EPA-450/2-77-021a
USES, LIMITATIONS AND TECHNICAL
BASIS OF PROCEDURES
FOR QUANTIFYING RELATIONSHIPS
BETWEEN PHOTOCHEMICAL OXIDANTS
AND PRECURSORS
Monitoring and Data Analysis Division
Office of Air Quality Planning and Standards
LIBRARY
U. S. cNVlRONMENT '. PROTECTION AGENCY
EDLSOs. 088.,
ENVIRONMENTAL PROTECTION AGENCY
Office of Air and Waste Management
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
November 1977
-------
This report is issued by the Environmental Protection Agency to report
technical data of interest to a limited number of readers. Copies are
available free of charge to Federal employees, current contractors and
grantees, and nonprofit organizations - in limited quantities - from the
Library Services Office (MD-35), Research Triangle Park, North Carolina
27711; or, for a fee, from the National Technical Information Service,
5285 Port Royal Road, Springfield, Virginia 22161 .
This document has been reviewed by the Office of Air Quality
Planning and Standards, U. S. Environmental Protection Agency, and
approved for publication. Subject to clarification, the contents
reflect current Agency thinking.
Publication No. EPA-450/2-77-021a
-------
Table of Contents
Page
Executive Summary iii
List of Figures ix
List of Tables x
1.0 Background 1
1.1 Introduction 1
1.2 Applicability of Available Procedures 2
1.3 Organization 4
2.0 Methods for Relating Ozone to Precursors 5
2.1 Use of Photochemical Dispersion Models 5
2.2 Non-Data-Intensive Methods for Relating Ozone to 9
Precursors
2.2.1 Use of an Empirical Kinetic Modeling Approach 9
(EKMA)
2.2.la Use of Standard Ozone Isopleths 12
2.2.1b Use of City-Specific Isopleths 28
2.2.2 Use of Linear Rollback 32
2.2.3 Use of Statistical Models 37
3.0 Consideration of Transport and Natural Background 40
3.1 Possible Roles of Transported Ozone/Precursors in the 40
Urban Oxidant Problem
3.1.1 Transport of Precursors 40
3.1.2 Transport of Ozone at Ground Level 42
3.1.3 Transport of Ozone Aloft 42
3.2 Measurement and Interpretation of Transported Ozone 46
3.3 Procedures for Accounting for the Impact of Transport 49
and Natural Background in Oxidant/Precursor Relationships
3.3.1 Photochemical Dispersion Models 52
-------
Table of Contents (Cont'd)
Page
3.3.2 EKMA 53
3.3.3 Standard Isopleth Version of EKMA and Rollback 58
3.4 Data Limitations and Resulting Consequences 64
4.0 Caveats 69
5.0 Acknowledgments 75
6.0 References 76
Appendix A: Monitoring Network Design and Instrumentation A-l
Appendix B: Guidelines for Use of a Kinetics Model in B-l
Estimating City-Specific Ozone-Precursor
Relationships
n
-------
Executive Summary
The purpose of this document is to describe the technical basis,
uses and limitations of several approaches for relating photochemical
oxidant (expressed as ozone) to organic compounds and oxides of nitrogen.
This document is not intended as a statement of EPA policy concerning
which method to use in relating ozone to precursors. By reporting the
nature and present status of various analytical techniques, it is hoped
that the document will prove useful to agencies and/or individuals
wanting (1) to estimate the amount of precursor controls needed to
attain the National Ambient Air Quality Standard (NAAQS) for photochemical
3
oxidants (160yg/m hourly average concentration not to be exceeded more
than once per year), and (2) to estimate the reduction in ozone concentrations
accompanying specified reductions in precursors. Some of the methods
described also provide additional measures of improvement in ambient air
quality which may accompany precursor controls. These additional capabilities
are identified where applicable.
Ambient levels of ozone reflect a complex interaction of locally
emitted organic pollutants and oxides of nitrogen with transported ozone
and precursors within the framework provided by prevailing meteorological
conditions. Conceptually, the role of ambient oxides of nitrogen (NO )
/\
is to provide the basic means whereby ozone is formed. Ozone results
from a series of reactions initiated by the irradiation of nitrogen
dioxide by sunlight. In the absence of appreciable amounts of organic
compounds, resulting levels of ozone remain low as the result of a
chemical equilibrium which is established among ozone, nitric oxide (NO)
and nitrogen dioxide (NOp). Presence of appreciable amounts of organic
-------
pollutants influences this equilibrium so that higher concentrations of
ozone prevail. The dependence of ozone concentrations on organic and
NO precursors can be dominated by meteorological conditions. Sunlight
/\
intensity and temperature influence chemical reaction rates and, therefore,
the equilibrium among reactants. Mixing height and its diurnal variation
determine the extent to which transported ozone and precursors are
likely to affect local ozone concentrations. The interaction of locally
emitted organic and NO precursors with transported pollutants is also
A
influenced by prevailing wind speeds and trajectories.
Given the complexity of the photochemical oxidant problem, it is
clear that, in order to be tractable, analytical techniques must in-
corporate a number of simplifications. The accuracy of methods in-
corporating such simplifications would be best determined in verification
studies which compare estimates against data observed in an urban area
which is of interest. Unfortunately, there are few locations in which
the data base is sufficient for a comprehensive verification of analytical
methods relating oxidant to precursors. A major effort has begun to
utilize the data collected in the Regional Air Pollution Study (RAPS)
conducted in St. Louis for verification purposes. Still barring the
availability of appropriate data for comprehensive verifications, it is
possible to estimate the range of uncertainty associated with estimates
if the accuracy with which certain key input variables are measured is
known. The results of such sensitivity studies suggest that the accuracy
of predictions obtained using a photochemical dispersion model is likely
to be no worse than +_ 50% of the actual ozone concentrations. The level
of uncertainty associated with the most detailed simple approach described
in this report (i.e., the Empirical Kinetic Modeling Approach) is probably
no worse than +_ 70%. Because of the manner in which proportional rollback
-------
and statistical approaches are used or derived, it is not possible to
speculate about their absolute accuracy. Although the uncertainty bands
associated with model predictions of ozone concentrations could under
some circumstances be high, there are indications that estimates of
precursor reductions needed to reduce oxidant levels from present levels
to the NAAQS can be made with considerably greater confidence. Sensitivity
studies suggest that the error bands associated with such estimates may
be in the order of +_ 25%, with most of the potential error arising from
uncertainty about prevailing hydrocarbon/NO ratios. Uncertainty associated
A
with methods which take explicit account of accurately known hydrocarbon/NO
X
ratios may be in the order of +_ 10%.
Four analytical approaches are discussed: photochemical dispersion
models, the Empirical Kinetic Modeling Approach (EKMA), linear rollback
and application of statistical models. Since EKMA has not been described
extensively elsewhere, it is treated in greater detail in this report
than are the other approaches. Because of some fundamental questions
about the nature of the rural oxidant problem, all the approaches
described in this report are only useful in addressing the urban oxidant
problem.
Photochemical dispersion models have the greatest potential for
evaluating the effectiveness of oxidant control strategies. This
potential arises primarily from the spatial and temporal resolution
possible with such models and from the ability to relate emissions
directly to ambient ozone concentrations as a result of chemistry and
atmospheric dispersion. However, data requirements associated with
models may be prohibitive in some cases. Simple approaches may therefore
be of use.
-------
The Empirical Kinetic Modeling Approach (EKMA) makes use of a
kinetics model to express maximum afternoon ozone concentration as a
function of morning ambient levels of nonmethane hydrocarbons (NMHC) and
NO . EKMA is empirical, because it requires the use of observed second
/\
high hourly ozone concentrations and morning NMHC/NO ratios to estimate
V\
control requirements. The EKMA reflects an effort which has been
underway since December 1975 to develop a substitute for the Appendix J
approach.
Linear rollback, which is even less data-intensive than EKMA, may
also be a possible alternative for estimating bounds on control require-
ments for organic precursors. However, rollback's failure to consider
the dependence of NMHC-ozone relationships on the prevailing NMHC/NO
A
ratio during the base period limits applicability of rollback to a first
order approximation.
The fourth approach is really a class of approaches which has been
labeled "statistical procedures." Several key features for such procedures
are identified. Any one statistical relationship is seen as being site-
specific. Further, the use of statistical procedures is most appropriately
limited to considering moderate changes from the base control state,
since imposition of drastic changes could appreciably modify the functional
relationship derived from the base period data.
Although the previously described ozone/precursor relationships can
be applied in large urban areas, long range transport still may significantly
alter control estimates in some cases. Provided transport can be
measured, it can be readily incorporated into photochemical dispersion
models and into EKMA. Transport is composed of two components, a
natural background and a manmade contribution. Natural background of
ozone appears to be about .04 ppm, primarily from slow diffusion from
the stratosphere. However, the impact of natural ozone is considerably
less in urban areas due to scavenging. The impacts of naturally emitted
vi
-------
organics and NO on urban ozone are probably much smaller, if not neg-
x
ligible. A review of a rather limited available data base implies that
the principal impact of transport on the urban ozone problem is likely
to arise from unscavenged ozone which is trapped aloft overnight as a
result of surface-based nocturnal inversions. During the ensuing day-
light hours, the atmosphere becomes well mixed near the surface, and
ozone aloft reacts with locally generated precursors. This process has
been simulated using the chemical kinetics model employed in deriving
the EKMA together with several scenarios concerning the rate at which
ozone from aloft is mixed in the surface layer and with differing levels
of locally generated precursors. The results suggest that ozone trans-
ported aloft may be from 20-70% additive in urban areas. For example,
if transported ozone were .10 ppm and were 40% additive, the impact on
observed downwind maximum ozone would be .04 ppm. In order for control
requirements to be equitable, recognition should be allowed for reduced
levels of transported ozone from upwind cities having control programs
when estimating local control requirements in downwind cities. Procedures
for including transport considerations in various analytical methods are
presented in this paper.
It is recognized that the data and capabilities for implementing
procedures suggested for considering transport may not always be suffi-
cient. An examination of the impact of ignoring transport and natural
background, however, indicates that the calculated differences in pre-
cursor control requirements may frequently be of little practical sig-
nificance. An exception to this generalization may occur in the case of
a moderately sized city which is downwind from a major city and is
observing ambient levels of ozone which are less than about twice the
vn
-------
NAAQS. In such a case, simulations suggest that estimated control require-
ments may be higher than necessary if transport and background are
ignored. A second problem may result from using an approach, such as
proportional rollback, which suggests a large reduction in ozone may be
forthcoming from initial controls. Significant transport may serve to
diminish any actual improvement in air quality resulting from local
controls. Therefore, a risk in relying on any such approach is that
questions may arise about whether controlling organic emissions will
ever be effective in reducing ambient ozone. Many such questions might
be the result of using a very simple model which fails to consider
several potentially important factors contributing to the urban ozone
problem.
Summarizing, a great deal of understanding of the urban photo-
chemical oxidant problem has been gained over the past several years,
and though subject to some error and uncertainty, the methods described
in this document provide means for making useful approximations in
attempts to confront the problem.
-------
List of Figures
Page
1 Sensitivity of Maximum Afternoon Ozone Concentrations to 13
Morning Precursor Levels Measured Upwind
2 Example Problem la 19
3 Example Problem Ib 21
4 Example Problem Ic 22
5 Sensitivity of Maximum Afternoon Ozone Concentrations to 25
Precursor Concentrations Given Default NMHC/NO Ratios
A
6 Conceptual View of the Column Model 30
7 Role of Ozone Aloft on Maximum Surface Ozone Levels: A 44
Conceptual View
8 Airborne and Ground Level Ozone Concentrations During the 47
Flight of Da Vinci II (June 8-9, 1976)
9 Diurnal Variation of the Vertical Distribution of Ozone at 48
Wilmington, Ohio, on August 1, 1974
10 Conceptual View of Base and Controlled States 51
11 Procedure for Considering Two or More Impacts of Control 56
Strategies Simultaneously Using City-Specific Isopleths
12 Consideration of Transport Using the Standard Isopleth Version 63
of EKMA
A-l Examples of Acceptable Monitoring Locations for Estimating A-6
Transported Ozone
IX
-------
List of Tables
Page
1 Ambient Data for Urban Sites or Suburban Sites Not Dominated by 24
Stationary Sources
2 Comparison of NMHC Reductions Needed to Attain the Oxidant NAAQS 34
Estimated by the Rollback and EKMA Approaches
3 Impact of Various Factors on the Additivity (A) of Transported 60
Ozone to Maximum Ozone Concentrations in Urban Areas
4 The Impact of Ignoring Transport on Estimated Organic Control 68
Requirements
B-l City-Specific Procedures for Determining Emission Control B-2
Requirements
B-2 Input Requirements of the Kinetics Model B-4
B-3 Derivation of Sample Values for Relative Emissions Rates B-10
-------
1.0 Background
1.1 Introduction
In the past, Appendix J to Title 40, Part 51 of the Code of Federal
Regulations has been used to estimate the degree of reduction in
organic emissions needed to achieve the primary National Ambient Air
3
Quality Standard (NAAQS) for photochemical oxidants (160 yg/m ~ .08 ppm
1-hour average not to be exceeded more than once per year). Appendix J
was derived from envelope curves enclosing points depicting 6-9 a.m.
ambient total and non-methane hydrocarbon concentrations and correspond-
ing daily maximum ambient oxidant concentrations measured at several
CAMP sites and in Los Angeles during the late 1960's. The Appendix J
approach has been criticized for a variety of reasons. As a result of
acknowledged shortcomings in the envelope curve approach as described in
Appendix J, an EPA working group was formed in 1975 to investigate the
viability of alternative approaches. Suggestions made as a result of
working group sessions have been reviewed periodically by a group of
experts outside of EPA. The conclusion reached as a result of these
efforts is that it is not possible to recommend a single approach for
all applications. The variety of applications, complexity of individual
situations, differences in data availability and resources all preclude
use of a single procedure nationwide. Therefore, this report describes
the technical basis, uses, advantages and disadvantages of several
currently available approaches for relating ozone to organic compounds
and oxides of nitrogen precursors. Four approaches are described:
photochemical dispersion models, the Empirical Kinetic Modeling Approach
(EKMA), linear rollback and statistical models. Because the EKMA has
evolved largely through the efforts of members of the EPA working group
and because it has not been as extensively discussed elsewhere, it is
described in greater detail than the other approaches.
1
-------
1.2 Applicability of Available Procedures
The procedures described in this document address the ozone problem
as it exists within and immediately downwind from large urban areas.
There are several reasons for limiting the applicability to urban areas.
First, although the ozone problem is better understood than it was at
the time Appendix J was conceived, certain aspects of the problem are
still not well enough understood to allow reliable quantifiable estimates
to be made for all situations. For example, present understanding of
chemistry occurring within portions of an urban plume trapped aloft by a
nighttime radiative inversion does not permit adequate quantitative
estimates. The role of less reactive precursors and ozone transported
aloft versus that of low concentrations of reactive precursors emitted
locally in the synthesis of maximum levels of ozone observed the next
afternoon is also difficult to quantify. Thus, situations in which
transported ozone and/or precursors are clearly dominant may not be well
described by existing methods. The methods described herein are there-
fore not applicable to the rural ozone problem, nor to cases in which
excursions above the NAAQS for oxidant occur during nighttime or early
morning hours. Second, the most serious oxidant problem generally
occurs downwind of large cities. Ozone concentrations are, as a rule,
higher in these locations, where a substantial portion of the population
lives. Further, it is likely that strategies which are implemented to
reduce ozone levels in and near major cities will produce some (though
perhaps less) benefit in more remote areas as well. Third, it is more
practical to address initially that portion of the ozone problem which
is most readily controlled (i.e., the case where ozone is synthesized by
high concentrations of locally emitted precursors subject to control
within relatively few political jurisdictions).
-------
An important limitation of the methods described in this report is
that adequate comparisons between their estimates and observed levels of
ambient ozone have not yet been made. In the absence of such verification
studies, definitive statements about the accuracy of the various procedures
are difficult to make. The problem of obtaining suitable data for
verification of the procedures described in this report is aggravated as
a result of the manner in which the procedures would be applied. The
analytical procedures are most appropriate for estimating the changes in
ambient ozone concentrations accompanying changes in precursor emissions
or concentrations. In order to evaluate how accurate procedures are in
estimating changes in ambient ozone levels accompanying control measures,
it is necessary to have collected ambient data over a period which is
long enough to have witnessed significant reductions in precursor levels.
Such data are not widely available. In the absence of adequate ambient
data, sensitivity studies can be utilized to estimate the range of error
in various analytical procedures resulting from uncertainties in key
determinants of ambient ozone levels. It must be emphasized that in the
absence of verification studies, sensitivity studies can only serve as
extremely crude indicators of the accuracy likely with any analytical
approach. As described in Section 4.0, sensitivity studies which have
been performed with photochemical dispersion models suggest an uncer-
tainty range in absolute predictions of less than +_ 50% in an area with
a reasonably good data base. As a result of some simplifications in the
way vertical dispersion is simulated in EKMA, uncertainty in absolute
predictions using that approach is probably less than +_ 70%. A consider-
ably higher degree of confidence is likely however, if photochemical
dispersion models and EKMA are applied to estimate changes in ozone
accompanying changes in precursor levels. Assuming base level emissions
-------
are accurately known, sensitivity studies performed with EKMA imply the
uncertainty level attributable to difficulties in estimating light
intensity, reactivity, atmospheric dilution rate and diurnal emission
patterns may be less than +; 10% in areas experiencing serious (e.g., >
.16 ppm) ozone problems. Failure to consider the appropriate ambient
hydrocarbon to oxides of nitrogen ratio (NMHC/NO ) would increase the
X
error band to about +_ 25% or more. Obviously, if emissions are not well
known, errors associated with all methods are likely to be greater. An
important point which needs to be emphasized is that the error bands
quoted above are likely to be greater if a technique is applied to
simulate a control strategy which is clearly not consistent with the
technique's underlying assumptions. For example, the error associated
with predictions obtained using the standard isopleths in EKMA might be
greater if the estimates were to be compared with observed improvements
in air quality accompanying a control strategy which results in sub-
stantial alteration of the spatial configuration of emissions.
Despite the limitations described above, a great deal has been
learned since the conception of the Appendix J curve. As a result, it
is possible to reduce a number of the shortcomings and uncertainties of
the Appendix J curve.
1.3 Organization
The remainder of this report is organized in the following mariner.
Section 2.0 provides a brief assessment of photochemical dispersion
models and a more complete description of three (EKMA, rollback, and
statistical procedures) non-data-intensive methods for relating ozone to
its precursors. Section 3.0 discusses the roles which transported
ozone/precursors are thought to play in the urban oxidant problem.
-------
Methods for including transport and natural background in some of the
techniques described in Section 2.0 are also presented. It is recognized,
however, that the available data base in many urban areas may not be
sufficient to implement suggested procedures for considering transport.
Thus, the consequences of not considering transport and natural back-
ground are also discussed in Section 3.0. Section 4.0 identifies some
caveats which should be kept in mind when applying the information
presented in earlier sections. This document contains two appendices.
Appendix A describes the monitoring network design thought to be most
appropriate for assessing a city's oxidant problem as well as salient
features of some of the key instrumentation. Appendix B describes in
detail the data needs and assumptions of a procedure which can be used
to estimate control requirements in specific urban areas. Further
documentation of recommended values for natural background and ambient
air quality and modeling analyses supporting many of the recommendations
3
presented in this report are contained in a separate report.
2.0 Methods for Relating Ozone to Precursors
2.1 Use of Photochemical Dispersion Models
Multisource air quality models employing appropriate chemical
mechanisms are believed to provide the best approach for relating ozone
to precursors. The photochemical dispersion modeling approach has the
potential for being theoretically more sound than any of the other
approaches described in this report. This results from the attempt to
simulate chemical" (e.g., transformation of primary species such as NO
and non-methane hydrocarbons into secondary species such as N0? and
ozone) and physical (e.g., atmospheric dilution, diurnal meteorological
variations) phenomena. Parameters subject to control (i.e., emissions)
are entered directly as input to such models. This facilitates examination
-------
of the impact of hypothetical control strategies.
There are two major types of models. Eulerian models calculate
concentrations at fixed locations in space at specified times. As a
result of practical limitations and computational expenses, calculated
concentrations represent typical concentrations likely to occur over
2
areas of 1 km or greater. The concentration estimated at each location
during each selected time period results from an interaction among
emissions, chemical reactions and the transport and dilution introduced
by prevailing meteorological conditions. It is then possible to superim-
pose iso-concentration lines (similar to contour lines on a map) over
the area of interest. Such iso-concentration lines can be drawn for
ozone or for other pollutants (e.g., NCL) which participate in the
photochemical process and are of interest for regulatory purposes.
Eulerian models are thus capable of providing the most complete estimate
possible of the impact likely to result from hypothetical control strat-
egies,, The spatial resolution afforded by such models is a particularly
desirable feature for several reasons. First, a control agency can make
judgments about whether control of equal amounts of precursors from
different sources will be equally beneficial. For example, is control
of a ton of organic emissions from a single point source as effective as
control of a ton of organic emissions from motor vehicles throughout a
city? For approaches having no spatial resolution, such a question can
only be addressed in a more limited fashion. Second, models with
spatial resolution provide greater flexibility in estimating the benefits
of precursor controls. For example, reduction in the population exposed
to ozone above certain specified levels can be assessed.
-------
The second type of photochemical dispersion model is the Lagrangian
model. Lagrangian models estimate the concentration of ozone and other
chemical species within a specified parcel of air. These models "follow"
an air parcel and estimate pollutant concentrations within the selected
parcel during subsequent times. The concentrations which are estimated
result from an interaction of chemical reactions among contaminants
which are within the parcel at the beginning of a simulation and emissions
encountered along the trajectory of the selected parcel. Meteorological
variables also play key roles in determining concentrations of ozone and
other pollutants in the selected parcel. The trajectory of the parcel
(and hence the emissions encountered), the vertical and horizontal
exchange with surrounding air and some key chemical reaction rates are
all determined by prevailing meteorological conditions. Lagrangian
models may be useful in addressing more limited types of questions, such
as:
(1) What is the maximum ozone concentration likely in the vicinity
of a city (i.e., will the NAAQS be met)?, or
(2) What is the impact of a major new source (e.g., a highway)
likely to be if it is built in an urban area?
A major advantage enjoyed by Lagrangian models over their Eulerian
counterparts is that they are likely to be considerably less expensive
to use. Both Eulerian (and to a lesser extent Lagrangian) photochemical
dispersion models can provide air quality predictions with temporal
resolution. Consequently, they can be used to evaluate strategies
resulting in different diurnal emission patterns and impacts on different
source categories.
-------
The major limitations to the widespread use of photochemical
dispersion models are:
(1) The data base required as input is extremely detailed and
extensive and may not be readily obtainable. For example, to obtain the
spatial and temporal distribution of ambient pollutants discussed
earlier, a large number of meteorological measurements may be needed to
characterize the wind field. In addition, spatial and diurnal emission
patterns for mobile and stationary sources, the mix of the organic
emissions, plume rise characteristics and upper air data, as well as a
fairly large oxidant monitoring network to operate and validate the
models, are also needed.
(2) In order to utilize photochemical dispersion models incorporat-
ing the information described earlier, the user must have access to
extensive computational facilities.
(3) Photochemical dispersion models are subject to several of the
same shortcomings of less data-intensive approaches (e.g., selection of
appropriate boundary and initial conditions to simulate transport).
(4) Photochemical dispersion models have not been extensively
verified. Therefore, their ability to simulate temporal and spatial
impacts has not been widely demonstrated. Extensive verification efforts
will be underway shortly by EPA using the recently completed St. Louis
Regional Air Pollution Study (RAPS) data base.
-------
Complete descriptions of individual models are best obtained from
the vendors of such models. Cursory descriptions of several models and
their prior applications are available elsewhere. '
2.2 Non-Data-Intensive Methods for Relating Ozone to Precursors
Approaches having less extensive data requirements than photo-
chemical dispersion models may provide useful approximate answers to
limited questions. For example, what level of organic emission controls
are needed to attain the oxidant standard, or what reduction in maximum
oxidant concentration will accompany a specified reduction in ambient
levels of hydrocarbons? Three basic approaches for relating ozone* to
precursors are described in this section: use of an Empirical Kinetic
Modeling Approach (EKMA), linear rollback and statistical techniques.
Of these three approaches, the EKMA is believed to have the soundest
underlying scientific basis. All of the approaches described in
Section 2.2 assume that meteorological conditions in the forecast period
are the same as during the selected base period.
2.2.1 Use of an Empirical Kinetic Modeling Approach (EKMA)
The Empirical Kinetic Modeling Approach (EKMA) utilizes a set of
ozone isopleths which depict maximum afternoon concentrations of ozone
downwind from a city as a function of initial (i.e., morning) concentra-
tions of NMHC and NO NMHC and NO emissions occurring later in the
x, x 3
day, meteorological conditions, reactivity of the precursor mix and
concentrations of ozone and precursors transported from upwind areas.
* Since the Federal Reference Method for photochemical oxidant is
specific for ozone, the methods described herein use ozone as a surrogate
for oxidant unless stated otherwise.
-------
The physical model underlying EKMA is similar in concept to the
previously described Lagrangian model. A column of air consisting of
initial concentrations of ozone and precursors is transported along an
assumed trajectory. As the column moves, it encounters fresh emissions,
which are assumed to be mixed uniformly within the column. The column
is assumed to act like a large smog chamber in which the precursors
react to form ozone.
EKMA is best used to determine the sensitivity of maximum hourly
ozone concentrations observed within or downwind of a city to changes in
ambient levels of non-methane hydrocarbon (NMHC) and oxides of nitrogen
(NO ) precursors. EKMA is most suitable for addressing questions like,
/\
"how much reduction in local prevailing precursor levels would be needed
to attain the .08 ppm standard for oxidants (measured as ozone)?" or,
"what reduction in ozone levels is likely to accompany a specified
reduction in precursor levels?" The method is less suitable for esti-
mating the impact of strategies which result in substantial changes in
source configurations or for questions about what the impact of control-
ling a single or small group of sources would be.
Many of the underlying principles of EKMA have already been reported
789
in the technical literature. ' ' The approach is an empirical one,
because it requires the use of observed air quality data. The relation-
ships among ozone and its precursors which underlie the approach are
based on the application of a chemical kinetics model. ' The kinetics
model used represents a detailed sequence of chemical reactions which
has been proposed for a mixture of propylene, n-butane and NO . The
A
chemical mechanism used in the kinetics model is based on information
10
-------
obtained in smog chamber experiments with propylene, n-butane and NO
A
conducted primarily at the University of California at Riverside. The
kinetics model predictions were matched against Bureau of Mines (BOM)
12
smog chamber data obtained by irradiating automobile exhaust. Initial
proportions of propylene and n-butane were then manipulated in the model
so that consistently close agreement was obtained with observations in
the BOM chamber. Of the available smog chamber studies, the ones using
automotive exhaust are thought to use a mix of reactants most represent-
ative of the mixes found in urban atmospheres. Automotive exhaust
contains a large variety of reactants which are difficult to include in
experiments using artificial mixes. One key category of reactants is
aldehydes. Aldehyde concentrations identical to those reported in the
BOM experiments were used in the kinetics model when the appropriate
proportion of initial propylene and n-butane was being determined.
There are two variations of EKMA. The first involves the use of
city-specific ozone isopleths. The second utilizes a standard set of
ozone isopleths in which fixed assumptions have been made about sunlight
intensity, atmospheric dilution rate, reactivity and diurnal emission
patterns. There are two main reasons why the city-specific approach is
preferable,. First, it is possible to use the city-specific version of
EKMA to evaluate a wider variety of control measures. Of particular
note is the capability to estimate the impact of control measures which
are initiated concurrently. For example, questions like, "what would be
the impact on maximum 0, if local organic levels were reduced X% and
11
-------
transported ozone were reduced Y%?", or "what would happen if the diurnal
emission pattern were varied?" can be addressed. The second advantage
of the city-specific approach is that it is more desirable to use locally
appropriate assumptions about atmospheric dilution rate, sunlight
intensity and diurnal emission patterns. This second advantage, however,
is probably not as crucial as it might initially appear. As discussed
later, proper application of standard ozone isopleths in EKMA should
provide estimates for control requirements which are essentially unaffect-
ed by use of assumptions about dilution rate, sunlight intensity and
diurnal emission patterns which are not locally specific. The advantages
of using the standard isopleth variation of EKMA are that less input
information is required and it is not necessary to use a computer.
Although, on balance, the use of the city specific isopleth version of
EKMA is most advantageous, for ease of presentation, the use of standard
ozone isopleths in EKMA will be described first.
2.2.la Use of Standard Ozone Isopleths
Derivation
Figure 1 relates initial concentrations of non-methane hydrocarbons
(NMHC) and oxides of nitrogen (NO ) to maximum ozone (CL) concentrations
X »5
formed within 10 hours. The curves in Figure 1 were derived using the
previously described kinetics model and the following assumptions:
(1) simulations were 10 hours long;
(2) the proportional mix of propylene and n-butane was
chosen so that the reactivity of the mix was comparable to that observed
for automobile exhaust in the BOM chamber studies (i.e., 25% propylene
(ppmC), 75% n-butane (ppmC));
(3) aldehydes were assumed to be 5% of the initial NMHC
levels (in ppmC);
12
-------
CO
O)
O)
S-
o
O
O)
S-
O-
S-.
O
O
-t-J
I/)
c:
o
+j
fO
OJ
o
c:
o
o
QJ
c
o
N
O
o
o
c
S-
OJ
4-J
X
(B
E •
ID
I*- C
O -i-
>-, Q-
-V-> Z3
•r- QJ
-M i-
•r- 3
co co
C ro
Ol O)
00 E
mdd
-------
(4) photolytic rate constants were varied in accordance
with diurnal variation in sunlight intensity during 8 a.m.- 6 p.m. LOT
for the summer solstice at 34°N latitude (i.e., similar to the latitude
of Los Angeles, Phoenix, Dallas, Atlanta);
(5) an atmospheric dilution rate (caused by lifting of an
elevated inversion layer) of 3% per hour was assumed until 3 p.m., with
a dilution rate of zero thereafter.
(6) all chemical reactions which had been incorporated
within the kinetics model to account for chamber-induced artifacts were
disregarded;
(7) no emissions were considered after 8 a.m. LOT.
(8) transported ozone was assumed to be negligible.
Use
Before describing how diagrams such as the one in Figure 1 are used
in EKMA, it is necessary to define what is meant by the expressions
"absolute sense" and "relative sense." Referring to Figure 1, it can be
seen that concentrations of NMHC and NO of 0.4 ppmC and .06 ppm respec-
A
tively correspond with an ozone concentraton of .16 ppm. Thus, applying
Figure 1 to say that morning concentrations of NMHC and NO of 0.4 ppmC
A
and .06 ppm will lead to maximum 0^ concentration of .16 ppm downwind
several hours later is an example of applying the diagram in an "absolute
sense." In Figure 1, each point on the .16 ppm ozone isopleth has
corresponding NMHC and NO coordinates. One could also start from an
A
identifiable point on the .16 ppm isopleth and ask the question, "what
would be the impact on the maximum 03 level if NMHC were reduced 50%?"
This would represent the application of the diagram in a "relative
14
-------
sense." In such an application, it is no longer important what level of
0^ will result from specified levels of precursors. What is of interest
is what is the change in ozone levels if NMHC and/or NO are changed a
X
relative amount. The isopleths in Figure 1 should not be used in an
absolute sense, since they reflect specific assumptions about dilution,
sunlight intensity, wind speed, emission configuration and patterns and
reactivity which may not coincide with what actually occurs on high
ozone days in a specific locale.
As will be seen later, however, widespread application of the
isopleths is feasible on a relative basis. Thus, Figure 1 should be
interpreted as depicting the sensitivity of maximum afternoon ozone
concentrations to changes in precursor levels. Consequently the diagram
may be used to estimate the degree of reduction in ambient precursor
levels needed to attain a specified level of ozone. Conversely, the
curves may also be used to obtain a rough estimate of the effect hypothe-
tical reductions in local precursor levels might have on the maximum
afternoon ozone concentration observed downwind.
Two pieces of information are needed to apply Figure 1. The first
is the design value* of ozone observed in or near the city. The most
appropriate location of the instrument measuring this value will depend
on prevailing local meteorology during periods conducive to oxidant
*The second highest hourly ozone concentration observed during the
base period. If there is more than one ozone monitor in the network,
the design value is the highest of the second highest hourly concen-
trations observed during the base period at all monitors.
15
-------
formation. Utilizing recommendations of the Standing Air Monitoring
Work Group (SAMWG) as a guide, the ozone concentration used as input
to the method should generally be the design value observed within 15-30
km downwind of the central business district. The SAMWG recommendations
are, in turn, based on field studies and reviews in which ozone gradients
14 15 16
downwind from urban areas were examined. »'^>IU since siting of instru-
ments is a crucial part of assessing the need for and effect of control
strategies, it is discussed more fully in Appendix A. Because of pre-
viously discussed limitations, only ozone concentrations observed during
the afternoon should be considered. The isopleths, as drawn in Figure
1, are appropriate for maximum afternoon 03 concentrations averaged over
periods of an hour. The second piece of information required is the
prevailing NMHC/NO ratio. Ideally, the most suitable ratio for use
A
would be the one occurring between 6-9 a.m. LOT within the urban core*
which is upwind from the site on the day observing the afternoon ozone
concentration corresponding to the design value. Because NMHC and NO
A
readings are apt to be relatively high in urban areas during 6-9 a.m.,
more confidence can be placed in the accuracy of these readings. It is
therefore recommended that ratios measured in the urban core be used in
EKMA. Furthermore, it can be shown that although NMHC and NO tend to
A
be higher in the morning, the ratio does not appear to vary nearly so
3
much as individual NMHC and NO levels during the course of the day.
A
*Urban core is defined as the central city. In large cities this
may encompass areas about 3-4 km from the center of the central business
district.
16
-------
The 6-9 a.m. NMHC/NO ratio used in the EKMA is viewed as a characteristic
A
of that city which would prevail during the remainder of the morning and
early afternoon in the absence of chemical reactions. Thus, use of the
6-9 a.m. ratios is considered to be analogous to the results of smog
chamber experiments in which sensitivity of maximum ozone levels to
initial combinations of NMHC and NO is described. Unfortunately,
X
individual NMHC measurements may be subject to significant error as
discussed in Appendix A. Further, 6-9 a.m. measurements of NMHC/NO are
X
frequently missing on the days in which highest ozone levels are observed.
Therefore, unless there are data from at least two separate monitoring
sites, and these data are comparable, a more robust (i.e., less affected
by extreme values) and available measurement of the NMHC/NO ratio is
A
preferred. It is recommended that the median 6-9 a.m. LOT ratio observed
on the days having the five highest ozone values with accompanying NMHC
and NO data be used. In the event data from more than one site are
A
available, the 6-9 a.m. ratios from each of these sites should be averaged
on each of the five days. The median of these five averages should then
be used in the EKMA. Close correspondence has been observed between
this five-day median ratio* and ratios on the day with maximum ozone in
3
a number of cities. It is probable that this close agreement is a
result of the similar meteorological conditions which prevail on days
experiencing highest ozone levels.
*In following examples, "median NMHC/NO ratio" will serve as
A
shortened notation for "median NMHC/NO ratio on days experiencing high
A
levels of ozone."
17
-------
Examples
The following example illustrates the use of standard isopleths in
the Empirical Kinetic Model Approach to relate reductions of ambient
precursor concentrations in a city to reductions in maximum ozone concen-
trations downwind. In the examples, letters which are primed reflect
the post-control state.
Given: The design value of 0, = .28 ppm;
Median NMHC/NOV ratio = 12:1,
X
Find:
(a) The reduction in ambient organic concentrations needed
to attain the .08 ppm oxidant standard if no change in
ambient NO levels is anticipated.
A
(b) Same as (a) but with a 50 percent reduction in ambient NO
A
anticipated.
(c) The reduction in the second high hourly ozone concentra-
tion likely if NO concentrations are reduced 30% and organic
A
concentrations are reduced 70 percent.
Solution:
(a) (1) Plot median 6-9 a.m. NMHC/NO ratio line on
X
Figure 2, and note the intersection of this line
with the .28 ppm 03 isopleth.
(2) Likely reduction needed is:
(NMHC)a - (NMHC)a' 1.26 - .26
~ (NMHC)a~ 1.26
R = .79 or 79%
-------
o
E
OJ
o
Q.
O)
a.
C\J
UJ
a:
CD
uidd
-------
(b) If a 50% reduction in NO accompanies the needed organic
A
emission controls, the procedure is to:
(1) Plot the given information on Figure 3 and note the
position of the points on the .28 and .08 ppm ozone isopleths.
(2) Likely reduction needed is:
(NMHC)d - (NMHC)d' _ 1.26 - .16 _ oc oco/
_ ——• —• — - — — - .00 - ob/o
(NMHC)d 1.26
Note that in this particular example, NO control slightly increased
A
NMHC controls needed to meet the standard.
(c) To estimate the impact of a 70% reduction in organic concen-
trations together with a 30% reduction in NO ,
X
(1) Plot the given information on Figure 4 (Point g). Note the
coordinates of point g (1.26 ppmC, .104 ppm);
(2) Calculate coordinates of post-control point, g' given the
specified reductions in precursor concentration.
9'NOx = (1 - .30)gNQ . g'NMHC = (1 - .70)
X
= (.70) (.104) = (.30) (1.26)
9'NOx = .073 ppm ; g'NMHC = .38 ppmC
20
-------
'3NOZO
o
E
o
I
cu
o
Q.
o>
Q-
rD
X
o:
uidd 'XQ|M
21
-------
uidd
-------
(3) Plot the post-control point on Figure 4. Results indicate that
the second high hourly ozone concentration would be about .16 ppm.
In the preceding examples, it has been assumed that sufficient data
exist to estimate the representative median ratio in an urban core area.
Frequently, however, this information may not be available. The issue
of how to estimate an appropriate NMHC/NO ratio for specific cities is
/\
discussed further in Appendix A. It may ultimately be necessary to
assume default values for the ratio. Data from a number of monitoring
3
sites are presented in a separate document. Fourteen of these sites
are considered to have a large enough sample size and are located either:
(1) in a highly populated urban area (which may or may not contain
substantial stationary sources of organic pollutants), or
(2) in a suburb which is not dominated by stationary sources.
These sites are presented in Table 1. Best estimates of typical 6-
9 a.m. NMHC/NO ratios, based on an examination of these sites, would
A
be a median ratio of about 9.5:1. The 10th and 90th percentile NMHC/NO
/\
ratios observed in the cities in Table 1 may be used to provide a measure
of the uncertainty introduced by using default ratios. These data
suggest 10th and 90th percentile ratios of about 6:1 and 16:1 respectively
are appropriate. In this case, Figure 5 may be used to relate the sensi-
tivity of maximum afternoon ozone levels downwind to changes in upwind
ambient precursor levels in the absence of appropriate ambient precursor
data.
23
-------
to
^
c.
ra
E
QJ
or
"^T
c: CL
T3 0 CL
E QJ 0—-
13 > cr
EE S— — CD
•i— GJ GJ d
X to 4-> O
0
-0 £0
OJ L3
4-1
ro 4-
U C
O
•— ^
d i/1
0 Of
ra £
4-5
CO CO
o
*
CD S-
0 0
O T-J
*• re
LO E
O
CXJ «
>1
11 C
CX E
o \
a- 2
CO
CD
in
QJ
U
S^
13
0
tn
40
.^
0
CL
.
E d
ro i—
1 -^
CXJ d
rO Q
d S- CQ
o u_ cj
ro o cr
4-5 4->
co co s:
LO
p —
LO
Z3
-o
d
^
>
rO
OJ
d
*~ '
O
Cxj
TU
c
fO
-a
(U
rsl
i —
•^
£_
4-J
ro
QJ
S-
ro
"O
QJ
4-5
fO
i —
CL
0
CL
-o
QJ <=£
4-> _J
ro
U C
0 3
r— O
4-J
d C
0 3
•— o
ra
4-5 d
CO *r-
-k
LO
CXJ
ID d
CD CO
+-> O
ra
U 4-
o o
3
d 2:
o
rO E
4-5
CO CO
r*.
0
4_
- a o
.0 CQ
S- CJ ~Z
r3 s
-Q 4-
rs o *
(/) -r-
^ E
c:
O • CO
4-> r- r —
13 in
O CXJ d
re <— ro
, —
CXJ
rO
S_
4-5
to
3 X
~a a>
d i —
•r— Q_
E
c o
rO U
>^
4-1
CJ
d
ro
JD
S^
13
JD
r3
VI
<:
—i
4;
CXJ
ro
E
Q
0 CQ
CXJ C_>
* «=E
0
0 4-
0 0
LO LU
CM ^
UJ
QJ
d
O
ro
2:
d
o
QJ
4->
CO
^
to Q
3 CQ
CLC_J
{=
ro "4-
c_j o
QJ ^:
cn
QJ .
r^ *E
o
C_> CXJ
w
OJ
4-J
LO
4-
• O
tn
4-5 UJ
nr ^
S-
S'e
13
nr cn
CO
• Q
CL CQ
O CJ
CL
O
cr o
O 0
>^co
ro ^f
cn CM
•o
cxj d O
"~^s rO U
O Q 4- X
«d- S- CQ O
4~ CJ
o o
QJ
4-> CO
•r- C
Ul O
o •«-
CX 4-5
(= fO
O 4-)
CJ CO
*3-
CXJ
•
CO
o
o
I—
-------
CJ
E
0.
o.
o"
a:
s-
o
CO
S-
^
o
ai
S-
a.
o
-M
CO
c
o
c:
CD CO
u o
c -r-
O -M
U tC
s_
O)
JZ X
o o
N "Z.
o ^^
o s:
o -z.
a>
01
-a
•i- O)
X >
ro -r-
•i- ro
> S_
•r- -I-J
+-1 C
•^ QJ
-------
For several reasons, isopleth diagrams such as those in Figures 1-5
are most properly interpreted when used in a relative rather than an
absolute sense. First, as described previously, the absolute position
of the isopleths depends upon a number of underlying assumptions concerning
meteorological conditions and emission patterns. Unless the meteorological
conditions and emission patterns corresponding to those assumed in
deriving the isopleths in Figures 1-5 are similar to those occurring in
the city of interest on smoggy days, there is no reason to expect the
absolute position of the isopleths to be correct. The relative positions
of the isopleths, however, should be less sensitive to these differences.
As discussed elsewhere, sensitivity tests have indicated that predicted
control requirements for organic emissions are not very sensitive to
changing dilution rates, solar intensity, diurnal emission patterns and
changes in reactivity when the isopleths are applied in a relative
sense. One advantage to beginning with the observed design value of
03 as input to the EKMA is that this parameter inherently reflects the
local prevailing meteorology on the smoggiest days. In this respect,
the standard isopleth approach is area-specific. It is crucial, of
course, that the CL monitoring data have been collected at sites likely
to observe the maximum 03. Therefore, it is important to consider local
meteorology in designing monitoring networks likely to observe high
ozone values. The important issue of instrument location is discussed
further in Appendix A. A second reason for applying the isopleths in a
26
-------
relative sense is that the isopleths reflect the behavior of one fixed
mixture of automotive exhaust. Therefore, use of these isopleths in an
absolute sense in areas having drastically different mixes of organic
pollutants may be unreliable. Relative positions of the isopleths are
less sensitive to the mix of organic precursors assumed. Third,
simulations in which the kinetics model was used to derive the isopleths
in Figures 1-5 did not include allowance for any injection of precursors
after 8 a.m. LOT. Sensitivity tests conducted with the model indicated
that the maximum CL concentration formed on the first day of the simulation
is increased by post-8 a.m. precursor emissions. However, these same
sensitivity studies indicated that estimated control requirements (using
the isopleths in a relative manner) were insensitive to differing
diurnal emission patterns. It is important to point out that in the
sensitivity studies described in Reference 17, post-8 a.m. emissions
were reduced in proportion to reductions in initial NMHC concentrations.
Thus, in using the standard isopleths, proportional reductions in a11
emissions (both 6-9 a.m. and post-9 a.m.) must be assumed in order for
the curves to be applied properly. No distinction can be made between
6-9 a.m. emissions and other emissions with the standard isopleths.
Fourth, it should be kept in mind that the isopleths have been primarily
validated against smog chamber data. Smog chambers represent a simpli-
fication of the urban atmosphere in that several contaminants which
may have a potential impact on maximum 03 may have been excluded from either
27
-------
the chamber or the model or both. Finally, a limitation in this method,
as well as in linear rollback or any other empirical technique, is that
there will always be some uncertainty about whether the maximum 0,
O
concentration is measured. On any given day it is unlikely that the 03
monitor is at the precise location experiencing the maximum Oo concen-
tration.
2,,2.1b Use of City Specific Isopleths
As indicated in Section 2.2.la, the diagrams presented in Figures
1-5 were derived using a fixed set of assumptions concerning sunlight,
atmospheric dilution rate, reactivity and diurnal emission patterns.
Figures 1-5 also assume that the impact of transported ozone and pre-
cursors is negligible. Possible means for considering transport using
the standard isopleths are discussed in Section 3.0, as are the conse-
quences of ignoring transport. The fixed assumptions inherent in using
standard isopleths limit the flexibility of that variation of EKMA in
considering the impact of precursor control strategies. Several of
these limitations can be easily overcome by using isopleths which have
been derived using city specific data. Generally, it is appropriate to
derive isopleths for two types of situations: (1) isopleths derived
specifically for the base period (i.e., that situation corresponding to
conditions on the day in which the design value of ozone is recorded in
a city), and (2) isopleths derived for a controlled state (those conditions
which are expected to prevail in the city being reviewed after controls are
initiated). This section describes a model which can be used to derive
28
-------
sets of ozone isopleths reflecting specific assumptions concerning
sunlight intensity, atmospheric dilution rate, diurnal emission patterns
and transport. The use of city-specific isopleths in EKMA provides one
with the important Capability of being able to consider different changes
(e.g., reduction in local precursors plus reduction in ozone transported
into a city or gross changes in diurnal emission patterns) which occur
concurrently. This capability will be illustrated in Section 3.3.
Figure 6 is a conceptual view of the model used to generate city-
specific isopleths for use in EKMA. The model considers an imaginary
column of air which extends from the earth's surface to the base of an
elevated inversion. The diameter of this column is large enough so that
horizontal exchange of air in and out of the column is ignored. Air
within the column is uniformly mixed at all times. The column is moved,
at a speed determined by the wind, along a specified trajectory. As the
column moves, its volume is increasing (decreasing) in accordance with
the diurnal raising (lowering) of the atmospheric mixing depth. As the
column encounters precursor emissions along its trajectory, these are
assumed to be mixed instantaneously and homogeneously within the column.
The column is assumed to behave like a large reaction vessel. The
previously described kinetics model is used to simulate the transformation
of precursors into ozone within the column. For a fixed set of assumptions
about sunlight intensity, dilution rate, diurnal pattern of emissions
injected into the column, reactivity and transported ozone, maximum
afternoon ozone can be plotted as a function of initial NMHC and NOV
29
-------
INITIAL
CONDITIONS
DILUTION RATE, x%/hr.
KEY
ti = TIME PERIOD i
Mi = MIXING DEPTH AFTER TIME i
Qi = PRECURSORS INJECTED INTO
COLUMN DURINGTIME i
-= SUNLIGHT INTENSITY DURING
TIME i
Figure 6. Conceptual view of the column model.
30
-------
concentrations, as was done in Figures 1-5. This procedure is then
repeated using a set of assumptions which is consistent with the post-
control situation.
The conceptual model underlying the generation of city-specific
isopleths is capable of considering data in various degrees of detail.
If the data are very detailed (i.e., the trajectory is determined by
wind field data, diurnal emission patterns are broken down by source
type and location, etc.), the model begins to approach in sophistication
the Lagrangian models described in Section 2.1. The most serious
limiting factor is the instantaneous homogeneous mixing which is assumed
in the column model. The model which was implicit in the derivation of
the standard isopleths described in Section 2.2.la represents a much
simpler application of the column model (e.g., no emissions after 8 a.m.
were considered).
Application of City-Specific Isopleths
A computer program to generate ozone isopleth diagrams using loc-
18
ally appropriate assumptions and an accompanying User's Manual is
expected to be available shortly. Appendix B discusses in detail consider-
ations used in estimating appropriate input parameters to this computer
model. Once an appropriate isopleth diagram is generated by the program,
it is applied in EKMA exactly the same way as the standard set of isopleths.
The procedure is somewhat more complicated if changes in transport,
diurnal emission patterns, etc., accompany a control strategy. This
procedure will be discussed in Section 3.3, where procedures for inte-
grating transport into approaches relating ozone to precursors are
addressed.
31
-------
As discussed above, use of city-specific isopleths in EKMA has
potential advantages over the standard isopleth approach. The most
important advantage of using city-specific isopleths is the ability to
consider several factors (e.g., changes in transport, local emissions,
diurnal emission patterns, etc.) concurrently. However, it should be
remembered that flexibility to consider locally appropriate factors such
as atmospheric dilution rate (caused by lifting of the mixing depth),
diurnal emission patterns and the amount of transported ozone is only an
advantage if one is able to estimate these parameters accurately.
2.2.2 Use of Linear Rollback
Linear rollback is simply the intuitively appealing concept that
ambient concentrations of a pollutant are directly proportional to
emissions of that pollutant. Furthermore, the relationship between the
ambient concentration of a pollutant and its emission is a constant as
shown in Equation (1).
x = aQ (1)
where x = concentration variable
Q = emission variable
a = constant of proportionality
32
-------
If a significant portion of the pollutant is transported into the area
of interest, Equation (1) becomes:
x = aQ + b (2)
b = Concentration of the pollutant which
is not attributable to local emissions,
If an air quality goal, x1, is established, emissions must be reduced
to the corresponding level, Q1
x1 = aQ' + b (3)
Combining equations (2) and (3), the reduction needed to attain the air
quality goal is
Q - Q' - (x - x') m
Q (x - b) (^>
Since there are no appreciable emissions of ozone, however, Q and Q1
must represent emissions of ozone precursors (i.e., organics).
For the linear rollback relationship to hold for organics and
ozone, Equations (2) and (3) must be valid. That is, there must be a
single proportionality constant, "a", between organic emissions and
maximum ozone. Re-examining Figure 1, shows that this is not always the
case. Even if it were assumed that the isopleths in Figure 1 are not
absolutely correct, the proportionality "constant" between NMHC and 0~
depends on (1) the prevailing level of NO and (2) the changes in NO
A A
accompanying hydrocarbon control strategies. For example, for 03 = .28
ppm, according to Figure 1 the proportionality constant, "a" can be
anything from .33 on down. Proceeding from the 0, = .28 isopleth down
to the 0, = .08 ppm isopleth at a constant NMHC/NO ratio of 10:1, one
0 X
finds proportionality constants between NMHC and 0-, varying between .25
and .50. Despite the questionable physical basis for rollback, linear
rollback can be shown to agree reasonably well with trend data reported
33
-------
TABLE 2. COMPARISON OF NMHC REDUCTIONS NEEDED TO ATTAIN THE
OXIDAKT NAAQS ESTIMATED BY THE ROLLBACK AND EKMA APPROACHES
d High Hourly
one Concentration
(a) Using
.10
.15
.20
.25
.21
.32
.14
.13
.18
(b) using
.10
.15
.20
.25
.21
.32
.14
.13
.18
(c) Using
.10
.15
.20
.25
.21
.32
.14
.13
.18
NMHC
NO Ratio
Constant NO
EKMA x
Rollback
Constant
.NMHC/NO
EKMA X
Typical NMHC/NOx Ratios
10:1
9.8:1
7.5:1
10.2:1
11.9:1
8.0:1
9.5:1
11.8:1
8.8:1
Low NMHC/NO
7.1:1
4.8:1
4.5:1
6.2:1
7.1:1
5.9:1
4.2:1
8.0:1
5.6:1
High NMHC/NO
20.0:1
13.2:1
15.0:1
16.8:1
40:1
15.0:1
13.8:1
23.8:1
14.7:1
36?
67%
64%
76%
m
72%
63%
62%
67%
Ratios
33%
44%
48%
62%
63%
63%
38%
56%
54%
Ratios
X
53%
73%
80%
85%
91%
85%
72%
78%
78%
22%
46%
59%
68%
62%
75%
42%
40%
55%
22%
46%
59%
68%
62%
75%
42%
40%
55%
22%
46%
59%
68%
62%
75%
42%
40%
55%
27%
62%
75%
85%
77%
90%
58%
55%
70%
33%
59%
72%
81%
77%
87%
59%
53%
70%
28%
65%
77%
85%
81%
89%
62%
53%
72%
34
-------
19
from Los Angeles. Table 2 compares estimates obtained with rollback
versus those obtained using Figure 1 for ambient data typical of those
observed in a number of cities. Rollback is compared with EKMA for two
scenarios:
(a) NO levels are kept constant while NMHC levels are reduced;
A
(b) the NMHC/NOv ratio is kept constant while NMHC levels are
A
reduced (i.e., NO is reduced proportionally).
A
Table 2 indicates that, with the exception of cases having very low
NMHC/NO ratios, needed NMHC reductions estimated with rollback are
A
almost always less than those obtained with the standard isopleth version
of EKMA. Thus, the main justifications for using linear rollback are
that it is simple, and that under NMHC/NO ratios believed to prevail in
A
most U. S. cities, it appears to be useful in serving as a lower bound
for estimates of hydrocarbon controls needed to attain the oxidant
standard.
Example
The example presented in Section 2.2.1 will be repeated.
Given: The Design Value of 03 = .28 ppm.
Find: (a) The reduction in ambient organic concentrations
needed to attain the .08 ppm oxidant standard if no
change in ambient NO levels is anticipated.
J\
(b) Same as (a), but with a 50% reduction in ambient
N0x.
(c) The reduction in second high hourly ozone concen-
trations likely if NO concentrations are reduced 30%
A
and organic concentrations are reduced 70%.
35
-------
Solution:
(a) Using Equation (4),
R = -28 - -08 = .71 = 71%
. Co
(b) Answer is the same as in (a), since the rollback approach
assumes differences in NO levels are irrelevant. It will
X
be remembered that the reductions calculated with the
isopleths in parts (a) and (b) were 79% and 87% respectively.
(c) if NMHC is reduced 70 percent,
7n - -28 - *
•/u .28
x = .08 ppm
This is more optimistic than the estimate of about .16 ppm obtained
with the Empirical Kinetic Modeling Approach.
An assumption in rollback is that the amount of organic emission
controls needed to attain the oxidant standard is independent of the
prevailing NMHC/NO ratio. However, smog chamber experiments suggest
X
that the lower the ratio, the more effective the hydrocarbon reduction
is in reducing the maximum ozone formed. Thus, at very low NMHC/NO
X
ratios, linear rollback may underestimate the effectiveness of organic
controls. Conversely, at high ratios, estimates obtained with linear
rollback may be overly optimistic. From Table 2 it can be seen that,
with the prevailing ambient conditions which appear to be typical in
many cities, the net effect of the rollback assumptions is to estimate
that less control is needed to attain the oxidant standard than would be
implied by the EKMA. Under the ambient conditions which apparently
prevail in most cities, rollback is likely to differ most from predictions
of kinetics models during the period in which initial control increments
36
-------
are exercised (i.e., when the NMHC/NO ratio is still relatively high).
X
Thus, for example, if organic emission reductions of 30% were implemented
in a city experiencing moderate (10:1) NMHC/NO ratios, the corresponding
A
improvement in maximum ozone concentrations may be considerably less
than the 30% predicted with linear rollback. The risk in relying on
rollback is that when the corresponding 30% reduction in ozone levels
failed to materialize, questions may arise about whether controlling
organic emissions will ever be effective in reducing ambient ozone.
Many of these questions would arise because a model which is not based
directly on cause-effect relationships between ozone and its organic and
NO precursors is used.
X
2.2.3 Use of Statistical Models
There are a number of statistical procedures which could conceiv-
ably be applied to attempt to describe ozone/precursor relationships at
20
specific sites. Examples of such procedures include simple log
linear regression equations, empirically derived envelope curves for
individual cities, stochastic models and complex multiple regression
systems. The use of statistical models offers two major advantages.
First, there exists a close relationship to the actual atmospheric data
upon which they are based, and second, because these models are rela-
tively simple, development and use costs are low. It is not the intent
of this document, however, to enumerate every possible approach which
could be pursued and its accompanying advantages/disadvantages. Instead,
a general discussion follows of the fundamental features that such ap-
proaches should incorporate.
37
-------
At the present time, statistical modeling of ozone/precursor
relationships can be viewed as both an art and a science. There is,
however, a considerable body of information on empirical relationships
between ozone and meteorological and emissions variables which is already
of) 01 po po
available to aid in the formulation of such models. '*' If a
statistical model is to be used for the purposes of control strategy
development and evaluation, then obviously the model must include vari-
ables that are subject to control by agency officials. Therefore, a
functional relationship between ozone concentrations and precursor
emissions and/or ambient concentrations must be a fundamental feature of
any statistical model used for control strategy purposes. As noted
earlier in this document, a number of studies have found that maximum
afternoon ozone concentrations are sensitive to prevailing meteoro-
logical conditions. Thus, some accounting for meteorological differences
should be incorporated into the statistical model, either directly or
indirectly, with particular emphasis on those times of the year and
those meteorological conditions which are conducive to high ozone levels.
In addition to the above considerations, some attempt should also be
made to account for the impact of the transport of ozone and its pre-
cursors as well as the impact of natural background.
Once a statistical model has been derived, it is relatively simple
and inexpensive to use, resulting primarily from the small input data
requirements. Perhaps the best example of small input data requirements
is the use of an empirically derived envelope curve where the only input
required is the ozone design value. As noted previously, this is one of
38
-------
the major advantages of statistical models. The problem with the use of
these models for control strategy development/assessment is the question
of reliability in extrapolating the model predictions beyond the range
of conditions contained in the data from which the model was derived.
This problem arises because the very purpose of a control strategy is to
change existing conditions such that emission levels are often substan-
tially reduced. Therefore, statistical models may be most appropriately
used to consider moderate movements from existing control states and to
indicate the direction of change rather than to make absolute numerical
predictions.
The motivation for the use of statistical models comes from the
assumption that a model derived from actual ambient data will enable one
to successfully predict ozone concentrations even when the understanding
of the complex photochemical process is incomplete. That these models
are derived from ambient data, however, is also a cause for concern.
Questions concerning instrument reliability, accuracy and representa-
tiveness (e.g., the absence of local NO interferences, etc.) must be
/\
resolved through a comprehensive quality assurance program. Similar
quality assurance procedures should also be applied to the emissions and
meteorological data. Where indicated, appropriate tests should be
performed for the basic assumptions in some of the statistical procedures,
such as linear regression, used to estimate model coefficients. The
most critical assumption in the use of a statistical model for oxidant
control strategy evaluation is that the functional relationship derived
from the base period data would hold for future time periods. If the
new control strategy drastically changes the emissions pattern from this
base period, then the error in the prediction of the change in ozone
level due to this new strategy is difficult, if not impossible, to
assess.
39
-------
3.0 Consideration of Transport and Natural Background
This section is organized in the following manner. First, possible
roles of transported ozone and precursors in contributing to high levels
of ozone observed in or near an urban area during the afternoon are
discussed. Second, means for measuring transported ozone and inter-
preting the measurements are described. Third, procedures for assessing
the impact of transport in an urban area and suggestions for incorpor-
ating transport into the methods described in Section 2.0 are presented
and illustrated. Finally, possible consequences of not considering the
roles of transport and background as contributors to the ozone problem
near large cities are addressed.
3.1 Possible Roles of Transported Ozone/Precursors in the Urban Oxidant
Problem
There are several possible means whereby transported ozone and/or
precursors could affect ozone concentrations observed in urban areas
during the afternoon.
3.1.1 Transport of Precursors
Observations concerning transported levels of organic pollutants
(including manmade and natural) are sparse. One means for estimating
concentrations of transported organic precursors is to measure levels of
organic pollutants at moderate distances downwind from urban areas.
Concentrations of organic pollutants measured 40 km downwind from St.
Louis and in Wilmington, Ohio, (about 60 km downwind of Cincinnati and
Dayton) revealed organic concentrations considerably less than those
found in urban areas. For example, organic concentrations downwind
from St. Louis are in the order of 20% of central city levels. Best
40
-------
estimates of transported organics based on these measurements would be
In the order of 0.1-0.3 ppmC. For example, highest 6-9 a.m. NMHC readings
40 km downwind from St. Louis appear to be slightly less than 0.3 ppmC.
Readings at sites which do not reflect the impact of any obvious urban
3
plume from an upwind city are in the order of 0.1 ppmC. The previously
described chemical kinetics model was used to estimate the impact of
such levels. As described in Reference 3, the impact of .1 ppmC of
transported organics was at most .02 ppm of ozone. However, the impact
was generally less than .01 ppm. Under some circumstances, transported
hydrocarbon concentrations of 0.26 ppmC were estimated to increase
maximum ozone concentrations by as much as .08 ppm. However, a more
typical impact was in the order of .03-.05 ppm. Available measurements
suggest that even in remote areas hydrocarbons emitted from vegetation
constitute but a small fraction of measured NMHC. Therefore, the impact
of naturally emitted organics on the formation of ozone in urban areas
is expected to be negligible. Scenarios which were tested with the
kinetics model also suggest that aldehydes which are transported from
upwind areas have a negligible impact on maximum ozone levels formed
3
downwind of a city.
Most available evidence suggests that ambient NO in areas not
X
within or immediately downwind from cities is at or below the detection
limits of commercially available instrumentation (.005 ppm). Hence,
the impact of NO which is presently transported over long distances on
A
the formation of ozone in urban areas is expected to be negligible.
Transported ozone is best considered as consisting of 2 layers; (1)
ozone transported within the morning mixing layer (i.e., "ground level
transport", usually within a few tens of meters of the surface) and (2)
ozone which is transported above the morning mixing height but below
41
-------
the afternoon mixing height (e.g., in a layer 100-2000 m aloft). This
second category of transported ozone is referred to as "transport aloft"
in the subsequent discussion. "Ozone transported at ground level" is
subject to surface reactions -ind scavenging by other species (e.g., NO)
emitted during the night. As a result of nighttime atmospheric stability,
"ozone transported aloft" does not come into contact with the earth's
surface, nor does it come into contact with scavengers emitted during
the night.
3.1.2 Transport of Ozone ,'., Ground Level
The chief impact of ozone transport near ground level is expected
to be the more rapid conversion of NO to NO,,. When incoming ozone near
the ground was simulated, using the kinetics model, the impact on
maximum ozone concentrations was generally found to be negligible. This
finding held true for incoming concentrations as high as .12 ppm. Only
when locally generated organic and NO precursor levels were decreased
X
to about .3 ppmC and .34 ppm respectively did the impact of transported
surface ozone become perceptible. Even in these extreme cases, the
incoming surface ozone was less than 20% additive (e.g., incoming 03 of
.12 ppm increased the downwind maximum 03 by .02 ppm). Thus the impact
of ozone transported overnight near the surface on maximum 0,, levels is
essentially negligible in the large urban areas for which the methods
described in Section 2.0 are most appropriate.
3.1.3 Transport of Ozone Aloft
A series of simulations was performed with the previously described
chemical kinetics model used to develop the EKMA. The kinetics model
was combined with the column model described in Section 2.2.1b to
simulate scenarios considering different dilution rates (resulting from
42
-------
increases in the atmospheric mixing depth during the course of the day),
different levels of NMHC, NO and NMHC/NO ratios. As described in
X X
Reference 3, all scenarios considered the impact of post-8 a.m. emissions
of NMHC and NO . A typical scenario is pictured conceptually in Figure 7.
X
Air trapped aloft overnight which contains relatively high ozone levels
is homogeneously mixed into an imaginary surface-based column of air as
the mixing depth lifts during the day as a result of surface heating.
The imaginary column of air is regarded as a reaction vessel in which
the kinetics model is applied. This reaction vessel is transported by
the mean wind in such a way that it reaches the edge of the most built-
up area by noon. Emissions into the column of air thereafter are assumed
to have a negligible impact, partly because of the very much lower
emissions assumed and partly because of the larger volume of air into
which these emissions are diluted. The results of these simulations are
reported in detail in Reference 3. Depending on the assumptions made,
ozone aloft was found to be generally between 20-70% additive. The
higher the dilution rate assumed, the greater the impact of ozone
aloft. If the local emissions decrease, the impact of ozone transported
aloft tends to increase as a result of less scavenging. This latter
observation implies the need to reduce local emissions and transported
ozone in order to most effectively reduce the urban ozone problem.
Simulations described in Reference 3, as well as simulations conducted
by Seinfeld,^ suggest that one of the impacts of post 8 a.m. emissions
is to diminish the importance of ozone transported aloft. However, it
should also be noted that post-8 a.m. emissions contribute to maximum
ozone levels as the result of increased synthesis which occurs.
43
-------
TIME
0600-0900
1200
MID AFTERNOON
(TIME OF MAX. 03)
NMHC AND NOX EMISSIONS)
Figure 7 . Role of ozone aloft on maximum sui face ozone levels: a conceptual view.
44
-------
Measurements conducted in remote locations suggest that natural
background ozone is about .04 ppm, primarily as a result of stratospheric
-i
sources. Combining this information with the previously described
simulation results suggests that the impact of natural ozone on peak
hourly ozone values in urban areas ranges from .01-.03, with .02 ppm
being most likely.
Summarizing the information presented in Section 3.1, it appears
that unscavenged ozone transported aloft is likely to be the component
of transport having the greatest impact on maximum afternoon ozone
levels observed downwind from cities. Measurements concerning trans-
ported organics are very sparse, but they indicate that these organics
may be in the order of 0.1-0.3 ppmC, with concentrations near 0.1 ppmC
typically found in rural areas upwind from major cities. Simulations
described in Reference 3 suggest the role of such levels would typically
be to contribute about .01 ppm to observed maximum 0~ levels near major
cities. The only significant source of natural background in urban
areas appears to arise from ozone transported above nocturnal surface-
based inversions. This ozone occurs as the result of slow diffusion
from the stratosphere, and may exert an impact of about .01-.03 ppm on
maximum urban surface concentrations of ozone.
45
-------
3.2 Measurement and Interpretation of Transported Ozone
In Section 3.1, it was concluded that unscavenged ozone remaining
aloft at night is the facet of transport likely to exert the greatest
impact on maximum ozone concentrations in urban areas, with a smaller
impact also arising from transported organics. While issues concerning
monitoring network design most suitable for estimating oxidant control
requirements are discussed more fully in Appendix A, it is appropriate
to enumerate some of the difficulties in measuring transported ozone and
in interpreting the measurements. The principal difficulty lies in
having to rely primarily on surface measurements to estimate ozone
levels aloft. As exemplified in Figure 8, there is a distinct difference
between ozone 2000 feet aloft and surface ozone which accompanies night-
25
time atmospheric stratification. A second complication is introduced
by photochemical synthesis which is most pronounced near the surface
where precursor sources are located. Figure 9 is contained in a review
by Vukovich and illustrates the respective roles of atmospheric mixing
9fi *}~]
and photochemical synthesis as the day progresses. ' To obtain
surface readings which are most indicative of ozone aloft, measurements
should be made upwind or outside of the urban plume after breakup of the
nocturnal radiative inversion, but prior to the time at which photo-
chemical synthesis results in higher ozone concentrations near the
surface than aloft. Because it is necessary to obtain a representative
indication of transported ozone, measurements of several hours duration
(i.e.., 3) are suggested. The exact time to make such measurements would
vary somewhat depending on local meteorological conditions. Data such
as those presented in Figure 9 suggest that surface measurements obtained
46
-------
UJ
z: en
o o
o s:
CQ
-
N-» *=C
CNI Q
U_
r-H CD
CNI
cn CD
cn
CD
o
CD
CD
i_n
CNI
CD
CD
CNI
CD
LO
CD
CD
O
LO
o
c
•H
>
(fl
Q
U-l
O
to
•H
m
QJ
Ml
C
cn
C
o
•H
C
OJ
o
C
o
o
01
c
o
N
o
0)
-------
3'iflo
3200
HID
OZONE, ;jg/rn3
Figure 9 Diurnal variation of the vertical distribution of ozone at Wilmington,
Ohio on 1 August 1974. These profiles have been smoothed using both the ascent
and descent portions of flights where possible. The indicated times are the mean
time (EOT) of each profile. Altitude is measured from the surface. (Unsmoothed
data are in Reference 27).
48
-------
between about 1100 and 1300 LOT would be most indicative of ozone levels
aloft. Therefore, in the absence of locally appropriate data on the
breakup of nocturnal inversions, surface measurements of ozone between
11-1300 LOT are suggested for use in estimating ozone which is advected
aloft. Some of the difficulties concerning the measurement of trans-
ported ozone could be circumvented by mounting an ozone monitor on some
tall structure, such as a TV tower. The monitor should be high enough
so that it samples air which is insulated from the surface at night.
Generally, elevations of 1000-2000 feet above the surface should suffice.
Siting of these monitors is described more fully in Appendix A, but as a
general guide, they should be located far enough upwind to minimize the
impact of pollutants which may be recirculated from the city under
review.
3.3 Procedures for Accounting for the Impact of Transport and Natural
Background in Oxidant/Precursor Relationships
The purpose of this section is to describe procedures for inte-
grating transport and natural background into ozone/precursor relation-
ships presented in Section 2.0. The role of transport and natural
background in estimating regulatory requirements will first be discussed
conceptually. This discussion will be followed by a more complete de-
scription of how transport and natural background can be considered in
photochemical dispersion models, EKMA, and rollback.
49
-------
Discussion
Figure 10 depicts the base state and the post-control state in a
hypothetical city. The .08 ppm reduction (i.e., .12 - .04) in the
upwind (transported) ozone concentrations pictured would have to result
from controls initiated upwind. The goal of local control programs
would be to ensure that the impact attributable to local emissions is
not greater than the difference between the NAAQS and the impact of
future transport in the urban area. If, for example, the impact of the
.04 ppm transported ozone concentration on the maximum afternoon concen-
tration of .08 ppm shown in Figure 10(B) were .02 ppm, the goal of the
city should be to reduce local emissions so that the net ozone produced
by the city does not exceed .06 ppm (i.e, .08-.02). If it is assumed
that incoming or transported ozone can never be less than natural background,
it is apparent that accounting for natural background should increase
the estimated local emission control requirements. However, it is also
likely that a portion of the .28 ppm ozone concentration shown downwind
of the hypothetical city in Figure 10(A) is attributable to the .12 ppm
shown as transported ozone. It should be noted that the impact of the
transported ozone shown in Figure 10(A) on maximum ozone observed downwind
from the city is likely to be considerably less than .12 ppm. This
less-than-100% additivity is largely due to chemical reactions which the
incoming ozone undergoes with precursors emitted within the city. If
transported ozone were reduced as shown in Figure 10(B), it is likely
that maximum downwind ozone would be reduced also (though to a smaller
extent). Therefore, the initiation of upwind control programs (thereby
reducing transport) should lower local control requirements. Conversely,
if transport remains constant, control requirements are greater, since
50
-------
(A) BASE STATE
WIND
(B) CONTROLLED STATE
WIND
O 012 ppm 03
CITY
0.04 ppm03
(NATURAL BACKGROUND)
CITY
Oo28p|>m 03
0.08 ppm 03
Figure 1 0. Conceptual view of base and controlled states.
51
-------
local controls alone must be used to reduce the ozone design value
(which reflects both local synthesis and transport) to .08 ppm. Since
the goal of control programs is to reduce ozone design values to .,08 ppm,
and it is unlikely that this design value will occur immediately upwind
of a second city after controls have been implemented in upwind cities,
assumption of post control transport levels which approach natural
background ozone concentrations (.04 ppm) may frequently be appropriate.
Potential implications of the apparently opposite impacts of natural
background and reducing transport will be elaborated upon in Section 3.4.
Because of the potential role played by reducing transport through
upwind controls, use of the second high hourly ozone concentrations as
the ozone design value may not always result in the highest estimated
control requirements. For example, if there were little transport
occurring on the day with the fifth highest hourly ozone value, more
local controls may be needed to attain the NAAQS than if there were a
significant contribution from transport on the day with the second
highest hourly value. Therefore, it may sometimes be adviseable to
review data from several days with high ozone to ensure that calculated
control requirements are sufficient to meet the NAAQS.
3.3.1 Photochemical Dispersion Models
Transport is considered in models by assuming boundary and initial
conditions which coincide with measurements or estimated transport.
Incorporation of these estimated or measured values in the models is a
straightforward process which proceeds within the computer software for
the model. The impact, of transport of manmade pollutants per se can be
simulated by a second model run in which transport has been reduced to
natural background. As described in Section 3.2, a far more difficult
52
-------
problem arises in trying to determine what boundary and initial values
to enter as input to the model being utilized.
3.3.2 EKMA
Transport can be readily considered by using the city-specific
isopleth version of EKMA described in Section 2.2.1b. Transported ozone
or precursor concentrations are just two of several specific assumptions
which can be made in utilizing the kinetics model underlying EKMA to
generate city-specific isopleths. If transported pollutants are assumed
not to change, the city-specific isopleths are applied exactly as de-
scribed in Section 2.2.1b. Reference 18 specifies how the transported
ozone concentration may be input into the kinetics model.
If a reduction in transported pollutant concentrations occurs
concurrently with emission reductions (as shown for the hypothetical
city in Figure 10), it is generally necessary to conduct separate model
runs for the base and controlled states. For reasons discussed in
Appendix B, it frequently shall be adequate to exercise this procedure
considering only changes in transported ozone concentrations. The
following is a step-by-step procedure which should be used in the de-
termination of emission reduction requirements based on city-specific
isopleths that consider reductions in transported ozone:
1. Gather the data described in Appendix B.
2. Using the data from Step 1, operate the kinetics model to
generate an isopleth for the design value of ozone. Unless locally
applicable information to the contrary is available, this isopleth
should be based on the assumption that the transported ozone concen-
tration is equal to the average from 11 a.m. to 1 p.m. LOT of the upwind
ozone concentration on the day the design value occurs.
53
-------
3. Using the data from Step 1, operate the kinetics model to
generate an ozone isopleth for CL = .08 ppm. This isopleth should
ordinarily be based on the assumption that the transported ozone concen-
tration is equal to natural background (.04 ppm).
4. Determine the 6-9 a.m. NHMC/NO ratio using procedures
X
outlined in Section 2.2.la of this report.
5. Using the ozone-precursor graph generated in Step 2,
determine the location of the intersection of the line for the NMHC/NO
A
ratio of Step 4 with the isopleth for the design value of ozone. Label
the coordinates of this location as the "base initial NMHC concentration1
and the "base initial NO concentration."
A
6. Estimate the percentage reduction of NO that will occur
X
by the time the ozone NAAQS is met. Reduce the "base initial NO
A
concentration" accordingly to estimate the "controlled initial NO
A
concentration."
7. Determine the location along the isopleth for .08 ppm
ozone which has the appropriate "controlled initial NO concentration."
A
The coordinates of this location are the "controlled initial NMHC con-
centration" and the "controlled initial NO concentration."
X
8. Hydrocarbon emission control requirements are determined
by this formula:
F "controlled initial NMHC concentration^
% Reduction Req'd = 1 - base initial NMHC concentration" x 100
Note that it is possible to interchange the roles of NMHC and NO in
A
Steps 6-8 of the above procedure. If one is solely interested in
assessing the impact of changing transport on local emission control
requirements, all other variables (e.g., diurnal emission patterns) must
stay constant. It is also possible, however, to assess the impact of
54
-------
several concurrent changes on local emission control requirements using
the above 8-step procedure. For example, one could generate the .08
isopleth described in Step 3 assuming reduced transport and a different
diurnal emission pattern.
Example
Figure 11 is used to illustrate how city-specific isopleths may be
used in EKMA to estimate the implications of concurrent changes on
precursor control requirements. The solid isopleths in Figure 11
depict ozone as a function of initial NMHC and NO in the presence of
X
ozone which has been transported aloft overnight. The dashed line
depicts the .08 ppm isopleth as a function of initial NMHC and NO , when
A
transported ozone has been reduced to natural background levels as the
result of upwind control programs. The amount of organic control needed
(keeping NO constant) to attain the .08 ppm standard is estimated by
X
line AB'. In the example shown,
ANMHC 1'02 ppmC' B'NMHC = .27 ppmC and B = .18 ppmC.
__
AB1 = 1.02 = .74 - 74%
If transported ozone had remained unchanged, the amount of organic
control needed to attain the oxidant standard would be represented by
line AB. In the example shown,
lJD2_-__il8
AB = 1.02 = .82 = 82%
55
-------
o
-WITH PRI:SI;;JT
TRANSPORT
X J = 10:1 —
1.8
Figure 11. Procedure for considering two or more impacts of control
strategies simultaneously using city-specific isopleths.
56
-------
In addition to the use of EKMA for determining emission control
requirements, Section 2.2.1 of this report also discusses its appli-
cation in determining the reductions in peak ozone concentrations
corresponding to a given reduction in precursor emissions. If trans-
ported ozone concentrations during the base period are assumed to be at
natural background levels, it is possible simply to generate a set of
isopleths all based on the transport of .04 ppm. These isopleths may
then be used as discussed in Section 2.2.la to determine the impact of a
given emissions reduction. However, if transported ozone concentrations
during the base period exceed natural background concentrations, such a
determination is made more difficult by the need to estimate the con-
centration of ozone that is transported at the time the given emissions
reductions have occurred. If estimates can be made of an appropriate
transported ozone concentration, it is then possible to operate the
kinetics model in order to estimate the ozone concentration that would
result, given the appropriately reduced initial precursor concentrations
and the estimated transported ozone concentration. As stated previously,
it is also possible to estimate the reductions in downwind ozone attrib-
utable to local emissions simply by assuming constant transport levels,
although such a procedure tends to overestimate the ozone concentration
that may be expected.
The following step-by-step procedure should be used in determining
the impact of a given emission reduction (based on the assumption that
current transport levels are greater than natural background levels):
1. Steps 1, 2, 4 and 5 of the procedure for determining
emission control requirements using city-specific isopleths should be
done to determine the "base initial NMHC concentration" and the "base
57
-------
initial NO concentration."
A
2. Reduce the base initial precursor concentrations by the
percent emission reductions that apply to the respective precursors in
order to determine the respective "intermediate initial precursor con-
centrations."
3. Estimate the transported ozone concentration that may be
expected at the time the given emission reductions have occurred. The
intermediate level of transport may be assumed equal to the base level
of transport or, preferably, will be some reasonable concentration
between the base transported ozone concentration and the natural back-
ground concentration.
4. Operate the kinetics model in order to determine the peak
ozone concentration that corresponds to the "intermediate initial pre-
cursor concentrations" from Step 2 and the transported ozone concen-
tration from Step 3. This peak ozone is the second highest concen-
tration of downwind ozone that may be expected when the given emission
reductions have occurred.
3.3.3 Standard Isopleth Version of EKMA and Rollback
In order to consider the impact of altering transport on the effect
of local emission controls using the standard isopleths or rollback, it
is first necessary to estimate the additivity of the transported ozone
to maximum ozone concentrations which are synthesized by locally emitted
precursors. As described earlier (and in greater detail in Reference 3),
additivity of transported ozone depends on the interaction of a number
of factors. Hence, the best approach for estimating the additivity of
transported ozone is through simulations with an atmospheric diffusion
model or with the column and kinetics models underlying EKMA. However,
58
-------
if one has the capability for implementing either of these procedures,
it makes little sense to use the standard isopleths or rollback. There-
fore, if available resources do not permit the use of photochemical dis-
persion models or the city-specific version of EKMA, less satisfactory
means for considering transport will, by necessity, have to be incorpor-
ated in the standard isopleth or rollback approaches. It is suggested
that Section 5.0 in Reference 3 first be reviewed to gain a more com-
plete appreciation of how factors such as prevailing NMHC/NO ratio,
A
local precursor levels, dilution rate, and diurnal emission patterns
affect the additivity of transported ozone. The range of additivity
reported in Reference 3 was roughly .20 - .70. Table 3 is adapted from
Reference 3 and indicates the impact of various factors on additivity.
The factors are listed in order of their impact on additivity (greatest
impact is listed first). The additivity of transported ozone can be
estimated subjectively by beginning at the midpoint of the observed
.20 - .70 range of values (i.e., .45) and adjusting this value upward or
downward according to the nature of the four factors presented in
Table 3. For example, if the dilution rate were low (e.g., < 13%/hour),
the city of interest were a major one with many sources, the prevailing
NMHC/NO ratio were low (e.g., 6:1 or less), and high ozone occurred on
X
days with atmospheric stagnation (i.e., so that air parcels tended to
stay near areas of high emission density all day), one would choose an
additivity value (A) near the low end of the scale (e.g., A = .20). In
some cases, several of the factors in Table 3 may be high while others
are low. In this event, greater weight should be attached to those
factors at the top of the table in choosing a value for the additivity
of transported ozone. Once the additivity of transported ozone has
59
-------
TABLE 3. IMPACT OF VARIOUS FACTORS ON THE ADDITIVITY (A) OF
TRANSPORTED OZONE TO MAXIMUM OZONE CONCENTRATIONS
IN URBAN AREAS
Factor
1) Dilution Rate (i.e., the
extent and rate at which
the diurnal mixing depth
increases).
2) Quantities of locally
emitted precursors
3) NMHC/NOV Ratio
X
4) Importance of post 9 a.m.
emissions (This reflects
both diurnal emission patterns
and the larger atmospheric
dilution capacity which
generally occurs during the
mid-morning and afternoon)
Factor Value
Relatively High
(e.g., > 13%/hour)
Relatively Low
(e.g., small city
^ 200,000)
Relatively Low
(e.g., < 6:1)
Relatively High
(e.g., significant
NO emissions in the
afternoon such as
would occur if an
air parcel remained
within the city
limits in the after-
noon during a stag-
nation period)
Additivity
Relatively High
Relatively High
(> .45)
Relatively High
(> -45)
Relatively Low
60
-------
been estimated using this rather subjective procedure, the second high
i
hourly ozone concentration observed during the base period should be
adjusted downward in accordance with the following formula.
<°3>ADJ = ^design ' A
where (Oo)nni is the adjusted (reduced) ozone concentration to
be used in estimating control requirements;
(CL) , . is the 2d high hourly ozone concentration observed
o u c o i y' i
during the base period.
A is the additivity factor for transported ozone.
T is the concentration of transported ozone estimated for the
base period.
T.T is the future concentration of transported ozone which,
presumably has been reduced from present levels as the result
of upwind controls. TV is frequently assumed to approach
natural background.
Controls required to attain a specified air quality goal, x1, may
be estimated with rollback using the following formula:
R = K03)ADJ-x']
(6)
[(03)ADJ - (A) (Tf)]
Example
Given: Design Value for 0_ = .28 ppm. Present transported
ozone is estimated to be .12 ppm. It is assumed that
upwind controls will reduce future levels of transport
to .04 ppm.
61
-------
Find: Amount of reduction of local organic precursors needed
to reduce the 2d high hourly ozone concentration to
.08 ppm.
Solution:
(1) Since nothing is given about any of the factors
influencing additivity, assume a value of A =
.45 (i.e., the middle of the observed range of
values) .
(2) From Equation (5) ,
(03)ADJ = .28 - (.45)(.12 - .04)
(03)ADJ = .24 ppm.
(3) Using Equation (6), the required reduction is
[.24 - .08]
R = __ -- - _ - = .73 or 73%
[.24 - (.45)(.04)] —
Note that in accordance with Table 3, the value of "A" in the denomi-
nator in Equation (6) should probably be somewhat higher than ".45." As
will be illustrated in the next section, however, this is probably an
unnecessary refinement in cities presently experiencing a serious oxidant
problem.
Using the standard isopleth version of EKMA, transport is considered
by first reducing the ozone design value in accordance with Equation (5).
Control requirements are estimated by entering the isopleth diagram
(e.g., Figure 12) at the intersection of the [0-5 LDi isopleth and the
prevailing NMHC/NO ratio line. Control requirements needed to attain
A
the standard are calculated by estimating the reductions needed to reach
the isopleth which is .08 — A(b). For example, if the additivity of
transported ozone were .45 and natural background were .04 ppm, control
62
-------
03
E
0.
Q.
et
LU
O
l/l
S-
cu
o.
o
l/l
-a
s-
ro
T3
C
rC
+->
OO
O)
CT,
c:
S-
o
Q-
O
£T
O
J_
a>
CO
c:
o
o
C\J
OL
uidd
-------
requirements would be estimated by NMHC and/or NO reductions needed to
X
reach the .06 ppm isopleth. This procedure is illustrated in Figure 12.
As previously discussed, transport can not be accounted for as
readily with the standard isopleths or with rollback as it can using
models or the city-specific version of EKMA. It is not possible to
readily account for the impact of transported precursors using the
standard isopleths or rollback. Neither is it possible to account for
the role of other changes, such as changes in diurnal emission patterns,
using these procedures.
3.4 Data Limitations and Resulting Consequences
Section 3.0 has described what appears to be the role of transport
in the urban ozone problem. Procedures for incorporating transport into
the methods presented in Section 2.0 have also been described. However,
when attempts were made to use existing air quality data to estimate
transported ozone, difficulties were frequently encountered. These
difficulties arose principally because of the lack of upwind data or
inappropriate siting of upwind instruments. Such problems are under-
standable, since the primary purpose of existing monitors is seldom to
obtain an estimate of representative ozone concentrations entering a
city. Therefore, it may not always be feasible to incorporate transport
considerations until monitoring programs have been developed. Hence it
is pertinent to examine how ignoring transport might affect estimated
control requirements. There are two situations which should be considered.
The first occurs if transported ozone concentrations remain unaltered.
If this situation prevails, ignoring transport will likely cause one
64
-------
to underestimate local control requirements using the simple approaches
described in Section 2.2. The reason for the low estimate is that, in
order to attain the NAAQS, local emissions would have to be reduced such
that the sum of locally synthesized ozone and the combined impact of
manmade and natural transported ozone is .08 ppm or less.
The second (and more important) situation occurs if transported
ozone is diminished concurrently with local emission control programs.
Ideally, this reduction would result in ozone transported to downwind
cities being reduced to levels approaching natural background. Since
transport and natural background affect maximum ozone via a similar
mechanism (i.e., from aloft), to be consistent, natural background
should be ignored if transport is ignored. For the situation in which
transport is actually reduced, ignoring both transport and natural
background may frequently result in calculated control requirements
which are similar to those in which both transport and natural back-
ground are considered. These assertions are illustrated in the following
example.
Example
Four hypothetical situations are presented:
Case I -- A moderate size city downwind from a major city in an
area where diurnal variations in the atmospheric mixing depth are
substantial. In this case, for illustrative purposes, transported
ozone will be assumed to be .20 ppm and 60% additive.
Downwind concentrations from the moderate size city will be
assumed to be .24 ppm, with NMHC/NO = 10:1
X
Case IA -- Same assumptions as Case I, except that the problem
is more moderate, with downwind and upwind ozone of .16 ppm and
65
-------
..14 ppm respectively.
Case II -- A major city within a megalopolis in an area with
moderate diurnal variations in atmospheric mixing depth. In this
case, transported ozone is assumed to be .12 ppm and 40% additive.
Downwind concentrations are assumed to be .24 ppm. NMHC/NO -: 10:1
A
Case III -- A major city which is relatively isolated in an
area with moderate diurnal variations in atmospheric mixing depth.
In this case, transported ozone is assumed to be .10 ppm and 20%
additive. Downwind concentrations are assumed to be .18 ppm
with NMHC/NO = 10:1.
A
Because the computer program for generating city specific isopleths
was not available at the time this report was being prepared, calcula-
tions for the above cases were made using the standard isopleth version
of EKMA and assumptions about additivity. As indicated in Section 3.3,
this is not the best way to account for transport in EKMA.
For Case I, the following data apply,
- Present transported ozone, T = .20 ppm
- Future transported ozone (natural background), Tf = .04 ppm
- Design value of ozone, (03) design = -24 ppm
- Assume that the fraction of additivity, A , has already been
determined to be .60, as given.
Solution
(1) Adjust ozone design value downward to reflect the impact of
transported ozone
<°3>ADJ = ^design ' A
66
-------
= .24- (.60)(.20)
(03)ADJ = .12 ppm
(2) Note that if, in the future, natural background were only 60%
additive, local controls would have to ensure that synthesis of local
emissions were less than
.08 - (.60)(.04) = .06 ppm
(3) Using the standard isopleth version of EKMA, enter Figure 12
(Section 3.3) and find the intersection of the .12 ppm ozone isopleth
with the NMHC/NOv = 10:1 ray.
A
(4) Note the NMHC and NOV coordinates of this point are (.32 ppmC,
X
.03 ppm). If NO is kept constant, and only NMHC is reduced, the
A
corresponding (NMHC, NO ) coordinates at the .06 ppm ozone isopleth are
A
(.10 ppmC, .03 ppm).
.32 - .10
(5) Required NMHC control = .32 = .69 = 69%
(6) Ignoring transport and background, Figure 1 would be entered at
the intersection of the .24 ppm ozone isopleth with the 10:1 NMHC/NO
X
ratio line, and the percent reduction needed in NMHC levels
estimated is: 00 oc
.OO - . £0
Required Control - .88 = .72 - 72%
Similar procedures can be followed for Cases IA, II and III. The
sensitivity to transport considerations for each of the three cases is
depicted in Table 4. In these examples, NO was assumed to remain
X
constant, and a natural background concentration of .04 ppm was assumed
for ozone.
It can be concluded that for the cases pictured in Table 4, the
calculated difference in needed controls is practically imperceptible,
except in the cases where a moderate size city is impacted by a major
city upwind such that measured upwind and downwind concentrations are
almost identical.
67
-------
TABLE 4. THE IMPACT OF IGNORING TRANSPORT ON ESTIMATED ORGANIC
CONTROL REQUIREMENTS
Considering Ignoring
Transport and Transport and
Case Natural Background Natural Background
I 69% 72%
IA 31% 69%
II 76% 75%
III 71% 70%
68
-------
4.0 Caveats
In addition to the limitations already described in connection with
each analytical approach, there are several which apply generally to all
simple approaches described. Some of these limitations apply to photo-
chemical dispersion models as well. Limitations are of three categories:
1) inability to make accurate, appropriate measurements; 2) inability
to account for measured phenomena properly in analytical approaches
(models); 3) limited availability of data for verification of the ana-
lytical approaches.
The EKMA, rollback and statistical approaches all require the use
of observed ambient ozone data to relate oxidant to precursors. Thus, a
key limitation arises if ozone monitors are not properly sited. Selecting
the proper location for instruments depends on local meteorology and the
configuration of sources. For economic reasons, ozone instruments are
frequently co-located with instruments measuring other pollutants.
Ideal siting criteria for ozone instruments are not, as a rule, entirely
consistent with the criteria for other instruments. Even if the instrument
locations were consistent with available guidance, however, it is conceivable
that the concentration measured at a point in space would be less than
the actual maximum 03 concentration. Further, if there are any local
sources of precursors, the monitoring data may not represent concentrations
predicted over the somewhat larger spatial scales assumed in available
models. These limitations have implications concerning control require-
ments estimated by using the simplified procedures described herein and
with regard to attempts to validate both complex and simple models.
Even the most sophisticed Eulerian grid models do not attempt to predict
o
pollutant concentrations for grid squares less than 1 km in area.
69
-------
Methods such as the EKMA and some statistical approaches also rely
on measured NMHC data. Existing instrumentation for NMHC produces
estimates which are subject to large errors. Because these errors
appear random and less severe on days experiencing high levels of
ozone, use of robust estimates like median 6-9 a.m. NMHC values on high
0., days reduces this error considerably. However, using such a procedure
reduces some of the advantage of the cause-effect nature of the EKMA.
All methods are limited by inadequate emission inventories. If known
emissions are responsible for only 50% of the ambient organic concentra-
tions found in urban areas, reducing the known emissions a given amount
will not have the impact anticipated b.y the analytical procedures
described herein. The impact of transported organic material is not
very well documented. This problem arises, in part, because of the
difficulties in measuring low levels of organics and various types of
organic pollutants. If the major impact of transported material comes
from ozone advected aloft, as suggested in Section 3.0, utilization of
surface measurements to derive numbers for use in analytical procedures
has uncertainties associated with it.
An important set of limitations of the simple models arises from
the lack of spatial resolution. Thus, these procedures do not, in
their present form, enable evaluation of strategies which result in
different spatial emission patterns. Further, the 1:1 relationship
assumed between ambient precursor concentrations and emissions does not
enable one to consider the relative effectiveness of controlling parti-
cular source categories. The simple procedures do not readily allow
consideration of other indicators of the oxidant problem, such as
population exposed to high levels, frequency of high concentrations,
70
-------
etc. Such other indicators may prove useful in assessing the benefits
accompanying precursor controls. All of the models proposed are limited
to urban areas. The rural problem and the impact of ozone and/or
precursors on areas whose local emissions are small is not well described
by the EKMA or rollback approaches (or by photochemical dispersion
models either, for that matter). Also, the impact of urban control
strategies on the transport of ozone and precursors overnight is not
considered. Certain smog chamber experiments, for example, suggest that
drastic control of organic precursors may increase the transport of N0?
po
downwind from cities. In addition, it may be difficult to simulate
complex meteorological phenomena such as land-sea breezes and urban heat
island effects using the procedures described herein. Difficulties of
this sort stem both from data availability and from simplifications
which have been made in computational schemes.
Perhaps the most important shortcoming of the methods described in
Section 2.0 is the limited extent to which their accuracy has been
assessed. Major efforts are underway at EPA to verify several Eulerian
and Lagrangian photochemical dispersion models, using the St. Louis RAPS
data base. As reported in Reference 3, many uncertainties are intro-
duced when one attempts to compare EKMA and rollback with Los Angeles
trend data. Further investigation of the Los Angeles data is underway.
Another possible way to assess the accuracy of the EKMA approach would
be to use ambient NMHC, NO and ozone data collected in and near a city.
X
This may be very difficult to do in practice, however, since the iso-
pleth curves in EKMA are based on the premise that meteorology does not
71
-------
vary. If different meteorological conditions occur from day to day,
there is little reason to expect the isopleths to agree with the ob-
servations in the data base. The reason for this is that changing ozone
levels are not solely a function of changing precursor levels, but of
changing meteorological conditions as well.
Given the lack of extensive observations with which to compare
estimates obtained with the procedures discussed previously, it is
difficult to generalize about the accuracy of these procedures. One
means for estimating the uncertainty inherent in various procedures is
to test the sensitivity of estimates to errors in input values. If the
potential error in input variables can be estimated, the degree of
uncertainty in estimated ozone concentrations can also be assessed. The
more fundamental question about the appropriateness of an analytical
procedure as applied to a specific area can only be completely resolved
after verification studies have been conducted, however. Errors asso-
ciated with key input variables to photochemical dispersion models have
been summarized in a review by Seinfeld for the Los Angeles area. This
information is then used to infer that errors in Eulerian photochemical
dispersion model estimates of absolute ozone concentrations are less
than + 50% if input variables are known with a reasonable degree of
accuracy. Comparisons between single cell and multi-cell models (in the
vertical) suggest that single cell models (such as the column model in
EKMA) may be subject to an additional error as high as 20%. This
information implies that errors associated with predictions of absolute
levels of ozone using EKMA are less than +. 70%. Such numbers, of course,
are extremely rough estimates. It is likely that careful collection
72
-------
of appropriate input data would result in more accurate predictions.
Conversely, if key variables such as emission inventories are poorly
known, errors can be much higher. Since it is not possible to apply
rollback in anything but a relative sense, it is not possible to comment
on the absolute accuracy of that approach. Neither is it possible to
comment on the accuracy of statistical approaches using anything other
than observed data.
The degree of error or uncertainty can be appreciably reduced when
models are applied in a relative sense. Additional sensitivity studies
conducted utilizing the kinetics model underlying EKMA bear this assertion
out. Reviewing the results of the sensitivity studies reported in
Reference 17, uncertainties concerning reactivity, diurnal emission
patterns, sunlight intensity and atmospheric dilution rate result in
less than a 10% difference in estimated organic control requirements
(with constant NO ). As shown in Reference 17, the degree of uncer-
X
tainty concerning control requirements is greater if there is uncer-
tainty concerning prevailing NMHC/NO ratios. Errors in estimated NMHC
X
control requirements in the order of +_ 25% are implied if NMHC/NO
A
ratios lie somewhere between 5:1 and 20:1. Since rollback makes no
attempt to consider prevailing NMHC/NO ratios, atmospheric chemistry or
A
changes in transported ozone concentrations, estimates obtained using
this approach may be subject to somewhat greater error than those which
attempt to consider these factors (e.g., models, EKMA).
A final caveat to keep in mind is that estimates obtained with
simple approaches may be subject to additional errors if they are used
to assess the impact of strategies which are clearly inconsistent with
73
-------
their underlying assumptions. Thus, if the standard isopleth version of
EKMA or rollback were used to assess the impact of strategies involving
a substantial rearrangement of source configurations, additional errors
could occur.
For the reasons cited above, there are uncertainties associated
with the estimates obtained by using simple approaches and complex
models as well. However, these uncertainties are more of degree than of
direction. For example, all methods indicate that substantial reductions
in organic precursors are needed to reduce ozone levels appreciably. In
addition, there is no evidence that controlling organics and NO at the
X
NMHC/NO ratios which appear to exist in most cities will not be bene-
X
ficial in reducing maximum levels of ozone. Therefore, the procedures
described herein can be regarded .as approximations, subject to some
error, which may nevertheless be useful in assessing the magnitude of
the oxidant problem.
74
-------
5.0 Acknowledgements
The principal authors of this document are Edwin L. Meyer, Jr.,
John E. Summerhays, and Warren P. Freas, all of the Monitoring and Data
Analysis Division, Office of Air Quality Planning and Standards, U. S.
EPA. The assistance of members of the EPA Appendix J Working Group and
the Appendix J Review Group for their efforts in reviewing and commenting
on ideas and earlier drafts leading to the development of this document
is also acknowledged. These comments, as well as a number of meetings,
provided a very beneficial exchange of information. Particular note is
taken of the close support the Office of Air Quality Planning and Standards
has received from the staff of the Environmental Sciences Research
Laboratory, EPA, since the inception of the Appendix J working group.
Much of the material in this document reflects an attempt to apply in a
regulatory mode developmental work which has been performed inhouse or
under contract for ESRL.
75
-------
6.0 References
1. "Requirements for Preparation, Adoption, and Submittal of Implementation
Plans." Title 40, Code of Federal Regulations: Part 51.
36FR15502 (August. 14, 1971).
2. Air Quality Criteria for Nitrogen Oxides. AP-84, U.S. EPA, Research
Triangle Park, North Carolina, January 1971, pp. 4-4, 5.
3. U.S. EPA, OAWM, OAQPS, MDAD: Procedures for Quantifying Relationships
Between Photochemical Oxidants and Precursors: Supporting Documentation.
EPA-450/2-77-021b, in press.
4. N. deNevers and J. R. Morris, "Rollback Modeling: Basic and Modified."
JAPCA, 25, 943 (1975).
5. J. E. Summerhays, "A Survey of Applications of Photochemical Models."
International Conference on Photochemical Oxidant Pollution and Its
Control, Proceedings: Volume II, EPA-600/3-77-001b, U.S. EPA, Research
Triangle Park, North Carolina, January 1977, p. 805.
6. J. H. Seinfeld, "International Conference on Oxidants 1976--Analysis
of Evidence and Viewpoints. Part VI. The Issue of Air Quality
Simulation Model Utility." EPA-600/3-77-118, U.S. EPA, Research Triangle
Park, North Carolina, November 1977.
7. Assessing Transportation-Related Air Quality Impacts. Transportation
Research Board Special Report 167, National Academy of Sciences,
Washington, D.C., 1976, pp. 8-20.
8. B. Dimitriades, "Oxidant Control Strategies, Part 1: An Urban Oxidant
Control Strategy Derived from Existing Smog Chamber Data." Environmental
Science and Technology, 11, 80 (1977).
9. B. Dimitriades, "An Alternative to the Appendix J Method for Calculating
Oxidant-and N0?-Related Control Requirements." EPA-600/3-77-001b,
ibid., p. 871.
10. G. Z. Whitten and H. Hogo, Mathematical Modeling of Simulated Photo-
chemical Smog. EPA-600/3-77-011, U.S. EPA, Reasearch Triangle Park,
North Carolina, January 1977.
11. M. C. Dodge, "Combined Use of Modeling Techniques and Smog Chamber
Data to Derive Ozone-Precursor Relationships." EPA-600/3-77-001b,
ibid., p. 881.
12. B. Dimitriades, "Effects of Hydrocarbon and Nitrogen Oxides on Photo-
chemical Smog Formation." Environmental Sciences and Technology
6, 253 (1972).
13. "Oxidant/Hydrocarbon Monitoring." Strategy Issue Papers (Draft),
Standing Air Monitoring Work Group, U.S. EPA, Research Triangle Park,
North Carolina, October 1976.
76
-------
14. Guidance for Air Quality Monitoring Network Design and Instrument
Siting (Revised).OAQPS No. 1.2-012, U.S. EPA, Research Triangle Park
North Carolina, September 1975.
15. E. L. Martinez and E. L. Meyer, Jr., "Urban-Nonurban Ozone Gradients
and Their Significance." Proceedings, Ozone/Oxidant Interaction with
the Total Environment Speciality Conference, Air Pollution Control
Association, March 1976.
16. Report of the Air Monitoring Siting Workshop. U.S. EPA, Las Vegas,
Nevada, July 1976.
17. M. C. Dodge, Effect of Selected Parameters on Predictions of a
Photochemical Model. EPA-600/3-77-048, U.S. EPA, Research Triangle
Park, North Carolina, June 1977.
18. G. Z. Whitten, H. Hogo, User's Manual for a Kinetic and Ozone Isopleth
Plotting Package, prepared under EPA Contract 68-02-2428, U.S. EPA,
Research Triangle Park, North Carolina, in press.
19. J. C. Trijonis, et al., "Emissions and Air Quality Trends in the
South Coast Air Basin." EQL Memo No. 16, California Institute of
Technology, Pasadena, California, January 1976.
20. Assessing Transportation-Related Air Quality Impacts, ibid., pp. 46-62.
21. T. D. Hartwell and H. L. Hamilton, Examination of the Relationships
Between Atmospheric Oxidant and Various Pollutant and Meteorological
Variables. EPA Contract 68-02-1096, Research Triangle Institute,
Research Triangle Park, North Carolina, December 1975.
22. E. L. Meyer, Jr., e t a1., "The Use of Trajectory Analysis for Determining
Empirical Relationships Among Ambient Ozone Levels and Meteorological
and Emission Variables." EPA-600/3-77-001b, ibid., p. 903.
23. F. L. Ludwig, et aj_., "Important Factors Affecting Rural Ozone Con-
centration." International Conference on Photochemical Oxidant
Pollution and Its Control, Proceedings: Volume I, EPA-600/3-77-001a,
U.S. EPA, Research Triangle Park, North Carolina, January 1977, p. 425.
24. Effectiveness of Organic Emission Control Programs as a Function of
Geographic Location. U.S. EPA, Office of Air Quality Planning and
Standards, Research Triangle Park, North CArolina, April 1977.
25. C. E. Decker, et al., Ambient Monitoring Aloft of Ozones and Precursors
Near and Downwind of St. Louis. EPA-450/3-77-009, U.S. EPA, Research
Triangle Park, North CArolina, January 1977.
26. E. Robinson, F. Vukovich and D. H. Pack, "International Conference on
Oxidants 1976--Analysis of Evidence and Viewpoint. Part V. The Issue
of Oxidant Transport." U.S. EPA, Research Triangle Park, N.C. November 1977
77
-------
27. Investigation of Rural Oxidant Levels as Related to Urban Hydrocarbon
Control Strategies. EPA-450/3-75-036, U.S. EPA, Research Triangle
Park, North Carolina, March 1975.
28. H. J. Jeffries, et al., Outdoor Smog Chamber Studies: Effect of
Hydrocarbon Reduction on Nitrogen Dioxide. EPA-650/3-75-001, U.S. EPA,
Research Triangle Park, North Carolina, June 1975.
78
-------
APPENDIX A
MONITORING NETORK DESIGN AND INSTRUMENTATION
A comprehensive discussion of this topic is beyond the scope of
this document. Nevertheless, proper design, collection and interpre-
tation of ambient monitoring data is a crucial part of the previously
discussed approaches for relating ozone to precursors and for consider-
ing the role of transport. The purpose of this Appendix is to outline
considerations for designing an appropriate air quality monitoring
network to enable assessment of control requirements for oxidant. In
addition, some limitations in available instrumentation and their re-
sulting implications are identified.
A.I Network Design
This discussion will proceed by first outlining appropriate siting
guidance associated with the least data-intensive method (linear rollback)
and then build upon that discussion with advice concerning additional
data which are desirable for other approaches (e.g., the Empirical
Kinetic Modeling Approach with transport considerations). The funda-
mental piece of ambient air quality information required by linear
rollback (and all other simple approaches) is a good estimate of the
design value of ozone.* Hence, it is necessary to exercise great care
in siting the ozone monitor(s) so that there is a reasonable chance of
observing the higher 0, levels near the city. The prime sites should be
in areas which are most frequently downwind on summer days between mid-
morning and mid-afternoon.
* The highest second highest hourly concentration observed at all
monitoring sites during the base period.
A-l
-------
"Downwind" should be determined from meteorological data which are not
unduly influenced by very localized phenomena such as land-sea breezes.
Each ozone site should preferably not be located within 200 meters of
major ground level sources of NO. The monitoring site should not be
obstructed by buildings or have any restrictions to uniform flow in at
least 3 of the 4 major directions. The monitoring instrument itself
should be a reference method or equivalent. The distance of the monitor
from the center of the urban core* should be determined by how far a
parcel of air over the urban core might be expected to travel between 6-
9 a.m. to mid-afternoon with summmer/fall winds in the prevailing
direction. Therefore, the most appropriate distance and orientation
would vary according to the local meteorology. As a general guide, the
Standing Air Monitoring Work Group (SAMWG) has recommended that two
permanent ozone monitoring sites be established in each city to be
reviewed. On days with light winds, one site should be within 15-30 km
of the center of the urban core in the predominantly downwind direction
o
(afternoon). A second monitoring site might be located on the fringes
of the urban core in the predominate summer/fall downwind direction. A
review of air quality data collected in St. Louis as part of the Regional
Air Pollution Study (RAPS) has revealed that highest 03 concentrations
may occur very near or within the urban core during atmospheric stag-
3
nations.
*
Urban core is defined as the central city. In large cities, this
may encompass areas as much as 3-4 km from the center of the central
business district.
A-2
-------
Obviously, the greater the number of monitoring sites, the better the
chance of observing high concentrations of ozone in the vicinity of a
city. Furthermore, imposition of controls may have as one of its
effects a shifting of the location of maximum ozone concentrations.
Therefore it is recommended that additional ozone monitors be deployed.
For very large cities (e.g., New York City), maximum 03 concentrations
frequently appear to occur somewhat farther downwind than the 15-30 km
SAMWG guidance suggests. Therefore, if sufficient resources exist, a
third ozone monitor might be placed somewhat farther (40 + km) from a
city in the predominate summer/fall daytime wind direction.
In addition to the requirements for monitoring 0^, it may be
desirable to monitor non-methane hydrocarbons and NO if the EKMA or
A
certain statistical approaches are utilized. Since NMHC and NO data
X
are needed to calculate prevailing NMHC/NO ratios, the instruments
A
should be co-located. Because it is important to obtain representative
values for the ratio in locations likely to have the highest smog-
forming potential, it would be preferable to have more than one NMHC and
NO monitor located in the urban core and preferably at least 200 m from
A
any important sources. Analysis of the RAPS data confirms that NMHC/NO
A
ratios in the St. Louis urban core do not vary appreciably from each
other during the day. It is recommended that NMHC and NO ratios
A
observed during 6-9 a.m. be used. The reasons for this are that con-
centrations measured at this time best reflect initial conditions prior
to the onset of photochemical reactions. Further, concentrations of
NMHC and NO are apt to be higher during these hours. Hence, the reli-
X
ability of the monitoring data is greater. As described later, there
are serious problems with the available measurement techniques for
A-3
-------
ambient NMHC. For this reason, the Agency cannot recommend the use of
individual NMHC readings from single monitoring sites in analytical
approaches relating oxidant to precursors. As will be discussed later,
more robust measures such as the median NMHC/NO ratio on days experiencing
X
high ozone may be used, with smaller associated errors.
If it is not possible to monitor NMHC and/or NO , default values
A
for the NMHC/NO ratio based on ambient observations presented in Table
/\
1 could be used in EKMA. These default values were obtained using NMHC
and NO data from a number of cities. The data were obtained either as
X,
part of EPA field studies or as a result of State or local monitoring
efforts.
In order to consider the role of transported ozone, as discussed in
Section 3.0, additional ozone monitoring sites are needed. It is
suggested that at least one such site can be located predominantly
upwind of the urban core of the city during late morning (e.g., 1100-
1300 LOT) on sunny summer/fall days. The site should be located in as
rural a location as possible so as not to be appreciably affected by
local sources of precursors. The goal to strive for is to obtain a
representative indication of the ozone concentrations in incoming air
aloft. The distance such upwind sites should be located from a city
would depend on the degree of urban sprawl. It is desirable not to
measure pollutants which are recirculated from the city under review.
Limited experience with the RAPS data base suggests that a distance of
40 km or more upwind from the urban core should be sufficient,, This
distance perhaps could be reduced for smaller cities. Figure A-l
A-4
-------
depicts orientations for acceptable upwind sites. Note that it is not
essential for an "upwind" site to be "upwind" provided it is not influenced
by emissions from the city under review. The inherent assumption is
that the "upwind" sites in Figure A-l are measuring representative air
quality levels resulting solely from remote sources. Since SAMWG has
indicated that only two ozone monitoring sites in high concentration
areas need to be maintained on a permanent basis, additional sites may
be operated on a seasonal basis or as part of a special study.
Use of mobile monitors should be subject to the same guidance as
that previously discussed for stationary sites. In the event tower data
are used to estimate transported ozone, as suggested in Section 3.0, the
tower should preferably be located as far upwind from the city as
feasible to reduce the measured impact of recirculated pollutants.
Instruments located 1000 feet or more above the ground should provide a
good indicator of ozone transported aloft.
A.2. Limitations of Existing Instrumentation
The instrumentation used to measure ambient NO should be capable
/\
of estimating hourly concentrations. The most widely available NO
X
instruments are sensitive down to levels of about .005 ppm. This degree
of sensitivity does not present serious difficulties in estimating urban
NO levels. However, it exists as one of the technical barriers to
A
understanding fully the rural oxidant problem where exceedingly low
levels of NO prevail.
X
A-5
-------
(A)
WIND
(B)
STAGNATION
>40 km
>40 km
CITY
>40 km ___ ^ 1
^ 1
)
CITY
^40 km
Figure A-l . Examples of acceptable monitoring locations for estimating transported ozone.
A-6
-------
Far more serious problems exist with the methods available for
estimating ambient non-methane organic concentrations. Within the past
several years, there has been increasing awareness of the problems
associated with measuring ambient non-methane hydrocarbons. These
problems center on the Flame lonization Analysis (FIA) method used in
continuous field instruments for NMHC. According to this method, iJMHC
are not measured directly, but are computed by determining the difference-
between measured methane concentrations and the total hydrocarbon (THC)
concentration of the sample. Thus, possible errors in the individually
measured methane and THC are compounded in the calculated NMHC value.
When methane and THC are similar in magnitude, there can be a large
error in their difference. This may occasionally result in negative
estimates of NMHC.
The causes of error in the methane and THC measurements are attri-
butable to a wide range of mechanical and operational problems with the
5 6
FIA instruments. ' For example, system flow valves tend not to regulate
effectively the pressure in the sampling loop. This inability to
maintain stable hydrogen, combustion air, and carrier gas flow rates
results in an unstable flame and inaccurate instrument response. Other
problems such as deterioration of the analytical columns, contamination
of the sample due to adsorption and desorption of hydrocarbons in the
system, and sample loss in inlet lyies and plumbing contribute to the
response unreliability.
A-7
-------
These problems are further complicated bv the highly sensitive
5 6
nature of the instruments. ' Instrument response varies if, during
calibration, span gases are not in air, not correctly analyzed for
methane, or contain unknown amounts of higher hydrocarbons. Additionally,
THC measurements are sensitive to the oxygen content of the calibration
gas and tend to be depressed by high ambient relative humidity. The
mixture of light and heavy hydrocarbons in the sample also affects the
instrument's response. The FIA will produce a greater response in air
to methane than to the same number of carbon atoms of higher hydrocarbon
species. The response to carbon-oxygen bonds is weaker still. These
mechanical fluctuations, together with the sensitivity problems of the
instruments result in drift and instability of response. In addition,
the instruments are highly erratic, requiring that a skilled technician
be assigned to maintenance during all operations. Small oversights and
errors in operating and calibrating the analyzer contribute to the
inaccuracy in recorded values. Further details on the problems arid
complexity of operating the FIA field instruments are discussed in
References 5 and 6.
The inconsistent response of hydrocarbon instruments leads to
imprecision and repeatability errors in recorded NMHC data. Because no
commercially available instruments measure NMHC directly, existing field
instruments must be evaluated with laboratory test mixtures or against
the response of other field instruments. This presents some diffi-
culties in quantifying the errors, since each test method has substantial
limitations. Using laboratory mixtures can test instrument response to
A-8
-------
known hydrocarbon levels, but that eliminates variability due to ambient
conditions. Comparing data from different analyzers sampling ambient
air solves this problem but eliminates a reference NMHC value.
An EPA study using the latter approach estimates (by assuming most
instruments have equal total error) that the error in hourly NMHC data
may be 5 to 10 percent of the 10 ppm full scale range of the instrument.
This leads to the assumption that there may be at least +_ 100% error in
values below 0.5 ppmC, +_ 50% error in values around 1.0 ppmC, and +_ 25%
error in values around 2.0 ppmC. A look at instrument response to
known NMHC concentrations gives slightly different results. Based upon
a study by Scott Labs for EPA, the difference between known NMHC con-
centrations and values (peak response) measured by 13 users of Flame
lonization Devices (FID) ranged from +350% to -110% (1.04 ppmC maximum
observed, -0.02 ppmC lowest observed) for an actual concentration of
0.23 ppmC, and +28% to -32% (3.70 ppmC maximum, 1.98 ppmC minimum) for
2.90 ppmC.
In summary, measurement of NMHC by the available commercial FID
instruments yields unreliable individual measurements. Inaccuracy in
individual hourly values may be as high as +_ several hundred percent in
the range of concentrations below 0.5 ppmC. At higher levels, the error
is considerably less, but may still be +_ 25% at up to 2.0 ppmC.
Because of the unreliability of individual NMHC measurements, there
was cause for concern about whether any approach (such as EKMA or potential
statistical approaches) which relies on ambient NMHC data could be used.
NMHC data which were observed on days experiencing high ozone levels
A-9
-------
were reviewed for several cities. Ambient NMHC levels occurring on
sucli days tend to be high (with correspondingly small relative errors).
As demonstrated in the examples in Section 2.2.1, for NMHC/NO ratios
/\
prevailing in most cities, this magnitude in estimating the prevailing
ratio would not be of serious consequence in relating ozone changes to
precursor chanc.es with the EKMA.
In summary it has been demonstrated that the error in individual
NMHC data recorded from FIA instruments does not, in most cases, produce
a significant impact upon the application of the EKMA methodology under
actual ambient conditions, occurring on days experiencing high ozone
levels. Because of this, the FIA instruments could be used by State/local
agencies to obtain NMHC data, but only if the data are used to compute
median and mean NMHC/NO ratios. Alternatively, if close agreement is
X
found among several sites on the day in which the ozone design value is
observed, the mean of such measurements could be used in the EKMA as
well. Any use of individual NMHC concentrations at single sites for
single days is not recommended, because of the large error that is
present in the data. Existing field instruments should only be used
until a more precise technique is developed. Therefore, if a city does
not have FID instruments, it should rely on default ratios presented in
this document or on estimates obtained in special contractural studies
performed by qualified specialists.
A-10 *
-------
References for Appendix A
1. "Requirements for Preparation, Adoption and Submittal of Implementation
Plans." Title 40, Code of Federal Regulations: Part 51. 36FR15502
(August 14, 1971).
2. "Oxidant/Hydrocarbon Monitoring." Strategy Issue Papers (Draft),
Standing Air Monitoring Work Group, U.S. EPA, Research Triangle Park,
North Carolina, October 1976.
3. U.S. EPA, OAWM, OAQPS, MDAD, Procedures for Quantifying Relationships
Between Photochemical Oxidants and Precursors: Documentation for
Recommendations. EPA-450/2-77-021b, in press.
4. W. S. Cleveland, et al., "Analysis of Ground-Level Ozone Data from
New Jersey, New York, Connecticut, and Massachusetts: Data Quality
Assessment and Temporal and Geographical Properties." APCA Paper
75-51.6, presented at the APCA Convention, Boston, Mass. June 1975.
5. Hydrocarbon Measurement Discrepancies Among Various Analysers Using
name-Ionization Detectors. EPA-600/4-75-010, U.S. EPA, Research
Triangle Park, North Carolina, September 1975.
6. Survey of Users of the EPA Reference Method for Measurement of
Non-Methane Hydrocarbons in Ambient Air. EPA-650/4-75-008, U.S. EPA,
Research Triangle Park, North Carolina, December 1974.
A-ll
-------
APPENDIX B
GUIDELINES FOR USE OF A KINETICS MODEL IN ESTIMATING
CITY-SPECIFIC OZONE-PRECURSOR RELATIONSHIPS
B.1 An Overview of City-Specifie Studies
This appendix provides guidance in collecting the nr-<.>:.<.) data to
use the city-specific isopleth version of EKMA as descr ,'-•». .. bect\un:> 2.2.1b
and 3.3. This appendix also discusses the use of thes : -iu•:.•! ;n i kinetics moao'.
to derive the necessary isopleths required by EKMA. The kineses model which
should be used is described briefly in Section 2.2.la and ,r, .Jerri!, by Dodge
?
and Whitten and Hogo. This model has been successfully validated against smog
chamber data and represents the state-of-the-art for mathematically simulating
the production of photochemical oxidant without requiring large quantities of
input data.
The procedures discussed herein use the kinetics model to generate isopleths
representing ozone-precursor relationships for the specific conditions applicable
to a given city. Procedures are discussed separately for (1) analyses which con-
sider pollutant transport, and (2) analyses which do not consider pollutant
transport. Table B-l summarizes the procedures necessary to determine emission
control requirements both with and without the consideration of pollutant
transport. This table shows the three phases that comprise each such analysis
(i.e., data collection, kinetics model use, and determination of emission control
requirements). Similar analyses may also be conducted to determine the reduction
in downwind ozone concentrations that occur with a given emission reduction,
although the analysis is more difficult if pollutant transport is being considered,
B-l
-------
, —
1
CQ
QJ
J2
ro
I
^^
co
4->
C
OJ
f\ \
(U
•i —
3
CT
QJ
o:
r-~~
O
s-
4->
C
0
CJ
c
0
co
CO
•r—
£
LU
cn
c
•i —
E
•r"-
CD
4— '
QJ
Q
S-
O
4»
CO
QJ
S-
3
-o
QJ
CJ
o
s-
Q.
o
•r-
4-
O
OJ
D-
00
>,
4->
• i —
O
S-
0
Q.
CO
c:
QJ ro
X- i—
3 i—
T3
QJ 4->
0 C
O ra
S- 4J
Q- 3
O r^
•r- 0
4- Q-
., —
CJ cn
CD C
Q--I-
00 S-
1 QJ
">». "n
^ TJ
4-> T-
•i- CO
O SZ
o
CJ
4->
5-
4-> O
O Q_
C CO
C
QJ ra
S- S-
3 1—
-o
QJ 4->
O C
O ra
S- 4->
Q- 3
CJ ^
•r- O
4- 0-
O CD
QJ C
CL-i-
00 S-
1 QJ
>^"O
4-J •!-
•r- co
0 E
O
CJ
-
co
QJ
4->
ro
s_
C
o
>•, T-
4-; co
*^" QJ CO
CO 4-> -r-
sz ro E
QJ S- QJ
.J_l
C C QJ
•^ 0 >
•r— *i —
4-J 4-1 4->
JZ 3 ro
CT>i— i—
••- -r- CD
_j a o;
c:
o
rO
4-)
ro
-a
4->
CJ
QJ
'o
O
QJ
C
o
N
O
t3
t data on Transporte
o
QJ
, —
O
O
-D
C
QJ S- o
3 4-> O S-
i— i- 4- CJ)
rO O -i Q.-C: cj
CO 4-> ra
c: c CD 4-> jo
CD ra i — c
•i- s- a_ to i —
co 4-> O 4-> ra
CD co 3 S-
-O 4-> -i- •— 3
C t— 4->
II ra Q) o ra
4-> > a. c:
CO 3 -p-
O i— S- C cn
i— CD O £Z
S- O "O -r-
O Q- T3 -C
4- 0 CD U
4J CO CO ra
-c c i — ra o
4-> Q) <£ J3 S-
CD S- Q.
i— s- E a.
Q. 3 . Q. ra
O O CO Q.
CO C 4->
•r- C O CO S-
O -r- O O
QJ 4-1 • Q_ CO
> -0 -r- CO r-
•i- CD "O M CO)
S- co c ra >
QJ ra O OOS-. QJ
a _a o o 4-> i —
c
-Q
CD co i—
> CD •
o i- c\j
_Q 3 •
ro -O CXI
QJ
CJ TD O CO
•r- CD o c:
4- > i- 0
•r- -i- Q_-r-
0 S- 4->
CD CD O 0
CLTD 4-> CD
CO OO
1 CO CT)
>~>-c: c c:
4J 40 n- .r- OO
•r- CD "O
0 r— S- C CO
Q- O QJ
QJ o o > -a
co co o -r- c:
ID -i- ra cn ra
1
i;_
4-> E
S- Q.
OQ D-
•I— Q_ CO
4- cn co -C CO
••- c: c: 4-> o
O T- ro CD
CD CD E S-. i— c:
i— a. 3 4-> D. cn ii
CD CO CO O •!—
c: i co 4-> co co oo
•r- >, ra !Z -i- CD O
co 4-> ra T3
•r— CO 4-> CD "D
ra o j^ 3 c ii c
4-> i — -i- ra
QJ 4- QJ i — "O OO
> O i — O 3 O CD
•i- Q. Q-i— 3
S- 4-> 0 0 S- i—
QJ QJ CO O C O ra
Q co -i- c: -r- 4- >
OO
OO
o T- -a
0 T3 4-> C
•i- CD C ra
4- > CT) QJ
•i- -r- £: > J3
CJ S- T- -r- r—
QJ QJ T3 Cn .
a.-o s_ cxi
co O co
1 CO CJ QJ CXI
>,JC CJ S-
4-> 4-> ra 3 c
•r- QJ "O O
o i — QJ CD •!-
Q. > O 4->
QJ O O O O
co CO JI2 S- CD
13 -r- 03 Q.OO
o
O)
0)
T3
O
O C CO
•r- O +->
-P
(T3
Q
CO
O QJ
•r- CO
+J ZD
QJ
B-Z
c: c: E
1 .2 2!
S- 00 -r-
O) co rs
+-> -r- cr
QJ E QJ
Q I.U C£
-------
The specific procedures that should be followed in each type of analysis
have been discussed in Sections 2.2.1b and 3.3 of the main text. As shown in
Table B-l, the first phase of any analysis is data collection. All analyses
require data on light intensity, dilution rate, and relative emission rates.
Data on light intensity for the day the design value of ozone* occurs are
necessary to simulate the diurnal variation of photolytic rate constants.
Data for determining dilution rate are necessary to simulate entrainment of
relatively clean air as the mixing height increases. Relative emission rates
are necessary to simulate variations of emissions into the air parcel over space
and time. If the information is available, it is also desirable to consider the
concentrations of ozone and other pollutants being transported into the city.
B.2 Data Requirements
The first step in any determination of the city-specific ozone-precursor
relationship is data collection. Table B-2 shows the data required in order to
use the kinetics model. Information should be obtained on light intensity,
dilution rate and relative emission rates. Information should also be obtained
on transported pollutant concentrations (particularly ozone) if pollutant
transport is being considered. Dodge discusses the variables in more detail
and discusses the values that are used in generating the standard set of
2
isopleths presented in Section 2.2.la. Whitten and Hogo discuss the format
required for use of these variables in the kinetics model. The following
subsections provide procedures for determining city-specific values of these
variables.
*The design value of ozone is the highest second-highest ozone
concentration measured in or near a city.
B-3
-------
Table B-2
Input Requirements of the Kinetics Model
Data Required
Values Used in
Standard Isopleths
Source of City-Specific
Value
Light intensity as
defined by latitude,
longitude, and day
of year
34°N, 118°W
June 21
Latitude and longitude is
determined using a local
map; day of year is based
on times of highest
measured data
Mixing height data
Dilution rate = 3%/hr
Local mixing height
measurements or data
from Hoi zworth^
Relative emission rates
Emission rates after
8 a.m. LOT are zero
Local emissions and
meteorological data
(default values for
missing data provided
this Appendix)
in
Ozone concentration
transported into city
(optional)
0.0 ppm
Ambient measurements
B-4
-------
B.2.1 Light Intensity
2
The kinetics model discussed by Whitten and Hogo requires data on
latitude, longitude and day of the year. These data are used to calculate
the appropriate diurnal variation of photolytic rate constants. Local
latitude and longitude (for the center of the city) should be readily avail-
able. The date used should be the day on which the design ozone concentration
occurs.
Adjustments for attenuation of light intensity due to cloud cover are
not provided for in the kinetics model package. If quantitative information
exists which suggests a different level of light intensity is more appropriate,
this information may be used by modifying the kinetics model. However, it
should ordinarily be assumed that the highest ozone concentrations occur on
days without significant cloud cover.
B.2.2 Dilution Rate
Dilution in the kinetics model represents the entrainment of relatively
"clean" air from aloft into the primary "dirty" column of air. The kinetics
model uses data on mixing height to determine the rate of dilution. The
model assumes uniform mixing throughout the column of air at all times. Thus,
for example, if the mixing height doubles over some time period, dilution
by a factor of two is assumed to occur. The extent to which the air above
the original mixed layer contains pollutants is considered separately and
is discussed in the section on data requirements for considering pollutant
transport. Wind speed and lateral diffusion are not considered, because
the pollutant concentrations in the air surrounding the imaginary column of
air are assumed to be similar to those within the column. Thus, it is
assumed that horizontal mixing does not cause any significant dilution of the
concentrations within the column.
B-5
-------
2
The kinetics model as described by Whitten and Hogo requires data for a
minimum mixing height, a maximum mixing height, and a time period over which
the mixing height is increasing. Ideally, these data should be observed on
the day experiencing the design value of ozone. However, it is usually
necessary to use some other indicator of these mixing height data.
If the information is available, a suitable surrogate for design day
data is median data based on several of the days with the highest ozone
concentrations. Nevertheless, the only available data often are seasonal
mean morning and afternoon mixing heights as provided by Holzworth. If no
local information is available on the time of day over which the mixing height
is increasing, it may be assumed that increases in mixing height typically
occur primarily between the hours of 8 a.m. and 3 p.m. "local daylight time.
B.2.3 Relative Emission Rates
Relative emission rates express the rate of emissions into the imaginary
column of air for each hour of simulation relative to the emissions into
the air column prior to the beginning of the simulation (8 a.m. LOT). In
the kinetics model, emissions prior to the time a kinetics model simulation
begins (8.a.m. LOT) are represented by initial precursor concentrations.
Post-8 a.m. emissions are represented by additions to these precursor con-
centrations. Relative emission rates are used by the kinetics model to
determine the appropriate rate at which to add to the initial precursor
concentrations in order to simulate post-8 a.m. emissions.
Relative emission rates are required for each hour of simulation. A
rigorous calculation would require a precise specification of: (1) the location
of the air parcel, (2) the spatial variation in emissions, and (3) the temporal
B-6
-------
variation in emissions. Unfortunately, none of these items can be specified
accurately with the limited data normally available. The task then becomes to
derive reasonable approximations with data which are available.
The most practical approach is to develop a hypothetical air parcel
trajectory, and then to roughly approximate the spatial and temporal varia-
tions in emissions along this trajectory. Normally it is appropriate to first
consider spatial variations along the trajectory, and then separately adjust
for temporal variations. The following discussion indicates the types of
assumptions that may be used in estimating these variations. Note that it
is important to use information and assumptions that pertain to the local area
wherever possible. The assumptions and the values discussed below are included
primarily for illustrative purposes and should be used only if local information
is unavailable.
A typical set of assumptions might be that: (1) the emission distribution
2
is Gaussian in the direction of the wind (i.e., emissions = Q0e~ax , where
"Q0" is the emissions in the center of the city, "a" is a constant, and "x"
is the distance from the center of the city); (2) the air parcel meanders around
the center of the city until 8 a.m. and then moves at a constant velocity
until it leaves the urban area three hours later (a conservative estimate),
and (3) the emissions distribution is calculated on the basis that the emissions
_p
at the edge of the urban are e~ (approximately 0.14) of the emissions in the
center of the city. These assumptions result in the following hourly factors
expressing the emissions at the location of the air parcel relative to the
emissions in the center of the city:
1-7
-------
prior to 8 a.m. LOT : 1.0
8-9 a.m. : .95
9-10 a.m. : .61
10 - 11 a.m. : .25
after 11 a.m. : negligible
These emissions are calculated at the midpoint of each hour (e.g. 8:30 a.m.).
This generally provides an adequate estimate of the average emissions for
each hour.
In the absence of site-specific data, a city-wide average may generally
be used for the temporal distribution of emissions. This distribution may be
calculated as an average of the temporal distribution of mobile and stationary
sources. Nevertheless, the temporal distribution of automotive emissions
should ordinarily be an adequate representation of the temporal distribution
of the total emissions. If local information on the temporal distribution
A
is not available, Tittemore et. al. suggest that the following hourly percentages
of a day's automotive emissions may be typical (times are in LOT):
Midnight - 8 a.m.: 16.9% (the concentration at 8 a.m.
is assumed to reflect emissions over these eight hours)
8-9 a.m.: 6.4%
9 - 10 a.m. : 4.4%
10 - 11 a.m. : 4.8%
(for the sample calculation, the air parcel is assumed
to be outside the city after 11 a.m. and thus no
longer receiving emissions).
In this example it is not necessary to consider the hours prior to 8 a.m.
separately since the air parcel is assumed to remain stationary and because
the mixing height is assumed constant during this time period.
B-8
-------
Once the above data are obtained, relative emission rates may be determined
by adjusting the relative emissions for the air column location each hour
according to the fraction of the day's emissions that occurs during that hour.
An important subsequent step is to normalize these emission rates relative to
the emissions that have entered the air column prior to 8 a.m. The calculation
of normalized relative emission rates is illustrated in Table B-3 with sample
values of spatial and temporal emissions variations. Normalized relative emission
rates are required as input to the kinetics model. Note that the values shown
in Table B-3 should be used only if no city-specific information is available.
The rates of precursor concentration addition derived by the kinetics
model are dependent upon the emission rate and the height over which these
emissions are mixed. The kinetics model uses input data on normalized relative
emission rates and changes in mixing height to determine post-8 a.m. rates
of concentration addition as a fraction of the 8 a.m. concentrations. If the
mixing height is variable prior to 8 a.m., the pre-8 a.m. emissions (against
which later emissions are normalized) must be adjusted accordingly. It may be
noted that changes in wind speed affect the location of the air parcel,
the corresponding emission density, and the length of time spent traversing
an area; however, changes in wind speed do not affect the mass per unit time
added to the air parcel due to emissions from a given area. That is, if
the air parcel remains for an hour over an area with a given average emission
rate, the total mass added to the air parcel is independent of the wind speed.
Thus the kinetics model requires only normalized relative emission rates and
the mixing height data used in determining dilution rate to calculate the rates
of concentration addition appropriate for simulating post-8 a.m. emissions.
B-9
-------
CD
4-> *
re cu
,— 4->
CD re
o; o:
•a c
CD o
N -I—
•r— CO
i— CO
re -I—
E E
i. UJ
O
o
o
00
CD
S
CD
CD
CD
OO
1
CO
cu
-Q
fO
\—
CO
CD
4->
re
o;
c:
o
•i—
CO
CO
• r—
E
UJ
CD
>
4->
re
i —
CD
rf
S_
O
q-
CO
CD
rs
i —
re
>
CD
0.
p:
re
OO
(+_
o
c
o
-I—1
re
>
•i—
<*.
CD
Q
CD
>
4->
re
i —
CD
a:
CD
-M
•1 —
CO
o
Q.
E
O
c_>
If..
o
4-^
ercen
o
OJ
4-1
re
o;
f—
o
•i —
CO
CO
• 1 —
E
UJ
CO
c
0
•r—
CO
cn
*>^
E
UJ
co
>-
(O
Q
4J
L"
Q)
+J
CO
_J>
TD'
«-'JC
're
+j
ns
Q.
CO
x — ^
Q
—I
E
I—
II
s^gf
otri
r^> •
°U
X
CD
r^9 -r^p -^p ~^^i
O^ — o^^ o\ c^- l~~ '
Ol i — 1 C\J J3
• • • • 'r—
tD tD C\J i— CD
r— ., —
,
CD
CD
C
II II II II
&s &^ s«
CTi ^J- ^j- CO
• • * •
UD Id ^- ^±
x x x x
CD
[Q
un i — LO •!-
O CTl C£> C\J CD
D)
CD
E
.
E E E' E
re
re re re re
OO CTi O i — i —
s~
1 1 1 1 CD
i__)
CM CO CTl O "4-
. — i — rC
CD
1 \
re
s-
CD
4_)
CO
o
a.
E
o
o
.
£
re
00
CD
_c
+J
o
4->
CD
^.
4->
re
CD
T3
QJ
NJ
re
E
O
c:
CD
4->
re
S-
E
O
CO
CO
E
CD
CD
-£
re
CD
S_
CD
4->
CO
O
a.
E
0
u
CD
-C
I—
B-10
-------
B.2.4 Transported Ozone Concentrations
Section 3.1 discusses the transport of ozone in two atmospheric layers,
viz. transport in a layer near the ground and transport aloft. Section 3.2
discusses means of estimating the ozone concentrations transported into a city
in these two layers. Ideally, one should have measurements of the transported
ozone concentration at several heights. However, ground-level measurements
are generally the only available data. The best ground-level measurements for
indicating ozone concentrations transported aloft are generally upwind measure-
ments taken in the late morning and early afternoon, after mixing has extended
well above the morning mixed layer. Therefore, unless better information is
available, the ozone concentration transported aloft into the urban area for
the current situation should generally be assumed to equal the upwind ground
level measurements between 11 a.m. and 1 p.m. LOT.
Ozone transported in the surface layer in the early morning is generally
much less significant that ozone transported aloft. Scavenging may at times
significantly reduce ozone concentrations transported in the surface layer.
5
Furthermore, a kinetics model analysis suggests that even high ozone concen-
trations transported in the surface layer have little effect on the peak ozone
concentration in the afternoon. Nevertheless, if information is available on the
current ozone concentrations transported in the surface layer early in the
morning, e.g. if upwind measurements are available between 6-g a.m., this
information may be considered. Substantially similar results should be obtained
using any reasonable assumption, for example, that the ozone concentration
transported in the surface layer equals the concentration transported aloft.
When ozone isopleths representing future conditions are being generated,
it is necessary to estimate the corresponding transported ozone concentrations.
When an isopleth for the NAAQS for oxidants is being generated, it is reasonable
B-ll
-------
to assume that upwind cities are also meeting the NAAQS. Therefore, for this
case, the ozone concentration transported aloft can be assumed to be generally
close to the natural background concentration but in some cases as high as
0.08 ppm. An examination of the ozone concentrations found in isolated areas
suggests that natural ozone concentrations are approximately 0.03 - 0.05 ppm,
with 0.04 ppm. most likely. Thus, in deriving the isopleth for 0^ = 0.08 ppm,
the ozone concentration transported aloft should normally be assumed to be
0.04 ppm but may be assumed to be as high as 0.08 ppm. The relationship
between the ozone concentration transported in the surface layer and the
ozone concentration transported aloft should be similar to the relationship
used in simulating current conditions.
B.2.5 Other Variables
The discussion of pollutant transport in this Appendix has focused
exclusively on the transport of ozone. If data are available to indicate
both current and future transported precursor concentrations, these data may
also be considered. However, consideration of precursor transport is not
considered essential for several reasons. First, transported precursor con-
centrations, particularly NO concentrations, tend to be significantly
x
c
below the concentrations attributable to an urban area. Second, due to
rapid reactions of the more reactive hydrocarbon species, the mix of trans-
ported organic species is likely to be less reactive than an urban emissions
mix. Third, transport of precursors from an upwind city decreases as the
upwind city controls its emissions. To the extent these reductions parallel
the emissions reductions of the city under review, precursor transport
has little if any impact on the relative effect of local emission reductions.
B-12
-------
In order to assess the potential impact of precursor transport, a
kinetics model analysis of transport considered the impact of precursor
transport on a hypothetical city located within 40 kilometers of St. Louis.
This was assumed to represent a worst case example of transported precursors.
Simulations were conducted based on two different assumptions: (1) the
concentrations transported aloft were equal to those transported at ground
level, and (2) the concentrations aloft were minimal. These simulations
suggested that the impact of transported precursors is usually less than
the impact of transported ozone. Thus for any reasonable assumption about
future transported precursor concentrations, consideration of precursor
transport is likely to have only minimal effect on estimated emission control
requirements. If data on precursor transport are available, these data may be
considered. However, in the absence of these data, the failure to consider
precursor transport in most cases should not greatly affect the estimated
impact of local emission reductions.
In general, consideration of other variables is not specifically
recommended, either because consideration of the variable is likely to have
little, if any, effect on the estimated peak ozone concentrations or because
it is difficult to quantify the variable. One such variable is the initial
N0~/N0 ratio. The standard isopleths presented in Section 2.2.la are based
£• /\
on the assumption that N09 is 25% of MO . Ground level transport of ozone
L- X
is likely to be the most significant factor affecting the initial N09/N0 mix,
(~ A
but the kinetics model analysis has suggested that the effect on peak ozone
concentrations is small. Other factors that affect the initial N02/MOX mix
are likely to have less effect on this mix and negligible effect on peak
ozone concentrations, as shown by Dodge.
B-13
-------
Consideration of reactivity is not recommended, due to the serious
difficulties involved in determining an appropriate hydrocarbon mix. These
difficulties stem both from the limited knowledge about the mixes of
hydrocarbons that exist in urban areas and from the uncertainties about how
to represent these mixes in the kinetics model. In particular, there are
generally no data available—either through measurement or through estimation
using an emissions inventory—on what mix of paraffins, olefins, aldehydes and
aromatics can be expected in individual urban areas. Even in cases where
the data are available, the kinetics model cannot directly use the data. The
kinetics model is based on a detailed chemical mechanism using two hydrocarbon
species, specifically n-butane and propylene. Unfortunately, there is very
little information with which to specify what butane-propylene mix should be
used to simulate different atmospheric hydrocarbon mixes. The only atmospheric
mix for which a corresponding butane-propylene mix has been established is the
automotive mix used in the Bureau of Mines smog chamber, which is best
represented with a 75% butane-25% propylene mix. Unless correspondence
between other atmospheric and kinetics model mixes can be established, the
75% butane-25% propylene mix should ordinarily be used in generating city
isopleths.
Consideration of transported aldehydes or variations in the initial
aldehyde mix is also not recommended. The standard isopleths assume that
formaldehyde represents 2% and acetaldehyde represents 3% of the total initial
nonmethane hydrocarbon concentration (all in ppmC). In some situations, these
proportions might be slightly higher. However, consideration of these variables
5
is likely to have little effect on peak ozone concentrations. In addition,
aldehydes are difficult to measure reliably. Therefore, unless contrary
B-14
-------
information is available, the assumptions about aldehyde concentrations
used in deriving the standard isopleths should be used.
B.3 Summary
This appendix has outlined the data requirements necessary to generate
city-specific isopleths. In particular, various assumptions and potential
sources of data have been discussed for determining light intensity, dilution
rate, relative emissions rates, and transported pollutant concentrations.
Once the appropriate data are obtained, the kinetics model described by
2
Whitten and Hogo may then be used to generate city-specific isopleths.
These city-specific isopleths may then be used to estimate emission control
requirements and the benefits of given levels of emission reduction as
discussed in Sections 2 and 3.
B-15
-------
4.0 References
1. M. C. Dodge, "Combined Use of Modeling Techniques and Smog Chamber
Data to Derive Ozone-Precursor Relationships," International Con-
ference on Photochemical Oxidant Pollution and Its Control. Proceedings
Volume II, EPA-600/3-77-001b, p. 805. U.S. EPA, Research Triangle
Park, N. C., January 1977.
2. G. Z. Whitten, Hogo, H., User's Manual for a Kinetics Model and
Ozone Isopleth Plotting Package, prepared under EPA Contract
58=02-2428, U.S. EPA, Research Triangle Park, N. C.
3. G. C. Holzworth, Mixing Heights, Wind Speeds and Potential for
Urban Air Pollution Throughout the Contiguous~Dnited States,
A.P. 101, U.S. EPA, Research TriangTe Park, N.C. January~T372.
4. L. H. Tittemore, et. al., An Analysis of Urban Travel by Time
pf_Day, Federal Highway Administration No. FH-11-7519, "U.S. '
DOT, Washington, D.C., January 1972.
5. Environmental Protection Agency, Procedures for Quantifying
Relationships Between Photochemical Oxidants and Precursors:
Documentation for Recommendations. EPA-450/^-77^021b, U.S. EPA,
Research TriangTe PaFR, N.C., November 1977.
6. M. C. Dodge, Effect of Selected Parameters on Predictions of a
Photochemical ModeTEPA-600/3-77-048. U. S~ EPA, ResearcF
Triangle Park, N.C., June 1977.
B-16
-------
TECHNICAL REPORT DATA
(/'lease read Imlrvcnons oi< 'he reverie before rump/etui/;)
REPORT NO.
_EPA-450/2-77-_Q2Ja_
TITLE ANn ci IBTI
3 RECIPIENT'S ACCESSION NO.
Uses, Limitations and Technical Basis of Procedures for
Quantifying Relationships Between Photochemical Oxidants
and Precursors
5 REPORT DATE
November 1977
6. PERFORMING ORGANIZATION CODE
AUTHOR(S)
8. PERFORMING ORGANIZATION REPORT NO.
PERFORMING ORGANIZATION NAME AND ADDRESS
EPA, Office of Air Quality Planning and Standards
Monitoring and Data Analysis Division
Research Triangle Park, N.C. 27711
2 SPONSORING AGENCY NAME AND ADDRESS
10 PROGRAM ELEMENT NO
2AA635
11. CONTRACT/GRANT NO.
13. TYPE OF REPORT AND PERIOD COVERED
14. SPONSORING AGENCY CODE
5 SUPPLEMENTARY NOTES
6. ABSTRACT
Approaches for relating ozone to non-methane hydrocarbon and oxides of nitrogen
controls are discussed. All procedures are only applicable in addressing the oxidant
problem in urban areas. Available procedures include use of Eulerian and Lagrangian
photochemical dispersion models, a new approach called the Empirical Kinetic Modeling
Approach (EKMA), proportional rollback and statistical approaches. EKMA, rollback and
statistical models are most useful in estimating the degree of reduction needed to at-
tain the National Ambient Air Quality Standard (NAAQS) for oxidant and for estimating
reduction in maximum hourly ozone concentration accompanying specified levels of pre-
cursor controls. Models are useful for these purposes as well as others. The issue of
long range transport of ozone/precursors is discussed. It is concluded that ozone
transported above the surface-based nocturnal inversion layer is likely to be the facet
of transport exhibiting the greatest impact on maximum ozone levels in urban areas.
Procedures for integrating transport considerations into the previously mentioned
analytical approaches are discussed.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Air Pollution
Ozone
Non-methane Hydrocarbons
Nitrogen Oxides
Meteorology
Transport
Atmospheric Models
Kinetic Models
b IDENTIFIERS/OPEN ENDED TERMS
c COSATI Field/Group
8. DISTRIBUTION STATEMENT
Unlimited
19 SECURITY CLASS (This Report)
Unclassified
21 NO. OF PAGES
117
20 SECURITY CLASS. (This page)
Unclassified
22 PRICE
EPA Form 2220-1 (Rev. 4-77)
VIOUS F-OITION IS OBSOLETE
------- |