should be noted that
instantaneous values are given for the models, but an hourly average value
is given for the observed data. However, as noted earlier and illustrated
in the trajectory model results shown in the figures, ozone tends to vary
rather slowly near the peaks. The results shown in the figures should
also be compared for timing effects. For instance, figure 5(b) shows that
the OZIPM-simulated ozone peak occurs an hour before the observed peak.
The terrain near the peak could involve some convergent wind flow, which
would raise the mixing height without diluting it; thus, the mixing height
used in OZIPM for that time (1600 to 1700 hours) might have been inapprop-
riate. This trajectory does, in fact, follow a path similar to that of
the 26 July 1973 trajectory 1 path discussed earlier. The results
presented in figure 6(b) for trajectory 2 of 24 July 1974 also show that
the timing and peak value might be much closer to that of the observed
ozone level if convergent flow effects had been involved during the
previous two hours. However, again, this possibility was not pursued.
3.2 SACRAMENTO APPLICATIONS
Air quality modeling of the Sacramento region was performed using a
sophisticated three-dimensional grid model (the SAI Airshed Model) as
described by Reynolds et al. (1979). In that study two 1976 ozone-episode
days were modeled (28 June and 24 August). Detailed descriptions of the
inputs to the Airshed Model were presented in that study. In this study,
we compared the results of Reynolds et al. with the OZIPM/CBM trajectory
model results and those of the regular city-specific OZIPP.
Figure 7 shows the Sacramento modeling region and the available moni-
toring stations. The modeling region is 50 km on each side and is divided
into a 625-grid square with each grid cell 2 km on a side. The modeling
region consists of four vertical levels. The first three levels are kept
within the mixed layer until around noon.
In this study, we compared the results of Reynolds et al. with those
of a series of less complex models. If we arrange these models on the
basis of model complexity, with a high degree of commonality among the
models, comparisons of results between adjacent models can be made in a
way that minimizes the number of model differences involved. A similarity
in the results of adjacent models would indicate that for the particular
case studied, the differences in complexity were not important.
By model complexity we mean the spatial and temporal detail with which
the modeling is executed and also the detail with which the physical and
chemical processes are represented.
32R/2
33
-------
LEGEND
* AIR MONITORING SITES
• SURFACE WIND MEASUREMENTS
• VERTICAL TEMPERATURE SOUNDING
VERTICAL HIND SOUNDING
SACRAMENTO
METROPOLITAN
AIRPORT
McCLELLAN
LI AIR FORCE BASE ~ ~~"
HATHER AIR
FORCE BASE
A AIR MONITORIHG SITES
ROSEVILLE
NORTHGATC
SACTO APCD
CALIF ARB
HEADOUVIEU
RANCHO SECO
FIGURE 7. LOCATIONS OF MONITORING SITES IN THE SACRAMENTO AREA
32R 3
34
-------
Correspondingly, a large difference in results would apparently indicate
that the features uniquely addressed by the more complex model were
significant for that case.
The discussion that follows first describes the models that were
applied and highlights important differences between them. Details are
then presented for two case studies performed for the Sacramento area.
The final section of this chapter summarizes the results of these studies.
3.2.1 Model Definition
The following list briefly describes the series of four models used
in this study. They are described in greater detail in the discussions
that compare them. Some models are distinguished only by the mode of
application.
> The SAI Urban Airshed Model. This model is similar in
concept to the LIRAQ model applied to San Francisco.
> The SAI trajectory model. In terms of inputs, computa-
tions, and results at points along a selected trajectory
this model differs from the Airshed Model only through
the elimination of horizontal dispersion and convergent
and divergent winds.
> An EKMA-style model (OZIPM/CBM), which uses the same
chemistry (Carbon-Bond) and mixing heights as the SAI
trajectory model. (In this case, the model is the same
as the OZIPM model described for the San Francisco study,
except that the regular EKMA chemistry has been replaced
with the Carbon-Bond chemistry.)
> The regular city-specific OZIPP model incorporating the
standard EKMA chemistry.
In this section we wish to highlight the major differences among the
models.
32R/2
35
-------
3.2.1.1 Comparison of the SAI Grid and Trajectory Models
These models were designed to complement each other and, therefore,
many algorithms are identical for both models (several subroutines of the
computer codes are actually shared); furthermore, they use the same input
files. The trajectory model begins on a square of the grid model at a
specific time. The path of the trajectory is then determined solely by
the wind speed and direction in the lowest level of the grid-model
files. Because the other levels in the grid model often advect with
velocities that are different from this lowest level, the two models often
generate different results as the simulations progress in time. Neverthe-
less, the trajectory model provides an inexpensive method for testing
parts of the grid model.
A special feature of the trajectory model is its ability to be
operated backward in time. This feature was used to generate the tra-
jectories used in this study. The time and location of either the
observed or the simulated ozone maximum were used to start the backward
trajectory to 0600 hours. Thus, the forward trajectories always arrived
at the desired spot at the proper time for comparison with either observa-
tions or grid-model results. Changes in emissions often lead to changes
in the time and location of maximum ozone concentrations. Small changes
in times of maximum ozone concentration at some specific monitoring site
can be associated with rather large differences in the location of the
origin, and in the actual pathway taken by, the air parcel producing the
specific ozone maximum. Therefore, when compared with a grid model, any
trajectory model used as a tool for control strategies possesses the
inherent shortcoming of being incapable of accounting for these changes.
3.2.1.2 Comparison of the SAI Trajectory Model and the OZIPM/CBM Model
The following similarities and differences exist between the tra-
jectory model and the OZIPM/CBM developed for this comparison:
> Chemistry. Both models used in this study employ Carbon-
Bond chemistry identical to that of the Airshed Model.
However, the photolytic constants are computed differently
in the OZIPM/CBM and SAI models. For this study, we
modified the regular OZIPM/CBM values for N02 photolysis
using a constant factor so that all models agreed at 1300
PDT. In the Airshed Model and trajectory model, the rates
Detailed discussions of this comparison are available elsewhere (see
Reynolds et al., 1979).
32R/2
36
-------
of all other photolytic reactions vary with N02 photoly-
sis, with a fixed ratio for each; but in the OZIPM/CBM the
photolysis rates vary independently. Aldehyde photolysis
was adjusted in the OZIPM/CBM to give agreement between
the models at 1100 PDT.
Actually, these differences in photolysis rates constitute
an advantage in sophistication for the OZIPM/CBM. The N02
photolysis rates normally used in the OZIPM/CBM are some-
what lower than those used in the SAI models because the
latter formerly used an algorithm based on older data.
Incidentally, the SAI models are now being updated to the
OZIPM/CBM N02 photolysis rates, which are those recom-
mended by Demerjian, Schere, and Peterson (1980). Since
the purpose of the present study was to compare the
OZIPM/CBM with previously computed results of the Airshed
Model, we were forced to alter the OZIPM/CBM models to
eliminate this difference rather than wait for the older
SAI studies to be rerun with the newer photolysis rates
for N02.
The use of fixed ratios to N02 photolysis in the SAI
models in place of the variable ratios used in the
OZIPM/CBM model may lead to some minor differences in
chemistry. The ratios are at a maximum at solar noon and
a minimum at sunrise and sunset (Whitten, Killus, and
Hogo, 1980). The photolysis rates for aldehydes change
more with solar zenith angle than the N02 photolysis rate
does because aldehydes photolyze at the short wavelength
end of the solar spectrum, whereas N02 photolyzes nearer
the long wavelength end of the ultraviolet spectrum. The
short ultraviolet wavelengths are more affected by the
ozone in the stratosphere than the longer wavelengths. At
high zenith angles (morning and evening), the solar rays
must pass through a thicker layer of ozone. Therefore,
the ratio of aldehyde photolysis to N02 photolysis varies
throughout the day, reaching a maximum at solar noon. For
this particular study, the OZIPM/CBM aldehyde photolysis
values were multiplied by a constant value throughout each
day so that the models agreed at 1100 PDT. Therefore, in
the early morning the photolysis rate of the aldehydes in
the OZIPM/CBM would be somewhat less than that used in the
SAI models.
For this particular study, we chose to modify the OZIPM/-
CBM to make it as much like the Airshed Model and trajec-
tory model as possible. By removing all possible
32R/2
37
-------
potential variables between models, we hoped to be able to
better compare the fundamentals underlying these models.
The CBM version of OZIPM/CBM has normally been applied to
assist users of the Airshed Model to estimate potential
control scenarios. For such applications the photolysis
adjustments discussed here are normal. Perhaps for this
study an additional series of tests with OZIPM/CBM using
the regular EKMA photolysis rates would have been helpful,
but they were not performed during that phase of this
contract.
> Reactivity. The SAI trajectory model and Airshed Model
are operated with reactivity splits that vary with each
source category. However, the OZIPM/CBM does not have
this feature; instead, reactivity is assumed to be
constant for all hydrocarbon emissions throughout the
day. For this study, the OZIPM computer code was modified
so that three different hydrocarbon reactivities could be
used: one for the initial conditions, one for the aged
hydrocarbons aloft, and one for the emissions that occur
throughout the day. Hence, the only differences in reac-
tivity were in the temporal variations aloft and in the
variable emissions used in the SAI models compared with
the fixed reactivities employed in the emissions for the
OZIPM models. The reactivities used in the OZIPM/CBM were
determined from averages of emissions used in the tra-
jectory model; for averaging the hydrocarbons aloft, a
weighting factor was used each hour that was equal to the
positive change in mixing height.
In a sense, the N02/NOX ratio also acts as a reactivity
effect. The OZIPM/CBM uses a fixed ratio of N02/NOX (10
percent) for all NOX emissions, whereas the SAI models
allow this ratio to vary according to their source cate-
gory (the average ratio is typically about 5 percent). At
this time, we have not attempted either to change the
OZIPM/CBM or to fix the ratio in the SAI model, because we
feel the difference is trivial. For NOX aloft, the
OZIPM/CBM uses pure N02, which is close to typical values
found in the trajectory model simulations because the
N02/NOX ratio is almost always greater than 0.85; often,
the values are greater than 0.99.
> Integration Scheme. The OZIPM/CBM uses a Gear-type method
without any steady-state approximations, whereas the
trajectory model uses the same Crank-Nicholson finite
32R/2
38
-------
differencing scheme and steady-state approximations that
are used in the Airshed Model. Again, this difference in
model sophistication is opposite from the direction
intended by our series of models. However, our previous
tests using smog chamber experiments suggest that the dif-
ferences introduced by the numerical integration schemes
should be minimal.
> Vertical Layers. The Airshed Model and trajectory model
share the same vertical layers and algorithms for numeri-
cal integration, but additional contributions are intro-
duced into the Airshed Model for convergent or divergent
winds. The OZIPM/CBM, of course, is limited to providing
calculations of the mixed layer as a whole, but fixed
concentrations in the layer aloft can be introduced as the
mixing height rises. The SAI models account for mixing
within the mixed layer, mixing within the inversion layer,
and eddy diffusion among all layers. The OZIPM/CBM
assumes instantaneous mixing within the entire mixed
layer, and no eddy diffusion is considered across the
boundary defined by the mixing height.
In our comparison studies, the effectiveness of multi-
layers and eddy diffusion was one of the central issues to
be evaluated in this region of our model series. A
problem that we tried to avoid, which does confound the
comparison, is the entrainment of pollutants from aloft.
The OZIPM/CBM requires fixed concentrations aloft, so
these must be chosen from some "proper" average of the
time-dependent concentrations computed in the SAI trajec-
tory model. With the average value used in this study we
attempted to weight most heavily the concentrations
entrained when the mixing height rises most signifi-
cantly. A more complex weighting scheme might involve the
time when the greatest effect on the chemistry occurs;
however, this effect was not evaluated. Concentrations
aloft are not considered for the time before the inversion
rises or after it has reached some maximum. As the mixing
lid drops late in the afternoon, the only process affect-
ing the surface layer from concentrations aloft is eddy
diffusion. Eddy diffusion is treated in the Airshed Model
and trajectory model, but is neglected in the OZIPM/CBM.
> Mixing height. All models used in the OZIPM/CBM that are
based on the OZIPM computer code, as well as the Airshed
Model and trajectory model, can use identical mixing
32R/2
39
-------
heights with values that vary linearly in time between
specified values at each hour.
> Temperature. The temperature varies .in the Airshed Model
and trajectory model, but the OZIPM/CBM currently employs
a fixed temperature. This difference can be signifi-
cant. Temperature affects both vertical diffusion and
chemistry in the SAI models, but it can affect only
chemistry in the OZIPM/CBM because vertical mixing is
assumed to occur instantaneously within the mixed layer.
> Emissions. The Airshed Model and trajectory model allow
for emissions into all layers; the OZIPM/CBM treats all
emissions in the mixing layer alike and does not directly
address emissions above the mixing layer. For the layer
aloft in the OZIPM/CBM, the emissions must be accounted
for in the determination of the fixed concentrations
aloft. The sum of the emissions used in the lower layers
of the trajectory model was used for the mixing layer in
the OZIPM-CBM. For this study, the OZIPM computer code
was modified to treat emissions in an absolute fashion
rather than as a multiple of the initial conditions. This
was implemented by using the transported surface layer
inputs as the initial concentrations. The normal initial
concentrations (the CALC mode in OZIPM) were specified as
1.0 for both HC and NOX, but these values were not added
as initial concentrations. They were used only for
emission fraction development because emissions are
expressed relative to these values in the OZIPM code.
> Emissions for the SAI models are given as total moles
emitted into each cell during each hour. To arrive at
values appropriate for the OZIPM/CBM we used the following
conversion procedure: the mixing volumes of the trajec-
tory model and the OZIPM/CBM were assumed to be ^equal
" below the mixing height, such that
V = A x Zmix ,
where A is the area of the horizontal cross-section of the
air parcel defined in trajectory model simulations, and
Zm.jx is the mixing height. If the volume is expressed in
cubic centimeters, the moles emitted in one hour (the
number available from the trajectory model computer
output) can be divided by this volume to give moles-cm
32R/2
40
-------
units. This, in turn, can readily be converted to the
OZIPM-CBM units of concentration, ppm, by using a conver-
sion factor of 2.445 x 10^ ppm moles'^ cm^, which is
technically correct for a perfect gas at 298 K and 1
atmosphere of pressure. The numbers calculated at this
point are almost directly applicable to the OZIPM/CBM
model. First, the use of 1.0 for the initial concentra-
tions of HC and NOX in the CALC mode implies that each
one-hour emissions number would be the concentration added
during the hour if the mixing height were the same as the
initial value ZQ. However, the computer codes are written
so that, at each instant in time, the emissions factor is
multiplied by ZQ/Zmix, Therefore, the Zmix drops out and
the actual numbers (En) used are merely
En = A x 2.445 x 1010/Z0
where A and ZQ are expressed in centimeters.
In this particular study, it was occasionally necessary to
use special emissions factors. These factors were used
for those cases in which the trajectory model employed
substantial initial conditions for the first two layers
(up to 100 m) even though the mixing height was only 50
m. Since there was typically an order-of-magnitude
difference in concentration between the initial conditions
below 100 m and the layers above this level, there was no
way to compute a reasonable average value to be used in
the OZIPM/CBM for the layer aloft. Therefore, the initial
conditions between 50 and 100 m used in the SAI models
were treated as emissions in the OZIPM/CBM. These
"emissions" were added to the normal emission during the
hour when the mixed layer in the SAI models jumped from 50
to 100 m. For these cases the appropriate OZIPM/CBM
emissions factors merely add to the normal emissions,
since the initial mixing height is 50 m and the height
added is also 50 m.
3.2.1.3 Comparison of the OZIPM/CBM with the Standard City-Specific
EKMA (QZIPP)
The obvious intention of this comparison was to evaluate only the
differences between the Carbon-Bond chemistry used in the SAI models and
the standard propylene/butane chemistry used in OZIPP. Although the OZIPM
computer program was specifically written to handle such comparisons, some
32R/2
41
-------
complications were introduced when the OZIPP was operated with the pro-
pylene/butane chemistry. The Carbon-Bond chemistry used in this study
assumes that formaldehyde and ketones represent half of the total alde-
hydes found in the simulation, whereas the formaldehyde in the city-
specific EKMA is only 40 percent of the total amount of aldehydes. The
photolysis rates used in the Airshed Model runs are based on a theoretical
solar radiation function that does not contain the updated NC^ absorption
cross-sections. The city-specific EKMA simulations use the latest NC>2
cross-sections, which are approximately 13 percent lower than those used
in the Airshed Model.
The reactivity of the two mechanisms was not typically adjusted in
this study. The total HC concentrations and emissions rates were identi-
cal for these two models, but the standard EKMA propylene/butane chemistry
was typically used with the standard 25 percent propylene, 75 percent
butane, and 5 percent aldehyde reactivity.
Another difference between the OZIPM/CBM and the regular city-
specific EKMA is the use of variable mixing-height inputs in the former,
more sophisticated, version. In the OZIPM/CBM, instantaneous mixing
heights are computed by linear interpolations between hourly input values
rather than by the OZIPP method, which uses an exponential interpolation
between the morning and afternoon mixing heights. The exponential
increase in mixing height with time provides a constant dilution factor.
In addition to the potential problems associated with the assumed shape of
the mixing-height time profile, the standard OZIPP cannot consider a
decrease in mixing height that may follow an increase, nor can it consider
a resumption in mixing-height growth following a decrease or a pause.
In terms of modeling complexity, these models can be classified
according to the levels of analysis described in the 1980 Federal Register
notice on ozone modeling:
> SAI Airshed Model-Level I analysis using a complex grid
model.
> SAI trajectory model-Level I analysis using a complex
trajectory model.
> OZIPM/CBM-Level II analysis using a simplified trajectory
model.
> City-Specific EKMA-Level III analysis.
32R/2
42
-------
3.2.2 Sacramento Case Studies
As previously described, the SAI Airshed Model was applied in
Sacramento for two days--28 June and 24 August 1976. The SAI trajectory
model was run for each day to establish the air parcel trajectory leading
to each ozone maximum. Next, the results of the SAI trajectory model run
were then used to develop the necessary input for the OZIPM/CBM (i.e.,
Level II) model. Finally, portions of the results from this model
simulation were used to develop some inputs for the last model in the
series—the city-specific EKMA. Each case study is described below.
3.2.2.1 Trajectory Calculations for 24 August 1976
The peak observed ozone level for 24 August 1976 was 0.13 ppm, and
the SAI Airshed Model also predicted this level. The meteorology for that
date consisted of relatively uncomplicated wind fields (i.e., low wind
shear and little divergence). The SAI trajectory model was first run to
determine the trajectory path to the maximum ozone predicted by the SAI
Airshed Model; and the calculated trajectory is shown in figure 8. The
detailed meteorological and emissions data used as input for the SAI
trajectory model calculations are discussed in Reynolds et al. (1979).
For this trajectory, the model predicted a peak ozone level (i.e., maximum
one-hour average) of 0.13 ppm, which, as previously noted, agreed with
both the observed level and with the prediction of the more complex SAI
Airshed Model.
The SAI trajectory model simulation served as the basis for develop-
ing inputs for the OZIPM/CBM model simulation. From the SAI trajectory
model output, we obtained the emission rates and meteorology (i.e., mixing
heights) for each hour along the trajectory path. These data, used to
derive inputs to the OZIPM/CBM, are summarized in table 3. The emission
fractions, based on a column area of 4 km , were derived in accordance
with the procedure described in the previous section. The initial
conditions used in the OZIPM/CBM simulation are summarized in the top
portion of table 4, along with the hydrocarbon reactivities. (Note that
in this table, pollutants transported in the surface layer actually refer
to initial concentrations, as was described in the previous section.) For
this particular trajectory, the OZIPM/CBM predicted a peak ozone concen-
tration of 0.14 ppm, slightly higher than that predicted by the two more
complex models. Table 5 shows that the NOX, PAR, ARO, and ozone concen-
tration profiles for the SAI trajectory model and the OZIPM/CBM agree
fairly well. This agreement implies that such factors as multilevel
mixing, eddy diffusion, and elevated emissions are not particularly
critical in this application.
32R/2
43
-------
I t C t * D
A AIR N3NITOR1NG SITES
• SURFAC£ WIND MEASUREHNTS
• VERTICAL TEMPERATURE SOUNDING
1- VERTICAL HIND SOUNDING
SACRAMENTO
METROPOLITAN
AIRPORT
HcCLELLAN
AIR FORCE
HATHER AIR
FORCE BASE
A AIR MONITORING SITES
ROSEVILLE
NORTHGATC
SACTO APCD
CALIF ARB
HEADOWVIEW
RANCHO SECO
Times (PDT)
FIGURE 8. TRAJECTORY USED FOR 24 AUGUST 1976
32R 3
44
-------
TABLE 3. OZIPM/CBM EMISSIONS AND MIXING-HEIGHT INPUTS
FOR 24 AUGUST 1976
Mixing
Heights
(meters)
50
50
100
150
225
300
650
800
800
800
800
800
Taken from the SAI trajectory model calculation.
Time
(PDT)
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
Emi
NOX
(ppm)
0.
0.00198
0.03287
0.144
0.12225
0.0194
0.0074
0.0287
0.0089
0.01267
0.00709
0.01112
ssions
HC
(PPmC)
0.
0.0031
0.2008
0.904
0.556
0.157
0.123
0.0574
0.038
0.0463
0.0336
0.0766
45
-------
TABLE 4. INITIAL CONDITIONS FOR CALCULATIONS OF 24 AUGUST 1976
(a) OZIPM/CBM
Transported
Pollutants
Surface*
Aloft
N02/NOX
0.518
1.000
NOX
(ppm)
0.0386
0.00263
HC
(ppmC)
0.1594
0.0552
03
(ppm)
0.0017
0.0259
Hydrocarbon Reactivities:
OLE
PAR
ARO
CARB
Emissions
(percent)
of total)
3.8
67.2
25.8
3.2
Surface
(percent)
of total)
1.6
60.2
22.09
16.06
Aloft
(percent)
of total)
9.0
60.66
25.7
10.48
(b) City-Specific EKMA
->- Species
Transported
Pollutants
Surface'
Aloft
NOX
(ppm)
0.0372
0.0026
HC
(ppmC)
0.1881
0.0552
°3
(ppm)
0.0095
0.026
N02/NOX = 0.572.
Standard reactivity.
Taken from the SAI trajectory model calculations.
Transported pollutants in the surface layer refer to total
initial conditions (i.e., local generation plus background
plus any possible transported levels).
32R/m
46
-------
TABLE 5. COMPARISON OF THE SAI TRAJECTORY MODEL AND THE OZIPM/CBM CALCULATIONS
FOR 24 AUGUST 1976 FOR N0x, PAR, ARO, AND OZONE
Time
(PDT)
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
NU
Trajectory
0.039
0.040
0.037
0.068
0.071
0.053
0.033
0.024
0.016
0.015
0.013
0.013
X
DZIPM/CBM
0.039
0.041
0.037
0.071
0.071
0.052
0.023
0.017
0.015
0.013
0.012
0.011
KA
Trajectory
0.096
0.097
0.143
0.285
0.283
0.234
0.173
0.139
0.113
0.113
0.112
0.114
K
OZIPM/CBM
0.096
0.098
0.133
0.299
0.293
0.242
0.135
0.114
0.112
0.112
0.111
0.112
AK
Trajectory
0.018
0.018
0.026
0.055
0.053
0.041
0.028
0.020
0.015
0.013
0.012
0.011
0
OZIPM/CBM
0.018
0.018
0.025
0.056
0.052
0.040
0.020
0.015
0.013
0.011
0.010
0.009
U
Trajectory
0.002
0.006
0.010
0.011
0.020
0.041
0.061
0.078
0.089
0.108
0.121
0.129
3
OZIPM/CBM
0.002
0.00004
0.009
0.013
0.027
0.053
0.063
0.078
0.100
0.117
0.130
0.139
32R/14
47
-------
The results of the OZIPM/CBM simulation were used to provide inputs
for the city-specific OZIPP simulation. The lower portion of table 4
shows the initial conditions used in the simulation. City-specific OZIPP
calculations require an initial and final mixing height. These values
were 50 m and 800 m, and the inversion height lasted from 0800 to 1300
PDT. For OZIPP calculations we used standard reactivity (i.e., 25 percent
propylene) for the hydrocarbon emissions splits and initial conditions.
Figure 9 shows the results of the OZIPM/CBM calculation, including the
station observations and the OZIPM/CBM predictions. The comparison
between ozone observations and the OZIPM/CBM calculations shows that they
are similar in temporal behavior. The ozone predicted by the city-
specific OZIPP model was lower than the observed ozone concentration by
about 7 percent.
3.2.2.2 Trajectory Calculations for 28 June 1976
OZIPM/CBM calculations for the 24 August 1976 trajectory just
discussed were fairly straightforward because of the lack of wind shear.
In contrast, the winds for 28 June 1976 contained high wind shears. The
results and even the actual "paths" to ozone maximums can be quite
different for the Airshed Model and trajectory model if wind shear or any
horizontal dispersion occurs in the Airshed Model. The trajectory
calculations for 28 June attempt to address the problems of wind shear and
horizontal dispersion because these factors can be important in compari-
sons of trajectory and grid modeling results.
We used the trajectory model to generate a back-trajectory from the
point of maximum ozone. Even though this trajectory model employs the
same number and size of the vertical layers as the Airshed Model, the
computer algorithm that is used to run the back-trajectories follows only
the surface winds of the grid model. The wind shear within the mixing
layer was significant for this day, and ignoring the winds in the second
layer generated a back-trajectory that would have been inappropriate since
the surface winds did not pass over any major emissions. Because good
mixing typically occurs within the mixed layer, we modified the computer
algorithm of the trajectory model to average the winds within the mixed
layer and used the average value instead of the surface wind value only.
This produced the trajectory shown in figure 10, which led to an ozone
prediction of 0.15 ppm at 1700 PDT.
The maximum observed ozone of 0.16 ppm occurred near Northgate at
1500 PDT; the Airshed Model predicted a maximum of 0.15 ppm at 1600 PDT
between Northgate and Roseville. Since the trajectory model predicted an
ozone peak of 0.15 ppm at 1700 PDT, an hour later than the prediction of
the Airshed Model, we could not readily compare the EKMA models (OZIPM/CBM
32R/2
48
-------
o
o
o
o
^ i
o
o
o
°0
§
o
o
0
o
o
0
0
o
0
o
o
o
§
o
o
o
o
o
o
0
o
* ^?
o
o
0
o
c/o °
g o
K^H ®
•SI— Q
^ ~ 4 §
_,, iix o
Q- O- Ci
g ^ §o
z: ^ §
5! UJ O
< o ®
o
g
o
Q
o !
e » * n a w —
r N e co >o « o
• • * • * • •
s
o
, O
• co
o
• 0
r—
o
g
o
0
in
o
o
1 —
o
o ^-«
oo i—
r— Q
0-
o;
E
O -r-
O 1—
00
r—
O
0
O
o
o
0
o
o*^
o
o
§
o
)
^
s:
UJ
o
u_
»— t
0
[ ! )
D.
oo
1
1 —
1 — 1
C£.
O
u.
oo
O
I-H
O
Q
UJ
o:
D-
_J
UJ
^^1
o
2:
Q
"^
^^
oo
•z.
0
I—H
I—
<
UJ
oo
CO
o
zz
o
1__
^^
h- O
00 h-
•z.
•ZL UJ
UJ S
K* O
LU eC
CQ 00
1
1 1 \ \
O P*-.
O i—
U. 1—
O 00
o •=>
oo ca:
o: <^-
ca: oo
D-
0 0
O U-
CTl
1 1 1
en
o
ONOKU C.M&
32R 3
49
-------
AIR KWlITORlNG SITCS
SURFACE WIND HEASUREfCNTS
VERTICAL TEMPERATURE SOUNDING
VERTICAL HIND SOUNDING
TIME = PDT
SACRAMENTO
METROPOLITAN
AIRPORT
NcCUELLAN
AIRVORCE BASE
3
0900
1000
HATHER AIR
FORC£ BASE
AIR MONITORING SITES
MSEVILIE
MORTHGAT[
SACTO APCD
CALIF ARB
KADOUVIEU
RANCMO SECO
FIGURE 10. TRAJECTORY USED FOR 28 JUNE 1976
32R 3
50
-------
or city-specific) with the airshed predictions. Instead, we compared the
predictions of the OZIPP-type models to the trajectory model predictions
only.
From the trajectory model results, we obtained the emissions rates
and mixing height information needed to perform the OZIPM/CBM calculations
as discussed for the 24 August 1976 calculations. Table 6 shows the
emissions rates and mixing heights used in the OZIPM/CBM calculations.
The top portion of table 7 shows the initial conditions used in the
OZIPM/CBM calculations. The OZIPM/CBM simulation resulted in a peak ozone
concentration of 0.16 ppm, which, like the 24 August simulation, was
slightly higher than the SAI trajectory model predicted.
As for the cases previously discussed, the inputs for the city-
specific EKMA were derived from the OZIPM/CBM simulation. To provide a
direct comparison between the models, 0800 PDT OZIPM/CBM conditions were
taken as the initial conditions for the city-specific OZIPP simulation.
Figure 11 shows the results of the city-specific OZIPP calculation
for ozone. The observed ozone at the monitoring stations along (or near)
the trajectory is also shown in this figure. Unfortunately, the trajec-
tory passed over only a few stations, and these do not record the ozone
maximum; therefore, we were unable to compare the observed ozone maximum
with the model-predicted ozone maximum. The city-specific OZIPP simula-
tion was carried out for 10 hours, from 0800 PDT to 1800 PDT, whereas the
trajectory model ran from 0600 PDT to 1700 PDT. The trajectory model
predicted an ozone level of 0.15 ppm at 1700 PDT. The OZIPM/CBM, a model
that is more general (i.e., it employs a variable mechanism, variable
mixing heights, etc.), predicted an ozone level of 0.167 ppm at 1700
PDT. The city-specific EKMA (standard OZIPP) predicted an ozone of 0.13
at 1800 PDT.
3.2.3 Summary of Results for Sacramento Case Studies
Table 8 summarizes the results of the case studies just described.
The important conclusion reached in this part of the study is that the
simple OZIPP-type trajectory model can produce simulations that agree very
closely with the simulations of highly sophisticated models such as the
SAI Airshed Model and the SAI trajectory model. Indeed, the present study
involved the use of extremely detailed input data common to all models;
yet, the main differences in results were due to the differences in
chemistry between the OZIPM/CBM and the OZIPP model with the regular
propylene/butane chemical mechanism.
32R/2
51
-------
TABLE 6. EMISSION RATES AND MIXING HEIGHTS USED ]N THE
OZIPM/CBM CALCULATIONS FOR 28 JUNE 1976*
Mixing
Heights
(meters)
150
150
220
250
280
310
370
460
600
800
800
800
Taken from SAI trajectory model calculations.
Emissions
Time
(PDT)
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
NOX
(ppm/hr)
0.
0.00067
0.00217
0.01171
0.01466
0.02545
0.02618
0.0333
0.04198
0.03253
0.03584
0.02019
HC
(ppmC/hr)
0.
0.0011
0.008
0.0388
0.07805
0.1173
0.10678
0.09757
0.20355
0.13898
0.10951
0.04486
32R/14 52
-------
TABLE 7. INITIAL CONDITIONS USED IN THE CITY-SPECIFIC EKMA
CALCULATIONS OF 28 JUNE 1976*
(a) OZIPM/CBM
Transported
Pollutants
Surface
Aloft
N02/NOX
0.845
1.000
NOX
(ppm)
0.0457
0.0025
HC
(ppmC)
0.2899
0.0541
°3
(ppm)
0.0019
0.649
Hydrocarbon Reactlvlties:
OLE
PAR
ARO
CARB
Emissions
(percent
of total)
4.0
67.5
26.0
2.5
Surface
(percent
of total )
2.6
48.6
13.0
35.7
Aloft
(percent
of total )
0.9
63.8
25.8
9.6
(b) City-Specific EKMA
Transported
Pollutants
Surface
Aloft
NOX
(ppm)
0.0319
0.002
Species
HC
(ppmC)
0.218
0.054
°3
(ppm)
0.035
0.065
N02/NOX = 0.773
Standard reactivity.
Taken from the SAI trajectory model calculations.
53
-------
•
•
•
•
•
et
H-
O
O
..,
fy'
UJ
CO
o
*
(•—"- +
J-
»
mt
»
O
o
o
§0
o
o
0
o
o
o
§
o
0
o
o
0
o
o
o
0
o
0
<
00
UJ
3
^£
^>
o
•
t^
o
0
UJ
C£
0.
o
•m-i + MM + MM.!.
C5 —
^ *^
o ©
H
•
•
o
§
8
o
8
o
o
0 H
o
o
8
o
o
O H
0
o
o
o
o
o
0
o
8
o
o
•
o
o
0
^°0
o
— O , tt C5 <
O> >£> •* CJ C
© © © © c
©*©'©'©«
k S
o
00
o
. ^2
fill"
o
k 0
UD
"~
, 0
0
Ln
'
0
• ™jT
r~
0
o
i- co
r~
o
• s
.§
o
• 0
0
o
•5?
0
o
' §
^
?
»
^^
1—
Q
0.
N_X
_E
h—
I
I
UD
r>-
en
00
CM
CO
•z.
o
I— t
t—
o
LU
o:
D-
Q
o
co
o
co
CO
o
GO
CO
UJ
o
tvl
o
2:0
Oh-
CLQ:
2:0
0<
o:
rs
CJ3
ONO
32R 3
54
-------
oo
§
CO
•a:
c\j
o
vo
r^.
CTi
> X
ro CM
t—* f—(
O O
UJ
ro
•-D
CO
OJ
2:
o
oo
1
1 1 1
1.LJ
o
o
f—
•y*
UJ
a:
6) tO O
§ OJ Q-
t- Q. -^^
0
-
*~ 1
ro
z o
CD
O E ^-
^ 1 1
a. •>- n.
«— i X • —
hsl ro
o z:
vc
^^
C3
o
*j . — .
-^ (—
0) «3 D
E O) Q-
•^ Q. ^
O
o
f^
»— 4
0
oo
CTi
n
t-H
•
0
§
r-*
«-H
>>
s-
O
O Q)
0) T5
El
DJ
•"-> O -^ Dj
"O S X ^3
CTi
CVJ
oo
z:
o
O
O
VD
O
O
o:
Q.
ro
O
IT) PO
t—I «-H
O O
.
O LU
.
o oo
00
•o o
o
E c
*^- o
o -IH
io m
»—* t—t
<3 C3
CO
UJ
_l
CO
O) t£>
4-> r-^
-------
Whereas the complicated emissions data constituted the unique feature
of the San Francisco study, the highlight of the Sacramento study is the
complex wind shear of the 28 June 1976 simulations. The simple averaging
of wind data for the two cells within the mixing layer simulated in the
SAI grid model seemed to generate reasonable ozone levels in the trajec-
tory models. Therefore, the studies of the simple trajectory models
conducted to date appear to show that if the precursor levels of HC and
NOX were similar in either the complex or simple models, the chemical
mechanism, especially if it's the same mechanism, tends to simulate
similar levels of ozone. Although this conclusion seems obvious, obtain-
ing similar precursor levels for the complex and simple models is by no
means straightforward.
The present study involved the use of detailed computer outputs from
the SAI Airshed Model and the trajectory model. These outputs served as
sources of input, standards of performance, and guides to the treatment of
inputs. Of course, typical applications of the EKMA would not be made in
such an environment. However, the indications of this study are that
careful application of a simple model such as the OZIPP-type trajectory
model can lead to reasonable ozone simulations. If the local situation
indicates complex problems such as diverse emissions densities or wind
shear, extra care is warranted. In the following section we explore the
sensitivity to some of the situations previously discussed; in section 5
we use isopleths to examine the relative applications of EKMA.
32R/2
56
-------
SENSITIVITY STUDIES
In this evaluation, the emphasis on sensitivity is related to
performance in an absolute sense rather than in the relative sense that is
used in the control-strategy application of the EKMA. Therefore, we are
concerned with the absolute ozone values simulated by the simple OZIPP
trajectory model. In the San Francisco region sensitivity to OZIPP
parameters and inputs was emphasized, whereas in the Sacramento study
(discussed later in this chapter) sensitivity to differences between
models was emphasized.
We did not use a systematic approach to selecting the input para-
meters of OZIPM; instead, we emphasized the regional aspects of the cities
studied. For example, the EKMA procedure for predicting control require-
ments necessitates a knowledge of the initial (0600 to 0900) HC/NOX
ratio. Yet, the application of the model to the San Francisco region
indicates that the initial HC/NOX ratio, or more important, the initial
concentrations of HC and NOX, do not affect the simulated maximum ozone
values. Hence, the use of observed HC/NOX ratios may be questionable for
the isopleth application of the EKMA to San Francisco, even though the
trajectory model used in the approach can provide reasonable simulations
of episode ozone levels in the area. The conclusion that observed HC/NOX
data for San Francisco are questionable hinges on the accuracy of the
trajectories determined for this study, the central issue being the origin
of all trajectories in the Pacific Ocean at 0800. When some trajectories
starting in the city at 0800 were investigated, the combination of
observed wind velocities and low HC/NOX ratios in both the observations
and the emissions inventory produced ozone peaks late in the day far
downwind of the region. These trajectories had passed the monitoring
sites before the occurrence of the observed maximum ozone concentra-
tions. Since there were no monitoring sites with which to test the model,
the trajectories were not studied further. Trajectories that originate in
the urban core and arrive at the monitoring site at the proper time do not
necessarily coincide with trajectories derived from surface wind measure-
ments. However, it must be emphasized that significant uncertainties are
associated with the latter, especially in areas of complex wind flow.
32R/2
57
-------
4.1 THE SAN FRANCISCO REGION
4.1.1 Initial Conditions
The main inputs to the OZIPP or the OZIPM that control the ozone
development for the San Francisco trajectories are the emissions and
mixing-height profiles. According to our analysis, all trajectories that
lead to high ozone formation originate over the Pacific Ocean. To
illustrate the lack of sensitivity to initial conditions, the trajectory
to Livermore was simulated using a wide range of initial concentrations
and ratios for HC and NOX from 26 July 1973. The results are shown in
table 9. The emissions profiles were recomputed for each calculation so
that the absolute level of precursor concentrations caused by emissions
was identical for all the cases. Similar insensitivity to initial
conditions would be seen in the other San Francisco trajectories since
they all originate over the ocean. This is not a case for generally
ignoring initial conditions in OZIPP models, because other cities do not
often experience the wind velocities and clean upwind air found in San
Francisco. The point we wish to emphasize here is that the OZIPP-type
model can still give reasonable results in spite of this lack of sensiti-
vity to initial conditions.
4.1.2 Mixing Heights
The mixing-height profiles, typically available directly from the
LIRAQ files, were investigated for sensitivity in two ways. First, the
necessity of using the OZIPM option of hourly mixing-height data was
explored relative to using the normal city-specific EKMA version in OZIPP,
in which only initial and final mixing heights are used. This area of
sensitivity proved to be irrelevant for this study, because the change in
mixing height with time could, in general, be closely simulated by the
proper choice of initial and final mixing heights for the standard OZIPP
input parameters. In other studies, such as those of Whitten and Hogo
(1981), trajectories have been considered in which the shape of the mixing
height with time could not be adequately simulated using the standard
OZIPP parameters. The second method involved the concept of mixing height
itself (discussed in detail in the previous chapter). The LIRAQ simula-
tion of 26 July 1973 was used extensively by the BAAQMD for the AQMP. The
main trajectory studied here, on the basis of the meteorological data
found in the LIRAQ files, arrived at the site of maximum observed ozone--
Livermore—on that date. Because of the significance of this trajectory,
special treatment of the OZIPM input data for mixing heights seemed
appropriate. However, determination of the "proper" mixing height data
was not pursued beyond the following sensitivity study. Several mixing
32R/2
58
-------
TABLE 9. VALUES FOR MAXIMUM OZONE VERSUS VALUES FOR INITIAL CONDITIONS*
NOX HC Maximum
(ppm) (ppmC) Ozone
0.003 0.03 0.197
0.003 0.01 0.196
0.003 0.003 0.194
0.001 0.03 0.198
0.001 0.01 0.196
0.001 0.003 0.195
0.0003 0.03 0.197
0.0003 0.01 - 0.195
0.0003 0.003 0.195
* 26 July 1973 trajectory No. 1.
59
-------
heights were adjusted for the later stages of the day when the air parcel
passed through a canyon area into the Livermore valley where the ozone
maximum was observed. Results using the different mixing-height profiles
or vectors are shown in figure 12.
4.1.3 Emissions Vector (Wind Speed)
Sensitivity to mixing-height input data is rather straightforward in
the OZIPM model. Precursor concentrations are reduced linearly as the
mixing heights are increased relative to the initial mixing height. Ozone
is reduced, but less than linearly, with dilution (Fox, Kamens, and
Jeffries, 1975). In some cases, a large concentration of ozone aloft can
actually lead to increases in ozone as the mixing height rises.
Sensitivity to the emissions-profile data actually involves input
preparation factors such as wind speed and model size. The wind-speed
factor was of paramount importance for the San Francisco trajectories
because this area is characterized by sharply contrasting emissions den-
sities. The dominant westerly winds typically bring clean air into the
area from the Pacific Ocean. The city itself lies on the northern end of
a peninsula and is thus surrounded by water on three sides. The peninsula
is a heavily populated suburban area. Across San Francisco Bay, to the
east of the city, are the large urban and suburban areas of Oakland. Just
to the east of this populated area are hills. Hence, a typical air parcel
in this area is subjected to sharply contrasting emissions levels. If a
low wind speed is associated with the air parcel, the time spent over
areas of high emissions is much longer.
For areas such as Los Angeles, there is much less contrast in emis-
sions levels along a trajectory path once the air parcel reaches land.
Wind speed would be less critical for an air parcel passing over a large
area of uniform emissions density. Hence, wind speed in the San Francisco
region is much more important than it is in the Los Angeles region.
This sensitivity to wind speed is demonstrated in the 26 July 1973
trajectory to Livermore by using twice the reported speed during the
morning hours. Figure 13 shows that a significant change is noted in the
ozone profiles for different assumed wind speeds.
4.1.4 Model Size
The grid squares of the LIRAQ model were of 5-km size, the value that
we used to define the narrowest width of the OZIPM trajectory model.
Since the path determined by our analysis of the LIRAQ wind data did not
32R/2
60
-------
O
CM
rv. LO
p^. CD
CTi CM
II II
CD CD CD
LU LU LU
Z Z Z
•ZL -ZL -Z.
o o o
»—I *—( I—t
o: ix o:
LU LU LU
> > >
o
o
o
o
ID
O
o
in
o
o
O
IVI
O
o
o
O
CL.
O 1X3
O O
O)
O
o
CM
O
O
O
o
o
CD
I—" >-
LU o:
z o
Z CJ
O LU
a:
i CO
U_ CTi
O i—
t/) LU
LU "
U-
LU CM
CM
CXL
^>
CD
o
I—(
•
o
O
o
01
o
(iudd)
32R 3
61
-------
O
O
\
\
\
\
\
\
O
O
LD
\
\
\
\
•»
»
•*
O
O
co
\
\
O
O
o:
D-
o
l^sl
O
O
O
\
\
N
O
Q_
O
O
C/5
O
o:
o
el
CL
OO
in
•
CO
O
O
in
i/o
O
tn i—<
O 3
u- o
oo u
»— z
UJCV
GO
cc.
UJ
CO
O
O
o
o
o
in
32R 3
o
*»•
t
o
o
PO
o
CM
•
O
o
^H
•
O
(iudd)
62
-------
generally follow the regular grid pattern of the LIRAQ model, we deter-
mined the emission information for the OZIPM inputs by including adjacent
grid-square information when the trajectory path width overlapped the
LIRAQ grid squares. The adjacent information was averaged using simple
linear weighting.
Although the adjacent grid trajectories produce large differences in
ozone concentrations, the 15-km trajectory produces results similar to
those for the central 5-km trajectory, as shown in figure 14. The size of
San Francisco—roughly 15 km—tends to place an upper limit on the nominal
size of a reasonable model. Also for this sensitivity analysis, the paths
of the trajectories were "forced" to follow the central, or standard,
path. Actually, an analysis of the wind patterns indicates that the path
to the north of the standard path turned toward the north and crossed San
Francisco Bay; thus, the high urban emissions did not affect the Livermore
valley. Therefore, this sensitivity analysis involves more than a simple
change in width: the high value of 0.48 ppm ozone predicted for the
adjacent trajectory to the north reflects a sensitivity to path definition
as well.
4.2 THE SACRAMENTO REGION
The SAI Airshed Model and trajectory model had previously been
applied to this area and, therefore, detailed inputs and outputs were
readily available and familiar to us. This access enabled us to conduct a
comparison study emphasizing individual differences between the SAI models
and OZIPP-type models. A similar study was recently completed for the Los
Angeles area by Whitten and Hogo (1981). The present evaluation involved
a series of four models that varied in complexity from the SAI Airshed
Model to the regular city-specific EKMA trajectory model found in the
OZIPP computer code. Two were OZIPP-type models, one was the regular
OZIPP model, and the fourth was based on the OZIPM version and was further
divided as follows:
> The city-specific model employing variable, instead of
constant, dilution rates. The OZIPM computer code was
needed for this exercise.
> The OZIPM model using variable dilution rates and the same
Carbon-Bond chemistry used in the Sacramento airshed
modeling study (i.e., OZIPM/CBM).
> An OZIPP-type model that we have called OZIPM/CBM. Along
with the variable dilution and CBM chemistry additions,
the starting time is changed to be early enough in the
32R/2
63
-------
\ \\
o o
>-
10
CM
UJ
O
g
a:
o
o
UJ
C3
'•udd)
auozo
64
-------
morning to maximize the influence of the emissions
inventory. In the regular, city-specific OZIPP, the 0800
starting time is intended to maximize the influence of the
0600 to 0900 precursor concentrations.
The model series was intended to demonstrate the sensitivity of
OZIPP-type models to three factors: dilution, chemistry, and starting
time. Although different starting times were used in practice, the
comparisons of this study remove such a difference. The model employing a
starting time of 0800 PDT used initial conditions that were generated at
0800 by the model that was started earlier. The model that had been
started earlier could, therefore, conceivably be restarted at 0800 to
produce the same result produced by the continuous model version. The
differences between models due to the dilution factor (variable versus
constant) were also unimportant in this study because the shapes of the
variable dilution curves could, in general, be closely approximated by
using a constant dilution rate. For instance, a calculation was made for
24 August 1976 in Sacramento using the city-specific inputs with both
variable dilution and constant dilution rates, but no difference in ozone
predictions occurred between the two simulations.
As indicated in chapter 3 of this report, though starting time and
variable dilution factors did not show any important sensitivity for the
Sacramento case, the change in chemical mechanism did have a noticeable
effect. More ozone was generated by the CBM chemistry than by the
standard EKMA chemistry, an effect that is consistent with the results
reported by Whitten and Hogo (1981). That study further demonstrated that
the standard EKMA chemistry could be easily modified to significantly
increase the calculated peak ozone. The only real option available in the
standard OZIPP model with which to enhance ozone is reactivity, since the
other options are limited by observations. The computer code was also
slightly modified to accept higher photolysis rates for aldehydes. Both
of the adjustments (reactivity and higher photolysis rates) merely
accelerated the production rate of simulated ozone during the morning
hours (which was already higher for the standard EKMA chemistry than for
the CBM chemistry), but the final ozone peak was not significantly
enhanced. No attempts to modify the CBM chemistry were made because all
models (the grid, trajectory, and OZIPM/CBM models) using it produced
results whose differences (when they occurred) could be explained by
factors other than chemistry.
An important aspect of the differences in chemistry is found in the
control requirements predicted by the two chemical mechanisms. This
aspect is discussed in the next section, which details EKMA applica-
tions. Absolute differences in simulated peak ozone between the mecha-
nisms are often less important when they are used with the relative method
of EKMA.
32R/2
65
-------
EKMA APPLICATIONS
When the EKMA is applied to an urban area, isopleth diagrams,
observed HC/NOX ratios, and observed ozone maximums are used to estimate
the emissions control requirements that are necessary to achieve compli-
ance with the ozone standard. The isopleth diagrams for use in EKMA
applications are constructed by interpolating between multiple simulations
of a specific trajectory model. Normally, the ordinate and abscissa of
the isopleth diagrams represent the initial conditions of HC and NOX con-
centrations used in the trajectory model simulations. The EKMA isopleth
lines represent constant maximum one-hour ozone simulated by the tra-
jectory model from the ordinate and abscissa value for initial condi-
tions. These diagrams are constructed by the OZIPP and OZIPM computer
programs.
In previous sections of this report, we discussed versions of the
OZIPP trajectory model as it was applied in the absolute sense. That is,
we were concerned with the absolute value of ozone predicted by the model
from given initial precursors, emissions, and meteorological factors.
When the trajectory model is used via the isopleth diagram, the model is
used in a relative fashion. The starting point on the diagram is the
intersection of the ozone isopleth line corresponding to the observed
ozone maximum with the radial line corresponding to the observed HC/NOX
ratio for 0600 to 0900 hours. The ordinate and abscissa values are
intended to relate to percentage changes in the HC and NOX emissions
inventories.
In this section we discuss the uncertainties associated with the
construction and use of the EKMA isopleth diagrams. Whereas we previously
compared (in an absolute sense) individual trajectory model simulations
with those of various other models and with observations, in this section
we compare (in a relative sense) the changes in ozone predicted by various
models on the basis of percentage changes in emissions.
32R/2
66
-------
5.1 CONSIDERATIONS RELATED TO THE USE OF ISOPLETH DIAGRAMS
5.1.1 Fixed Monitoring Sites
The time and location of maximum observed ozone concentrations can
change as a result of a change in emissions only. Grid models typically
show such effects. The probability of the coincidence of monitoring sites
and the actual location of maximum ozone concentrations can either
increase or decrease as a result of changes in emissions. The proper
deployment of several monitoring sites and the application of numerous
trajectory paths can, to some extent, mitigate this problem. The use of a
grid model is also helpful.
Another related problem occurs when winds vary in direction during
the course of a day because the air arriving at a monitoring site can
originate at several different locations. If the proposed control
strategy alters the time of occurrence of the ozone maximum, the base
trajectory utilized to construct the isopleth diagram then becomes
inappropriate for the monitoring site. When the wind directions are
constant throughout the day, a change in ozone peak timing merely leng-
thens or shortens the path to the monitoring site.
5.1.2 Uneven Changes in Emissions
Mobile-source emissions are, for the most part, regulated at the
federal level, whereas stationary sources are more often subject to local
control. Therefore, because the federal controls may be different from
the local requirements and because some areas are dominated by mobile
sources and others stationary sources, the uniform control of either HC or
NOX is rarely even in time and space. Individual trajectory-model
simulations might provide acceptable levels of accuracy for the base-case
year and the future year because these models can typically alter the
timewise distribution of emissions input data. However, the isopleth
approach requires some uniform emission pattern for all points on the
diagram.
Depending on the level of sophistication, some trajectory models can
also account for changes in hydrocarbon reactivity along the trajectory
path. However, in constructing isopleth diagrams, considering changes in
both emissions patterns and reactivity is beyond the scope of the present
algorithms.
32R/2
67
-------
5.1.3 Transported Pollutants
The regular city-specific EKMA isopleth algorithm uses the same
amounts of transported pollutants for all points of the diagram. In
practice, the regulation of mobile emissions at the federal level would
modify the transportation of pollutants. It was shown by Whitten and Hogo
(1981) that on the basis of the EKMA, significant differences in control
requirements can result from different assumptions about the transported
pollutants in the surface layer and the layer aloft. The latest guide-
lines for using the EKMA suggest employing two isopleth diagrams to deal
with this problem.
5.2 SAN FRANCISCO APPLICATIONS
The AQMP (1979), using the LIRAQ model, offers a unique means of
evaluating the EKMA. The control-strategy LIRAQ predictions for various
scenarios can be directly contrasted with comparable EKMA predictions. We
have also included brief discussions of two other topics involving the use
of EKMA in the San Francisco region:
> Isopleths prepared from some of the widely differing
initial conditions discussed in section 4, which demon-
strated a lack of sensitivity to initial conditions (see
table 9).
> A comparison of isopleths using the CBM and the standard
EKMA chemical mechanisms.
> A comparison of isopleths generated assuming different
trajectories.
5.2.1 Basic Comparisons between LIRAQ and EKMA
Several control scenarios were described in the AQMP (1979). For
cases in which comparisons could be made with the EKMA, we constructed
isopleth diagrams and applied various estimates of predicted ozone
concentrations using the EKMA technique. In this section, attention is
focused on those EKMA applications:
> The OZIPM trajectory analysis described in chapter 3
(i.e., a Level II analysis).
> A basic city-specific EKMA analysis using current guide-
lines for a Level III analysis (EPA, 1980).
32R/2
68
-------
A basic city-specific EKMA analysis in which the Carbon-
Bond Mechanism was substituted for the standard
propylene/butane chemistry contained in OZIPP.
Table 10 shows the comparative results for the different scenarios applied
in the LIRAQ analysis. A discussion of these results follows.
For the LIRAQ evaluations, 26 June 1973 was selected as the prototype
meteorological day. The trajectory corresponding to the prototype day was
shown earlier in figure 1. The OZIPM computer code was used to construct
the isopleth shown in figure 15 since the starting times and hourly mixing
height inputs were not the standard ones assumed for the city-specific
EKMA. The results indicate that the Level II predictions are generally
less sensitive to changes in hydrocarbons and/or NOX than are the LIRAQ
predictions. Because of the fundamental differences in the two models,
explanations for these differences cannot be postulated. However, the
sensitivity of Level II predictions is addressed in the next section.
The isopleth approach is, in effect, confined to one trajectory path,
whereas the grid models cover essentially all possible trajectories in
space and time. Thus, the worst case found over the entire gridded area
would normally be found in the grid model rather than (for this case) in
three trajectory paths. This reasoning leads to some surprising compari-
sons of LIRAQ (grid) and EKMA (trajectory) results. For 20, 40, and 60
percent HC reduction scenarios, plus the scenarios combined using reduc-
tions of 40 percent HC and 20 percent NOX, the ozone predictions from
LIRAQ are lower than those from the EKMA. The remaining three scenarios
(80 percent HC reduction, and 40 percent NOX reduction) reflect a situa-
tion in which the worst case over the gridded area is higher than, or
equal to, the worst case along the specific trajectory.
Two additional isopleth diagrams for 26 July 1973 were constructed to
reflect the current guidelines for a city-specific EKMA analysis (i.e., a
Level III analysis). In this approach, a trajectory is followed from the
urban core to the site of the observed peak ozone level (EPA, 1980). The
isopleth diagram shown in figure 16(a) was constructed using the city-
specific EKMA program (i.e., OZIPP), which reflects the standard
propylene/butane chemistry contained in city-specific EKMA. Figure 16(b)
was constructed using the Carbon-Bond Mechanism rather than the standard
chemistry. The inputs for both diagrams, which are summarized in table
11, correspond to conditions for 26 July 1973.
The HC/NOX ratio for the Level III applications was derived from
0600-to-0900-hour-averaged HC and NOX observations taken at the downtown
San Francisco monitoring station (MacCracken and Sauter, 1975). The
monitoring station reported total hydrocarbons only, but methane
32R/2
69
-------
TABLE 10. CONTROL-STRATEGY COMPARISONS FOR SAN FRANCISCO BETWEEN LIRAQ AND EKMA APPLICATIONS
Reduction
(percent)
HC
Models Studied N0x
0
0
20
0
40
0
60
0
80
0
40
20
80
40
0
40
LIRAQ predictions
0.189 0.140 0.082 0.069 0.055 0.119 0.064 0.2]
(-26%) (-5758) (-63%) (-7158) (-37%) (-66%} (+23
OZIPM (Level II)
(based on 0.189 max.)
0.189 0.161 0.145 0.102 0.036 0.141 0.062 0.1£
(-15%) (-23%) (-46%) (-81%) (-25%) (-67%) (-15
Level III EKMA (based on
0.189 max.)
0.189 0.130 0.080 <0.080 <0.080 0.10 0.08 0.2-5
(-31%) (-57%) «-57%) «-57%) (-47%) (-57%) (+31
Level III EKMA using CBM
(based on 0.189 max.)
0.189 0.112 <0.08 <0.08 <0.08 0.08 <0.08 0.31
(-4155) «-57%) «-57%) «-57%) (-57%) «-57%) (+66
The predicted ozone level is shown for each control scenario in addition to the percentage change from the
base-case level (i.e., no HC or NOX reductions).
32R/m
70
-------
II
o
car
a:.
o
LLJ
ce:
(—
cr>
10
CVJ
D:
o
O
oo
in
32R 3
(iudd)
71
-------
99'0
co
r-
>-
_J
ZD
O
^O
C\J
I
o
5^
LU
O
o u~>
d>
CL OL
00 O
I U-
&- C_)
(O LJ-I
I— o
o:
^>
o
rro 00-0°
Hdd'XCM
72
-------
LL'O
99 •"£)
SS'O
WO ££'0
Wdd'X0N
ZZ'Q
CO
c
o
QJ
c
o
CD
I
c
o
ro
o
o
d)
ex
>*,
*->
•i—
o
tro
00-0°
•D
0)
o
o
CJ
32R 3
73
-------
concentrations were usually between 1.3 and 1.9 ppmC. For the average
0600-to-0900-NMHC concentration, we assumed that the methane concentration
was 1.6 ppmC. The calculated design HC/NOX ratio for San Francisco is
3.3. This low value is similar to the low values estimated by de Mandel
et al. (1979), who reported in their study that typical HC/NOX ratios were
between 1.2 and 1.7 for the five worst ozone episode days.
The emission rates presented in table 11 were taken from ARB
(1976). However, information concerning hydrocarbon emissions reactivity
is needed to utilize the Carbon-Bond Mechanism. Even though this infor-
mation was not readily available, we decided to use the same reactivity as
that of the LIRAQ model. This resulted in an OLE fraction of 18.1
percent, PAR fraction of 80.4 percent, and CARB fraction of 1.6 percent.
Since the LIRAQ model does not contain any aromatics, we set the ARO
fraction to zero. Without aromatics the hydrocarbon reactivity value used
in this study may not reflect the true urban emissions mixture, nor does
it reflect a typical reactivity for the use of the Carbon-Bond Mechanism.
The control strategy comparisons shown in table 10 reveal that the
city-specific EKMA predictions using the standard EKMA chemistry most
nearly parallel those obtained using the LIRAQ model. When the Carbon-
Bond Mechanism is used, the sensitivity of ozone predictions to changes in
emissions is greater.
5.2.2 Further Investigations
In this section, we describe the results of some additional analyses
performed to assess the sensitivity of control strategy predictions using
the EKMA. First, the sensitivity to the HC/NOX ratio under particular
conditions is examined. Second, the sensitivity of control estimates to
particular trajectory path conditions is explored. Finally, an alterna-
tive method for generating isopleth diagrams and evaluating control
strategies is presented.
In section 4.1.1, we indicated that the predicted peak ozone levels
for trajectory 1, 26 July 1973, were insensitive to the assumed initial
conditions. Likewise, the control predictions for this special case
should be insensitive to the HC/NOX ratio. As described earlier, the
emissions input vectors were constructed from the LIRAQ emissions inven-
tory. For the sensitivity study concerning the insignificance of the
initial conditions (see table 9), we altered the emissions vectors to
maintain constant contributions from the emissions regardless of the
initial values or ratios of HC and NOX. We also demonstrated that entire
isopleth diagrams used with widely varying design HC/NOX ratios can be
32R/2
74
-------
TABLE 11. INITIAL CONDITIONS USED IN THE LEVEL III EKMA CALCULATIONS
FOR 26 JULY 1973 IN THE SAN FRANCISCO REGION
(a) Initial Conditions
Species
NOX
NMOC
Oo
Transported Pollutants
Surface Aloft
0.0 ppm
0.0 ppmC
0.0 ppm
0.0 ppm
0.0 ppmC
0.04 ppm
N02/NOX = 0.25.
Hydrocarbon reactivity: standard OZIPP conditions,
Morning mixing height: 150 m at 0800 PDT.
Afternoon mixing height: 977 m at 1600 PDT.
0600 to 0900 hours HC/NOX = 3.3.
0600 to 0900 hours NMOC = 0.4 ppmC.
0600 to 0900 hours NOX = 0.12 ppm.
Design 03 = 0.19 ppm at 1600 PDT near Livermore.
County
San Francisco
Alameda
(b) County Emissions
ROC
(tons/day)
97.7
187
NOX
(tons/day)
83.0
131.0
Area
(km2)
116.55
1898.5
Density
(kg/hr km2)
ROC NOX
31.69 26.92
3.72 2.61
(c) Emissions Fraction
Time
(PDT)
0800-0900
0900-1000
1000-1100
1100-1200
1200-1300
1300-1400
1400-1500
1500-1600
Trajectory
San Francisco/Alameda
Alameda
Alameda
Alameda
Alameda
Alameda
Alameda
Alameda
ROC Fractions
0.496
0.104
0.104
0.104
0.104
0.104
0.104
0.104
NOY Fractions
0.434
0.0767
0.0767
0.0767
0.0767
0.0767
0.0767
0.0767
32R/14
75
-------
constructed and can produce consistent results if the HC/NOX ratio
employed is the one upon which the emissions vector was constructed. Four
key points result from the investigation of this special trajectory:
> The EKMA can still be used as a control strategy tool,
provided that the special emissions input vectors and
design HC/NOX ratios are consistent.
> The observed HC/NOX ratio is not needed.
> The underlying effective HC/NOX ratio is found in the
emissions inventory.
> The isopleth-derived control implications are directly
related to the emissions themselves rather than to the
results of typical applications (where the initial
conditions of precursors dominate the model).
Two additional isopleth diagrams were constructed to illustrate the
technique developed for trajectories (such as those we found in San
Francisco) that are not sensitive to initial conditions. The base-case
isopleth diagram chosen was the one shown earlier in figure 15 in which
the initial condition HC/NOX ratio used in the base trajectory (i.e., the
26 July 1973 trajectory originally shown in figure 1) is 10/1. As
previously described, the design ozone value must be located on this
HC/NOX ratio line when this technique is used. The diagrams are shown in
figures 17 and 18. The initial-condition base HC/NOX ratio lines to be
used for these diagrams are 1/1 and 100/1, respectively. These diagrams
were constructed using the technique described earlier for showing
insensitivity to initial conditions—that is, the emissions vectors were
recomputed so that the contributions of emissions to the model concentra-
tions in the base-case trajectory were all the same. The trajectory model
results were presented earlier in table 9 and were discussed in section
4. The control implications of these additional diagrams, shown in table
12, demonstrate that consistent utilization of emissions inventory
information produces consistent isopleth-derived results, provided that
the proper HC/NOX ratio (i.e., the ratio used to construct the emissions
vector) is utilized with the design 03 when initial conditions are not
important.
Recall from chapter 3 that the OZIPM model (i.e., the Level II
analysis) was applied to several other trajectories in addition to the
basic 26 July 1973 trajectory that corresponds to the prototype conditions
modeled using the LIRAQ. Two additional isopleths were developed for the
trajectories shown in figures 3 and 5: trajectory 1 for 20 August 1973,
and trajectory 2 for 24 July 1974, respectively. The two diagrams are
shown in figures 19 and 20. The LIRAQ control scenarios were evaluated
32R/2 -c
/D
-------
II
o
eC
Qi
O
O
Lul
ra
eC
c:
ro
10
CO
o;
o
UJ
_i
ex.
O
oo
(Uldd)
32R 3
77
-------
o
o
II
o
o
o
«£.
CC
CO
>-
_J
rD
••3
a:
o
D-
o
oo
oo
Ul
D;
o
(uidd)
32R 3
78
-------
TABLE 12. SENSITIVITY OF CONTROL-STRATEGY EVALUATION TO HC/NOX RATIO FOR TRAJECTORY 1—26 JULY 1973
Cases Studied
Reduction
(percent)
HC
NO
X
0
0
20
0
40
0
60
0
80
0
40
20
80
40
0
40
OZIPM (Level II)
(based on 0.189 max.)
trajectory for 26 July 1973;
HC/NOx = 100
City-specific using OZIPM
(based on 0.189 max.) 1975
trajectory for 26 July 1973;
NC/NOx = 1.0
0.189 0.165 0.146 0.105 0.040 0.145 0.063
0.16C
0.189 0.169 0.142 0.098
0.036
0.143 0.060
0.16C
32R/m
79
-------
O
UJ
oo
O
CM
O
u.
:r
O.
O
oo
(uidd)
32R 3
80
-------
CM
O
UJ
*o
<£.
cx:
>-
_]
ra
o
*a-
CM
C£
O
LJ_
a:
I—
UJ
_i
D-
O
00
O
CM
32R 3
(uidd)
81
-------
using the EKMA technique with these diagrams. For these calculations, the
HC/NOX ratio corresponded to that of the assumed initial conditions, but
an ozone maximum of 0.189 ppm was selected to compare more easily the EKMA
predictions based on the different trajectories. Table 13 contains the
control calculations for these two new trajectories, as well as those for
the basic 26 July 1973 trajectory (i.e., trajectory 1). The results
demonstrate that control estimates can be sensitive to trajectory path, as
might be expected. Interestingly, some predictions using these new
trajectories agree more closely with the LIRAQ results than do the
trajectories previously discussed.
In the recent study by Whitten and Hogo (1981), a new type of EKMA
isopleth was introduced that uses the emissions inventories for NOX and HC
as the ordinate and abscissa of the diagram. The (1,1) point of such a
diagram corresponds to the trajectory simulation using the present
emissions inventory. Figure 21 shows the isopleth of 26 July 1973 in
which the (1,1) point is the trajectory shown in figure 1. The HC/NOx
ratio to be used with the new diagram is the diagonal line through the
(1,1) point. Using this HC/NOX line and the 0.189 ppm design, 03~value
ozone predictions were computed using the regular procedures. The results
are shown in table 14.
Note that the design ozone and the simulated ozone at the (1,1) point
do not need to be the same. The new method is used exactly as the
original EKMA except that the HC/NOX ratio line through the (1,1) point is
used in place of the observed 0600-to-0900-hour concentration ratio.
Thus, the new method in effect substitutes the emissions inventory
information for the 0600-to-0900-hour observed data. The similarity of
these results to those of the regular city-specific isopleth shown in
figure 15 is to be expected in this case because the initial conditions
are not important. The intended advantages of using the new type of
isopleth are the elimination of the initial HC/NOX ratio data require-
ment. The initial conditions can be handled in one of two ways as can the
concentrations aloft. In one method the concentrations are rolled back
(or ahead) away from the (1,1) point of the diagram when the isopleth
diagram is constructed. The initial conditions are thus treated as
concentrations transported in the surface layer. In the other method, the
initial conditions are still handled as surface-transported pollutants,
but they are held constant over the entire isopleth diagram. Then,
another diagram is constructed in which the expected future concentrations
are held constant. This is actually the current method for handling
present and future transported pollutants contained in the OZIPP computer
code used in the Level III applications of the EKMA. Thus, the initial
conditions may still be important, and their importance can be treated
using this new method.
32R/2
82
-------
TABLE 13. SENSITIVITY OF CONTROL-STRATEGY EVALUATION TO TRAJECTORY PATH
Cases Studied
Reduction
(percent)
HC
NO
X
0
0
20
0
40
0
60
0
60
0
40
20
80
40
0
40
OZIPM (Level II)
Trajectory 1 for
26 July 1973
City-specific using OZIPM
(based on 0.189 max.) 1975
trajectory for 24 July 1974
City-specific using OZIPM
(based on 0.189 max.) 1975
trajectory for 20 August 1973
Average for 3 OZIPM runs
0.189 0.161 0.145 0.102 0.036 0.141 0.062 0.161
0.189 0.162 0.127 0.070 <0.02 0.135 0.048 0.17;
0.189 0.168 0.128 0.082 0.043 0.132 0.058 0.18!
0.189 0.164 0.133 0.085 0.03 0.136 0.056 0.172
32R/14
83
-------
LU
rD
«C
a:
CO
01
«—i
>-
_i
^5
•-3
o;
o
a.
o
>-
I—
I
CM
(mdd)
32R 3
84
-------
TABLE 14. CONTROL-STRATEGY COMPARISONS USING A NEW-STYLE ISOPLETH
Models Studied
Reduction
(percent)
HC
N0v
X
0
0
20
0
40
0
60
0
80
0
40
20
80
40
0
40
LIRAQ predictions
OZIPM (Level II)
Trajectory 1 for
26 July 1973;
new-style isopleth
OZIPM/CBM (Level II)
Trajectory 1 for
26 July 1973;
new-style isopleth
0.189 0.140 0.082 0.069 0.055 0.119 0.064 0.23!
0.189 0.165 0.125 0.075 >0.02 0.132 0.063 0.16!
0.189 0.161 0.12 0.078 0.038 0.134 0.053 0.172
32R/1H
85
-------
We have also used this new-style diagram with the new Carbon-Bond
Mechanism, which was employed by Whitten and Hogo (1981). The diagram
generated using the trajectory of 26 July 1973 with the CBM chemistry is
presented in figure 22, and the control-scenario results are included in
table 14. Although the CBM chemistry generates more simulated ozone than
the regular EKMA chemistry for the (1,1) point, the control-scenario
predictions (which are based on EKMA-style relative application to a
single-design 03) turn out to be quite similar except for the case of 80
percent HC/40 percent NOX combined control.
5.3 SACRAMENTO APPLICATION
Part of the task of the original SAI Airshed Model study of the
Sacramento region was to examine the effects of controlling mobile-source
emissions (Reynolds et al., 1979). In order to perform the desired con-
trol simulations in the original study, the Sacramento region was rotated
to form a special air quality control region (AQCR) in which the main
highway runs in an east-west direction. Figure 23 shows this special
modeling region. Meteorology for 24 August 1976 was used for the study.
The wind fields for that day were also rotated in such a way that the
predominant wind flow was from west to east, parallel to the main high-
way. The modeling region was expanded to contain 729 grid cells, each
3 km in length. The total region size is 81 km by 81 km. The original
emissions inventory was increased to reflect the dimensions of the new
region and a rotation scheme having the same emissions density as the
original region.
Seven control strategies were defined, and airshed simulations were
performed for these strategies on the rotated region. Table 15 shows the
control strategies that were evaluated and the Airshed Model's predicted
peak ozone concentration for each. Note that several of the control
strategies cannot be easily simulated with the EKMA. Therefore, only the
first two control strategies listed in table 15 (30 percent reduction in
mobile-source HC and NOX emissions, and 30 percent reduction in mobile-
source HC only) were evaluated with the EKMA.
Because of the special rotation of the modeling region, new base-case
evalulations had to be made using the SAI trajectory model, OZIPM/CBM, and
city-specfic EKMA. As was done for the Sacramento simulations described
in chapter 3, the trajectory path leading to the maximum ozone concentra-
tion was defined with the SAI trajectory model (shown in figure 24). From
the trajectory model outputs, inputs for the OZIPM/CBM simulations were
generated, and the inputs for the city-specific EKMA (OZIPP) model were
then generated using the results of the OZIPM/CBM simulations. Tables 16
and 17 summarize the inputs for each model. Table 18 compares the
predicted maximum ozone for the four models.
32R/2
86
-------
oo
t—I
o
CO
o
CQ
o;
et
CD
Z
HH
oo
CJ
UJ
•-D
-------
REGIONAL
AIRPORT
CENTRAL BUSINESS DISTRICT
FIGURE 23. MAP OF THE HYPOTHETICAL URBAN AREA USED
TO STUDY TRANSPORTATION CONTROL STRATEGIES
32R 3
-------
TABLE 15. SAI AIRSHED MODEL PREDICTED PEAK OZONE CONCENTRATIONS COMPUTED
FOR DIFFERENT CONTROL STRATEGIES IN THE HYPOTHETICAL
REGION (base-case predicted peak ozone = 17 pphm)
Peak Ozone
Predicted
Control Strategy Simulation (pphm)
(la) Reduction of all mobile source emissions by 15
30 percent
(Ib) Reduction of mobile emissions of HC by 30 13
percent
(2a) 30 percent reduction in mobile, emissions 17
along a corridor parallel to prevailing wind
(2b) 30 percent reduction in mobile emissions 17
along a corridor perpendicular to prevailing
wind
(2c) 30 percent reduction in mobile emissions in 16
the central business district
(3a) Use of 3-hour running-average emissions 17
(3b) Delaying of all emissions by 3 hours 10
32R/14
89
-------
REGIONAL
AIRPORT
CENTRAL BUSINESS DISTRICT
1QOO 1100
0900\ \ 1200
1700 PDT
FIGURE 24. TRAJECTORY USED FOR THE CONTROL-STRATEGY
RUNS (MODIFIED SACRAMENTO REGION)
32R 3
90
-------
TABLE 16. EMISSION RATES AND MIXING HEIGHTS USED IN
EKMA APPLICATIONS TO HYPOTHETICAL REGION
CONTROL STRATEGIES*
Time
(PDT)
0600
0700
0800
0900
1000
1100
1200
1300
1400
1500
1600
1700
NO
(ppm/hr)
0.0133
0.0641
0.0629
0.06958
0.05804
0.0792
0.08836
0.0798
0.098
0.0677
0.0099
__
HC
(ppmC/hr)
0.039
0.3719
0.4171
0.493
0.2904
0.2239
0.1759
0.258
0.2445
0.1116
0.01728
__
Mixing Heights
(meters)
50
50
100
150
225
300
450
600
800
800
800
800
Taken from the SAI trajectory model calculations.
91
-------
TABLE 17. TRAJECTORY MODEL OUTPUTS USED AS INPUTS FOR THE
EKMA SIMULATIONS
(a) Initial Conditions and Hydrocarbon Reactivities Used in the
OZIPM/CBM Control Base-Case Run (starting time = 0600 PDT)*
Transported Pollutants
Species Surface Aloft
NOX 0.003 ppm 0.003 ppm
HC 0.057 ppmC 0.055 ppmC
03 0.019 ppm 0.026 ppm
Hydrocarbon Reactivities
Species Surface Aloft Emissions
OLE
PAR
ARO
CARB
(b) Initial Conditions Used in City-Specific EKMA Calculations
of Hypothetical Control Runs (starting time = 0800 PDT)
Transported Pollutants
0.014
0.616
0.282
0.088
0.009
0.607
0.257
0.105
0.041
0.698
0.239
0.023
Species Surface Aloft
NOX 0.041 ppm 0.003 ppm
HC 0.248 ppmC 0.055 ppmC
03 0.009 ppm 0.026 ppm
N02/NOX = 0.556.
Propylene reactivity = 0.25.
Dilution: initial mixing height = 100 m;
final mixing height = 800 m.
Taken from the SAI trajectory model calculation,
32R/14
92
-------
TABLE 18. COMPARISON OF MAXIMUM PREDICTED OZONE FOR THE SAI AIRSHED,
SAI TRAJECTORY, OZIPM/CBM, AND CITY-SPECIFIC EKMA MODELS:
BASE-CASE CONDITIONS USED FOR THE CONTROL-STRATEGIES
STUDY OF THE HYPOTHETICAL REGION
Maximum
Model (ppm)
SAI Airshed 0.17
SAI trajectory 0.12
OZIPM/CBM 0.12
City-specific EKMA 0.13
93
-------
The comparison of predicted maximum ozone shows that the trajectory
model and the two EKMA models give similar results, whereas the Airshed
Model prediction is approximately 40 percent higher than that of the
trajectory model. The special "rotated" emissions grid constructed for
the study turned out not to be compatible with the SAI trajectory model
computer code. Several attempts to construct a compatible emissions grid
failed to produce one that reproduced in the trajectory model the con-
centrations that were seen in the original study. Since the precursor
concentrations were low in the trajectory model simulations, low ozone
values were expected. Unfortunately, time and cost constraints precluded
the resolution of this emissions inventory problem.
The two control strategies were evaluated directly with SAI trajec-
tory model simulations reflecting the assumed emission reductions. For
the EKMA applications, isopleths were generated to evaluate the strate-
gies. Figure 25 shows the isopleth generated for the city-specific EKMA
application. The regional, initial HC/NOX ratio of 7.2 is taken from
Reynolds, et al. (1979). The initial HC/NOX ratio, determined from the
initial conditions for this specific trajectory is approximately 6/1.
However, we used the HC/NOX ratio of 7.2 because this value was derived
for the EKMA as it is more normally applied. Since a 30 percent reduction
was made in mobile emissions only, we needed to determine the reduction in
total emissions (sum of mobile and stationary sources). From the Airshed
Model emissions inventory containing the reduced mobile emissions and the
trajectory path, we found the emissions for each hour along the trajectory
path (see table 19). From the new emissions rates, we determined the
level of control of HC and NOX emissions from all sources to be reductions
of 19.6 percent HC and 16.3 percent NOX. Since the SAI Airshed Model and
the SAI trajectory model did not predict similar ozone maximums, we
decided to use the ozone maximum of 0.12 ppm predicted by the trajectory
model as the design ozone value.
The EKMA predictions resulting from application of the two control
strategies are presented in table 20. Controlling of HC mobile emissions
only shows a 20 percent difference in predicted maximum ozone between the
trajectory model and the city-specific EKMA model. This discrepancy
appears to be a result of the different chemistry used in the two models
because the CBM-EKMA simulation, with the same inputs as those of the
city-specific EKMA run, predicted an ozone maximum much closer to that of
the trajectory model.
Further investigations were made using Level III EKMA guidelines.
From the initial conditions presented in table 21, and from emissions
rates of ARB (1979), we generated two diagrams (one using the standard
EKMA chemistry, the other using Carbon-Bond chemistry). For the CBM
isopleth we used the same hydrocarbon reactivity as is presented in table
17. The isopleth diagram generated using the standard EKMA chemistry is
32R/2
94
-------
X
in
c
o
•r—
(S)
vt
OJ
-o
c u
•I- i-
(U
C D.
O
•r- O
4-> CO
o —
3
•o c
a; o
C
O
u
3
O IO -U
4->^ -- > O +J O
U >>•!- (J O -r-
QJ i— X3 a> ni -a
•r-j c a) -"-3 a>
(O O I- IO -E S-
$- Q. S- •!-> Q.
•!-»<_> -MO
ZEcX. -Q ^
I— < ^" I-H y8
ca: q- S <: 4- S
to o LU c/> o tu
CM CO
oo
•z.
o
I—I
g
o
o
C3
UJ
00
I
o
o
CU ^ CO
U CU U
S- S- t-
QJ O- Ol
O- O-
CO T3 XCO
«— O O
0.
o
UJ
Q.
oo
tn
CVJ
o:
Z3
CD
•(iudd) ON
32R 3
95
-------
TABLE 19. CONTROL-STRATEGY STUDY TOTAL EMISSIONS
RESULTING FROM A 30 PERCENT REDUCTION
IN MOBILE HC AND NOX EMISSIONS
NOX HC
(ppm/hr) (ppmC/hr)
0600 1.004 0.0384
0700 0.0463 0.2882
0800 0.0455 0.3076
0900 0.052 0.3808
1000 0.0456 0.2327
1100 0.0694 0.1826
1200 0.0803 0.184
1300 0.0688 0.2097
1400 0.0875 0.1952
1500 0.0622 0.0895
1600 0.0088 0.0163
1700
96
-------
TABLE 20. COMPARISON OF MAXIMUM OZONE PREDICTIONS BETWEEN THE
TRAJECTORY MODEL, OZIPM/CBM, AND CITY-SPECIFIC EKMA
WITH DIFFERENT CONTROL SCENARIOS (design 03 = 0.12 ppm)
Model
Trajectory model
OZIPM/CBM
City-specific EKMA
Level III EKMA
Level III EKMA using CBM I
Level III EKMA using CBM II
Predicted 03—
30 Percent Reduction
in Mobile Emissions
of HC and NO
0.103 (-14%)
0.097 (-19%)
0.108 (-10%)
0.104 (-13%)
0.10 (-17%)
0.101 (-16%)
Predicted 03~-
30 Percent Reduction
in Mobile Emissions
of HC Only
0.09 (-25%)
0.086 (-28%)
0.114 (-5%)
0.105 (-13%)
0.084 (-30%)
0.082 (-32%)
97
-------
TABLE 21. INITIAL CONDITIONS USED IN THE LEVEL III EKMA CALCULATIONS FOR
HYPOTHETICAL REGION FOR 24 AUGUST 1976 METEOROLOGY (emissions
based on Sacramento County for 1976)
(a) Initial Conditions
Species
NOX
NMOC
Oo
Transported Pollutants
Surface Aloft
0.0 ppm
0.0 ppmC
0.0 ppm
0.0 ppm
0.0 ppmC
0.2 ppm
N02/NOX = 0.25.
Hydrocarbon reactivity:
Morning mixing height:
Afternoon mixing height:
standard OZIPP conditions,
150 m at 0800 PDT.
800 m at 1400 PDT.
0600 to 0900 hour HC/NOX =7.2.
0600 to 0900 hour NMOC = 0.248 ppmC.
0600 to 0900 hour NOX = 0.035 ppm.
Design 03 = 0.12 ppm at 1700 PDT.
County
Sacramento
(b) County Emissions
ROC
(tons/day)
119.7
NOX
(tons/day)
70.9
Area
(km2)
2525.24
Density
(kg/hr km2)
NO,
1.29
(c) Emissions Fraction
Time
(PDT)
0800-0900
0900-1000
1000-1100
1100-1200
1200-1300
1300-1400
1400-1500
1500-1600
Trajectory
Sacramento
Sacramento
Sacramento
Sacramento
Sacramento
Sacramento
Sacramento
Sacramento
ROC Fractions
0.081
0.081
0.081
0.081
0.081
0.081
0.081
0.081
NOY Fractions
0.130
0.130
0.130
0.130
0.130
0.130
0.130
0.130
32R/H+
98
-------
shown in figure 26(a); the isopleth diagram with Carbon-Bond chemistry is
shown in figure 26(b). The ozone predictions using the two diagrams are
shown in table 20. For the 30 percent control of mobile emissions of HC,
the Level III CBM-EKMA using both the old and new versions of the Carbon-
Bond Mechanism [figures 26(b) and 26(c] predicts a lower ozone value than
the standard Level III EKMA, as was true for the case that compared the
CBM-EKMA with the city-specific EKMA. Thus, we see that the standard EKMA
chemistry is less responsive to hydrocarbon controls. Yet, when both HC
and NOX emissions are reduced, the standard EKMA models (city-specific and
Level-Ill) and the carbon-based trajectory model and OZIPM/CBM all give
similar results.
32R/2
99
-------
OZ-Q 9fo zro
o
\r>
t_) Ijj
fO
01
cu
o;
CT)
oo
CD
CSJ
CI3
LjJ
a:
C_J
o
a.
31
C£
O
CE:
o
UJ
3
a:
CM
GZ'O
OZ'O
sro zro
Hdd'XBN
90
oo-cP
32R 3
100
-------
zro
oro 80-Q
J _J
ZO'O
Ol'O 80-0 90'0
•a
O)
c
O
O
to
CM
H'O
zro
ZO'O
00-0°
Wdd'XBN
32R 3
101
-------
oro scro go'o wo
zro
oro
80'0 90'0
Wdd'X0N
WO ZO'O OO'O0
32R 3
102
-------
REFERENCES
ARB (1979), "Emissions Inventory, 1976," California State Air Resources
Board, Sacramento, California.
ARB (1976), "Emissions Inventory, 1973," California State Air Resources
Board, Sacramento, California.
BAAQMD (1979), "1979 Bay Area Air Quality Plan," Association of Bay Area
Governments, Bay Area Air Quality Management District, Metropolitan
Transportation Commission, San Francisco, California.
de Mandel, R. E. (1979), "Comparisons of EPA Rollback, Empirical/Kinetic,
and Physicochemical Oxidant Prediction Relationships in the San
Francisco Bay Area," J. Air Pollut. Control Assoc., Vol. 29, No. 4,
pp. 352-358.
Demerjian, K. L., K. L. Schere, and J. T. Peterson (1980), "Theoretical
Estimates: Actinic (Spherically Integrated) Flux and Photolytic Rate
Constants of Atmospheric Species in the Lower Troposphere," in
Advances in Environmental Science and Technology, Vol. 9 (John Wiley
and Sons, New York, New York).
Dimitriades, B. (1977), "An Alternate to the Appendix-J Method for
Calculating Oxidant and N02-Related Control Requirements," Proc:
International Conference on Photochemical Oxidant Pollution and Its
Control, Vol. II, EPA-600/3-77-001b, pp. 871-879, U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina.
Dodge, M. C. (1977), "Combined Use of Modeling Techniques and Smog Chamber
Data to Derive Ozone-Precursor Relationships," Proceedings: Inter-
national Conference on Photochemical Oxidant Pollution and Its
Control, Vol. II, EPA-600/3-77-001b, Environmental Protection Agency,
Research Triangle Park, North Carolina.
EPA (1980), "Guidelines for Use of the City-Specific EKMA in Preparing
Ozone SIPs," EPA-450/80-027, Research Triangle Park, North Carolina.
Federal Register (6 November 1980) 73-696.
32R/2 103
-------
Feldstein, M., et al. (1979), "Anatomy of an Air Quality Maintenance
Plan," J. Air Pollut. Control Assoc., Vol. 29, pp. 339-363.
Fox, D. L., R. Kamens, and H. E. Jeffries (1975), "Photochemical Smog
Systems: Effect of Dilution on Ozone Formation," Science, Vol. 188,
pp. 1113-1114.
MacCracken, M. C., and G. D. Sauter (1975), "Development of an Air
Pollution Model for the San Francisco Bay Area, Vol. 2. Appendixes,"
UCRL-51920, Lawrence Livermore Laboratory, University of California,
Livermore.
Reynolds, S. D., et al. (1979), "Photochemical Modeling of Transportation
Control Strategies—Volume I. Model Development, Performance
Evaluation, and Strategy Assessment," DOT Report FHWA-RD-78-173.
Whitten, G. Z., and H. Hogo (1981), "Evaluation of the EKMA Performance,"
10R-EF80-73R, Systems Applications, Inc., San Rafael, California.
Whitten, G. Z., H. Hogo, and J. P. Killus (1980), "The Carbon-Bond
Mechanism: A Condensed Kinetic Mechanism for Photochemical Smog,"
Environ. Sci. Techno!., Vol. 14, pp. 690-700.
Whitten, G. Z., J. P. Killus, and H. Hogo (1980), "Modeling of Simulated
Photochemical Smog with Kinetic Mechanisms," EPA-600/3-80-028,
Systems Applications, Inc., San Rafael, California.
Whitten, G. Z., and H. Hogo (1978a), "User's Manual for Kinetics Model and
Ozone Isopleth Plotting Package," EPA-600/8-78-014a, U.S. Environ-
mental Protection Agency, Research Triangle Park, North Carolina.
Whitten, G. Z., and H. Hogo (1978b), "User's Manual for Ozone Isopleth
Plotting with Optional Mechanisms (OZIPM)," EF78-30, Systems Applica-
tions, Inc., San Rafael, California.
32R/2
104
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO. 2.
EPA-450/4-81-005a
4. TITLE AND SUBTITLE
APPLICATION OF THE EMPIRICAL KINETIC MODELING APPROACH
TO URBAN AREAS
Volume 1: San Francisco/Sacramento
7. AUTHOR(S)
G. Z. Whitten, H. Hogo, and R. G. Johnson
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Systems Applications, Inc.
950 Northgate Drive
San Rafael, California 94903
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
MDAD, AMTB, MD-14
Research Triangle Park, North Carolina 27711
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
January 1981 (preparation)
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NC
SAI No. 32R-EF80-139
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-3376
13. TYPE OF REPORT AND PERIOD COVEREC
Final
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
EPA Project Officer: Gerald L. Gipson
16 ABSTRACT
The EKMA was evaluated by comparison with more complex models using common data
bases for the San Francisco and Sacramento areas. The basic model used in the EKMA
is a simple, moving-air-parcel, or trajectory, model that uses a detailed chemical
mechanism for the surrogate propylene and butane hydrocarbons. Time-dependent,
precursor-emission factors can be treated, along with expansion of the air parcel,
but entrapment is treated by assuming that constant concentrations exist outside
the parcel. This basic model is used to generate ozone isopleth diagrams for
estimating ozone response to hydrocarbon and nitrogen oxide controls. The
comparison study involved several levels of the EKMA system, from parts of the basic
model to control strategy predictions. For the San Francisco area, the EKMA was
compared with the LIRAQ model, which had been used for the 1979 AQMP. For the
Sacramento area, the EKMA was compared with the SAI Airshed Model. When carefully
applied, the EKMA produced results that typically agreed with the more complex
models. However, the evaluation confirmed that care must be exercised in applying
the model to situations involving wind shear, complex terrain features, and highly
nonuniform emission densities, and that caution must be used in interpreting the
results of model applications to situations involving these factors.
17. KEY WORDS AND DOCUMENT ANALYSIS ~~
a. DESCRIPTORS
Ozone
Control Strategies
Photochemical Pollutants
Models
EKMA
OZIPP
18. DISTRIBUTION STATEMENT
Unlimited
b. IDENTIFIERS/OPEN ENDED TERMS
19. SECURITY CLASS (This Report/
20 SECURITY CLASS (This page)
c. COSATl Field/Group
21. NO. OF PAGES
114
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION is OBSOLETE
105
------- |