Technical support document (TSD) for NO2-
related AERMOD modifications
-------
-------
EPA-454/B-15-004
July 2015
Technical support document (TSD) for NO2-related AERMOD modifications
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Air Quality Analysis Division
Air Quality Modeling Group
Research Triangle Park, North Carolina
-------
Preface
This document provides a review the scientific merit of the NO2 options in AERMOD, summarizes
existing evaluations of these options, and presents additional testing used to determine appropriate
application of these options for these options as part of the regulatory default version of AERMOD.
-------
Contents
Preface 4
Contents 5
Figures 6
Tables 7
1. Introduction 8
2. Background 8
2.1 NO2 chemistry 8
2.2 Ambient Ratio Method 10
2.3 Ambient Ratio Method 2 11
2.4 Ozone Limiting Method 13
2.5 Plume Volume Molar Ratio Method 13
2.6 Existing evaluations of OLM and PVMRM 14
3. Regulatory status of ARM, ARM2, OLM, and PVMRM 15
3.1 Regulatory requirements 15
3.2ARMversusARM2 16
3.3 Status of OLM and PVMRM 17
3.4 The 3-Tiered approach 17
4. Proposed regulatory options for NO2 modeling in AERMOD 17
4.1 ARM2 options 17
4.2 Default options for OLM and PVMRM 19
4.3 Appropriate applications of OLM and PVRMR 21
4. PVMRM2 21
4.1 Comparison of PVMRM vs. PVMRM2 Options in AERMOD 22
4.2 Comparisons of PVMRM and PVMRM2 for select field studies 26
References 30
-------
Figures
Figure 1 - NO2/NOx equilibrium ratios based on the PSS assumption from Equation 1 10
Figure 2 - NO2/NOx ratios from AQS for 2012 12
Figure 3 - Comparison of modeled NO2 concentrations from ARM2 and PVMRM 18
Figure 4 - Distribution of NO2 ISRs from EPA's database as of June, 2015 20
Figure 5 - Comparison of PVMRM and PVMRM2 for the Empire Abo north monitor 27
Figure 6 - Comparison of PVMRM and PVMRM2 for the Empire Abo south monitor 28
Figure 7 - QQ plot of PVMRM for the Palaau monitor 28
Figure 8 - QQ plot of PVMRM2 for the Palaau monitor 29
-------
Tables
Table 1 - Comparison of ARM and ARM2 estimated NO2 concentrations 17
-------
1. Introduction
The proposed revisions to Appendix W to CFR 40 Part SI—Guideline on Air Quality Models (Appendix
W), includes a new version of AERMOD (ADD VERSION #). This version of AERMOD reflects the proposed
new regulatory default options for NO2, including the new Ambient Ratio Method 2 (ARM2, (Podrez,
2015)), the existing Ozone Limiting Method (OLM, (Cole & Summerhays, 1979)), and a revised version of
the Plume Volume Molar Ratio Method (PVMRM, (Hanrahan, 1999a)). This TSD reviews the scientific
merit of these methods, summarizes existing model evaluations, and presents additional testing used to
determine appropriate application of these options for use as part of the regulatory default version of
AERMOD.
2. Background
The development of the original Ambient Ratio Method (ARM), OLM, and PVMRM precedes both the
development of AERMOD in the late 1990's and its promulgation for regulatory use in 2005. Though
there are aspects of the implementation of these methods in AERMOD that are specific to AERMOD, the
basic approach to NO2 speciation used by ARM/ARM2, OLM, and PVMRM are not inherent to AERMOD.
As a result, there is a notable body of literature focused on the more fundamental science behind these
NO2/NOx speciation approaches. This section provides a review of the scientific background available
for ARM/ARM2, OLM and PVMRM as well as the basics of NO2 chemistry relevant for these methods.
2.1 NO2 chemistry
Nitrogen oxides (NOx) consist of two species, NO2 and NO. Anthropogenic NOx is typically formed
during high-temperature combustion processes via three main reaction pathways. The most prominent
and consistent production route is the thermal oxidation of N2 via the Zeldovich mechanism (Zel'dovich,
Sadonikov, & Frank-Kamenetskii, 1946). The thermal production of NOx is fairly predictable and the total
amount of NOx and the speciation of NO and NO2 in the exhaust gas stream from "thermal NOx" is
controlled by the reaction temperature and the amount of available oxygen in the gas stream. NOx can
also be produced when nitrogen-containing fuels, such as oil and coals, are used and the nitrogen
contained in the fuel is liberated and subsequently oxidized. "Fuel NOx" can be minimal or a major
contributor to NOx emissions, depending on the fuel type (e.g., natural gas has virtually no nitrogen,
resulting in no fuel NOx production). The last production route is known as prompt production, which
involves the reaction of N2 with oxygen and hydrocarbon radicals resulting from fuel combustion.
"Prompt NOx" can involve hundreds of reactions of very short-lived species in the combustion chamber
and exhaust gas stream and is generally difficult to characterize and predict and instead is estimated as
the amount of NOx that is present that cannot be accounted for as thermal or fuel NOx. The speciation
of NO and NO2 in the exhaust gas stream from prompt and fuel NOx is a function of the fuels and the
type of combustion process and is very difficult to generalize. NOx production can continue in the
exhaust gas stream until temperatures drop low enough to end combustion.
Once NOx is in the atmosphere, there a number of potential chemical reactions that can occur,
depending on the relative amounts of NO and NO2, the total NOx, the ambient meteorological
conditions, and other atmospheric trace gasses available for reaction. In most cases, the fastest and
most important reactions of NOx involve ozone (O3). When there is sufficient ozone present, Reaction 1
dominates all other NOx chemistry:
NO + O -> NO + O
-------
In fairly concentrated plumes, this reaction is initially ozone limited, such that all available ozone will be
consumed. As a plume becomes more diluted and more ozone is mixed in, the reaction eventually
becomes NO limited. During daytime, NO2 can be photolyzed back to NO and subsequently reform
ozone:
NO2 + hv -> NO + O( 3P)
0( 3P) + O2 " 03
Though Reaction 1 is the fastest of the three, a quasi-equilibrium state, also referred to as the
photostationary state (PSS) due to Reaction 2 being driven by sunlight, can be achieved during the
daytime between NO, NO2 and ozone. While the PSS can occur within 1-3 minutes, it must also compete
with changes in plume concentrations due to mixing, i.e., entrainment of additional ozone and dilution
of NOx towards ambient levels. Thus, for a fairly concentrated plume, the rate of mixing and the rate of
the PSS reactions will compete on relatively equal time scales and Reaction 1 will usually dominate the
chemistry. Once a plume becomes relatively diluted, the PSS can be reached, which is dictated by the
total amount of NOx, background ozone and the ambient meteorological conditions. Assuming an equal
rate of NO2 formation and destruction, Reactions 1 and 2 can be used to determine the equilibrium
ratio of NO2/NOx (Seinfeld & Pandis, 2006):
Where Kl and K2 are the reaction rates for reaction 1 and 2:
The units for Kl are ppb"1*sec"1 and sec"1 for K2, where T is the ambient temperature in Kelvin and 6 is
the solar zenith angle (SZA) in radians from zenith1. The result from this equation is that the equilibrium
NO2/NOx ratio is solely a function of ozone, sunlight, and temperature and not a function of the total
NOx. The equilibrium NO2/NOx ratios determined from Equation 1 for a variety of conditions are shown
in Figure 1. There are clearly a wide range of potential equilibrium ratios. As the available sunlight
decreases (95% cloudy) or as there is more ozone available, the preference is for NOx to remain as NO2.
It is important to note that the solution to Equation 1 holds ozone constant, such that ozone titration is
not taken into account within the setting of a plume undergoing dilution and reaction. Thus, for any
background ozone concentration, the corresponding equilibrium ratio in Equation 1 is a conservative
estimate of conditions present in a plume in the atmosphere.
1 The rates for Kl and K2 given here are taken from (Hanrahan, 1999a). However, it should be noted that there is
notable variation in the reported reaction rates for Kl from various sources. Additionally, K2 is an estimate of the
photolysis rate of NO2, which was stated as accurate to about 10% for a SZA between 0 and 65 degrees (Dickerson,
Stedman, & Delany, 1982). More recent references (e.g., (Atkinson, et al., 2004)) indicate that the rate for K2
requires an integration of the absorptions rates at the range of wavelengths present for any scenario. Despite
these differences in rate constants, the resulting NO2/NOx ratio is not anticipated to vary greatly between
estimates of the rate for either reaction and that the results indicated in Figure 1 are reasonably representative.
-------
•SZA 0°, 95% cloudy, 30° C -«-SZA 0°, 50% cloudy, 30° C SZA 0°, 5% cloudy, 30° C
SZA 60°, 95% cloudy, 0° C -*-SZA 60°, 50% cloudy, 0° C SZA 60°, 5% cloudy, 0° C
100%
90%
.a 70%
« 60%
75 50%
o
40%
30%
//
10 20 30 40 50 60
ozone (ppb)
70
80
90
100
Figure 1 - NO2/NOx equilibrium ratios based on the PSS assumption from Equation 1.
Though generally ignored in the realm of NO2 modeling with Gaussian plume-based models, there are a
number of other chemical reactions that can convert NO to NO2, convert NO and NO2 reservoir species
(e.g., nitrous and nitric acid), and remove NOx from the atmosphere entirely. At nighttime, there is
essentially no ozone production, such that in urban areas (where NOx emissions continue throughout
the day and night), background ozone is generally quite low. As a result, there are fundamentally a
different set of daytime and nighttime reactions. In general, nighttime NOx chemistry favors conversion
of NO to NO2 and conversion of NO2 to NO2 reservoir species, which may be removed from the
atmosphere or converted back to NO2 at sunrise.
2.2 Ambient Ratio Method
The original ARM is based on (Chu & Meyer, 1991) and is focused on long-term averages of NOx
concentrations. The basis for the approach assumes that for long-term averages, the partitioning of NO
and NO2 at any location is controlled by the NO/NO2 partitioning in the ambient background air
because: 1) plumes are mixing with ambient air, and 2) as plumes photochemically age, they will
ultimately approach the background equilibrium. It can also be assumed than any sources near a
monitor will periodically impact a monitor and as a result, the average NO2/NOx ratio will reflect plumes
in various stages of mixing and photochemical aging. Since most sources are expected to have relatively
low in-stack ratios (ISR) of NO2/NOx, such periodic impacts of a source on a monitor would be expected
to reduce the long-term average NO2/NOx ratio. While this would make the NO2/NOx ratio from the
-------
monitor more representative of those local course impacts, it would make the NO2/NOx ratio less
conservative when predicting impacts from new or modified sources. Nevertheless, a NO2/NOx ratio
controlled by background conditions would still be expected to be conservative for estimating the long-
term average impacts from a new or modifying source. Thus, long-term averages of the NO2/NOx ratio
at any monitor can be used as a surrogate for estimating the long-term NO2/NOx ratio at other
locations.
The validity of the ARM assumptions were also confirmed with several sets of ambient data. First, (Chu
& Meyer, 1991) evaluated 10 years of ambient data from nationwide US networks to determine trends
of the long-term averages of the NO2/NOx ratios. Their analysis had four major conclusions:
1. The annual NO2/NOx ratios ranged from 0.48 to 0.74 over the 10-year period.
2. There was relatively little variation in the median, maximum and minimum ratios over the data
period.
3. The year-to-year variability at any single monitor was less than 10%.
4. The 90th percentile value for the most recent 3-years of data (1987-1989) was 0.75.
The fundamental conclusion drawn from the ambient data was that the annual average of the ambient
NO/2NOx ratio was fairly stable and robust, and 0.75 was recommended as a national default value.
Second, in addition to their own analysis of the available ambient data, several older studies, which
focused on sampling aging plumes as they were transported downwind, were also cited ((Keifer, 1977),
(Janssen, Wakeren, Duuren, & Elshout, 1988), (Builtjes, 1981). These studies generally showed that the
NO2/NOx ratio tends to increase with increasing distance from major sources.
Based on the behavior of the nationwide monitoring data and source-specific studies, the recommended
application for ARM included:
1. Use the 3-year average of annual NO2/NOx ratios in the vicinity of the new or modifying source
(within 15 to 90 km) to apply to annual averaged modeled NOx to determine the annual NO2.
2. Monitored values below 20 ppb of NO2 should be excluded when calculating the annual
averages due to uncertainty at these lower concentrations.
3. If the monitor is located in a major urban area, then only daytime measurements should be used
in the annual averages.
4. The national default should be used in absence of any nearby monitoring data.
ARM was implemented in the 2005 version of Appendix W with the national default value of 0.75 for
annual averages, with a mention of using regional ratios as an alternative. Given that there was no
hourly standard available at the time, there was no national default given for short-term standards.
However, the EPA issued a guidance memorandum that recommended a national default NO2/NOX
ratio for usage for the hourly standard (0.8) (U. S. EPA, 2011).
2.3 Ambient Ratio Method 2
The updated ARM2 was developed by (Podrez, 2015) and is focused on short-term (i.e., hourly) averages
of NO2 concentrations. The basis for the approach assumes that due to the nature of NOx chemistry and
typical NO2/NOx emission ratios, that the ambient NO2/NOx ratios will exhibit a predictable pattern,
which will in large part, be a function of the total NOx present. Given this assumption, a 10-year record
(2001-2010) of ambient data from the Air Quality System (AQS, 2014) was analyzed to characterize the
-------
behavior of these data. The analysis showed a decreasing NO2/NOx ratio with increasing total NOx
concentrations. In order to develop a NO2/NOx ratio response curve, the ambient data was binned
based on the NOx concentrations, the 98th percentile NO2/NOx ratio was determined for each NOx bin,
and a 6th order polynomial regression was performed on the selected values from each NOx bin. The
resulting polynomial is used to predict the NO2/NOx ratio based on the modeled NOx concentration,
though maximum and minimum limits on the NO2/NOx ratio are used at the upper and lower ends of
the range of potential NOx concentrations (0.9 maximum and 0.2 minimum). Figure 2 shows an example
of AQS data, comparing the NO2/NOx ratio to the total NOx, demonstrating the relationship found by
(Podrez, 2015).
0.0
100
200 300
NOx (ppb)
400
500
Figure 2 - NO2/NOx ratios from AQS for 2012
In addition to developing the ARM2 response curve, (Podrez, 2015) evaluated ARM2 against PVMRM
and OLM for three ambient datasets and performed a large number of sensitivity tests with model-
based evaluations. The ambient datasets include: 1) a gas processing plant in Empire Abo, NM; 2) a coal-
fired power plant in Palaau, HI; and 3) a small diesel-fired power plant in Wainwright, AK. The sensitivity
tests included scenarios taken from an earlier report on OLM and PVMRM performance (MACTEC, 2004)
and tests to address generalized facilities representative of several other source (diesel generators, an
oil refinery, a gas pipeline compressor station, natural gas production fields and processing plants, and a
large boiler). These comparisons showed that ARM2 gave generally conservative estimates of the
NO2/NOx ratio and the total NO2 as compared to the measurements as well as OLM and PVMRM. The
model evaluation report, (RTP Environmental Associates, Inc., 2013), provides more complete details on
the evaluation results than the journal article.
-------
2.4 Ozone Limiting Method
(Cole & Summerhays, 1979) is most often cited as the origin of OLM even though this particular paper is
a review paper of methods to estimate NO2 concentrations since there are no earlier citations for OLM.
As outlined by (Cole & Summerhays, 1979), OLM assumes NO conversion to NO2 by reaction with
ambient ozone (NO + 03 -> N02 + 02 Reaction 1), according to the
following procedure:
1. Estimate total NOx using the dispersion model of choice.
2. Divide the total NOx into two components:
a. Component 1 is the "thermal" NO2 that is emitted directly as NO2 from the stack
(assumed to be 0.10 for most sources)
b. The remaining NOx is assumed to be NO that is available for reaction with ozone.
3. If the ambient [ozone] > 0.9*[NOx] (the portion of NOx present as NO), then assume that all NO
is converted to NO2, such that [NO2] = [NOx]. If the ambient [ozone] < 0.9*[NOx], then assume
that the amount of NO converted is equal to the available ozone, such that [NO2] = 0.1*[NOx] +
[ozone].
4. The NO2 concentration from (3) is added to the background [NO2] to determine the total [NO2].
The reaction is assumed to be instantaneous and irreversible and can be applied on an hourly basis (or
shorter or longer, depending on the resolution of the model run and background [NO2] and [ozone]
data).
(Cole & Summerhays, 1979) presented no evaluations of OLM, but there have been a number of formal
and informal studies that have evaluated OLM based on AERMOD and other dispersion models. The
majority of these studies have simultaneously performed evaluations of OLM and PVMRM with results
of these studies being summarized below for both OLM and PVMRM.
2.5 Plume Volume Molar Ratio Method
PVMRM was developed by (Hanrahan, 1999a) and is based on the same fundamental chemistry as OLM
(NO + 03 -> N02 + 02 Reaction 1), which is also assumed to be
instantaneous and irreversible and can be applied on an hourly basis (or shorter or longer, depending on
the resolution of the model run and background [NO2] and [ozone] data). In contrast to OLM, though,
PVMRM attempts to estimate the amount of ozone that has been entrained into the plume(s), rather
than assuming all ambient ozone is available for NO conversions. (Hanrahan, 1999a) describes the
methodology for PVMRM as follows:
1. Calculated the plume segment volume at the receptor
2. Determine the moles of NOx present in the plume segment volume
3. Determine the moles of ozone present in the plume segment volume
4. Calculate the ratio of ozone moles to NOx moles
5. Multiple the ratio found in (4) to the NOx moles to determine the moles of NO2
a. The original PVMRM assumed that 10% of the NOx was initially NO2 upon release, so
0.1 was added to the ratio in (4) to account for the initial NO2.
Since PVMRM can account for NO conversion in overlapping plumes, PVMRM includes a methodology
for determining the volume of overlapping plumes. When considering multiple plumes, the plume
volume at each receptor is computed by first identifying the primary source contributing to the NOX at
-------
the receptor. Then, all additional sources contributing at least half of the amount of NOX as the primary
contributing source are identified as "major contributing sources." Third, the maximum width between
the major contributing sources is found. Finally, the maximum width found in the third step is added to
the width of the primary contributing plume to determine the combined plume width. Similarly, for area
sources, PVMRM uses the projected width of the area source as part of determining the combined
plume width.
2.6 Existing evaluations of OLM and PVMRM
The number of field studies available to evaluate refined NO2 modeling in the near-field are fairly
limited. As a result, the measurement-based model evaluations available for OLM and PVMRM tend to
include both models. There are also a number of "sensitivity studies", that are limited to comparisons
modeling scenarios without any measurement data to benchmark model performance. A selection of
the model evaluations are summarized here:
1. (Hanrahan, 1999b) presented an evaluation of PVMRM for several filed studies, including
airplane-based sampling of power plumes in the Netherlands and ground-based measurements
from power plants in Hawaii (Palaau) and New Mexico (Empire Abo) as well as a comparison
with a large eddy simulation (LES) model. The evaluations were focused on NO2/NOx ratios
rather than absolute NO2 and NOx concentrations and the underlying dispersion model was the
Industrial Source Complex (ISC) model. The results of their evaluation showed that the model
had some skill in replicating NO2/NOx ratios, though there were obvious deficiencies in the
model.
2. (MACTEC, 2004) presented sensitivity tests for a variety of single and multi-source scenarios for
OLM and PVMRM based on an early version of AERMOD and AERMET. The results of their
evaluation showed that both models were reasonably sensitive to source parameters, the
NO2/NOx ratios tended to be lower for PVMRM than for OLM, and that the NO2/NOx ratio
tended to be controlled by the volume of the plume for PVMRM versus being controlled by the
total NOx concentration for OLM.
3. (Hendrick, Tino, Hanna, & Egan, 2013) evaluated OLM and PVMRM using a relatively new field
dataset from a small diesel-fired power generation facility in Wainwright, AK with a relatively
recent version of AERMOD (11103). Additionally, (Hendrick, Tino, Hanna, & Egan, 2013)
suggested several improvements to the model code for PVMRM. This study was based on an
earlier report (Hendrick, Tino, Hanna, & Egan, 2012), which includes additional details on the
Wainwright comparison and a review of other ambient databases used for model evaluations.
Three main conclusions were drawn from the study:
a. OLM has a tendency to overpredict the NO2/NOx ratios, while PVMRM has a tendency
to overpredict the ratio at smaller NOx concentrations and underpredict the ratio at
larger NOx concentrations.
b. Neither OLM nor PVMRM have any skill in predicting NO2 when paired in time and
space with measurements. This appears to be driven more by the model's underlying
NOx prediction rather than with the NO/NO2 speciation.
c. When comparing only the highest unpaired concentrations, both models have a
tendency to overpredict the NO2 concentrations.
-------
4. (Schroeder, 2012) evaluated OLM and PVMRM for a single hypothetical source, but with
meteorology and background ozone taken from 4 different locations (Illinois, New Mexico, Utah,
and Wisconsin). The source was modeled with 2 emissions rates (1 g/s and 100 g/s) and with 2
ISRs (0.5 and 0.1), resulting in 16 sensitivity tests. The results of the study showed that at lower
emission rates, all options showed relatively similar maximum NO2 concentrations, as there was
sufficient ozone relative to NOx for high conversion rates, while the lower emission rates were
sensitive to the ISRs and the ambient ozone.
5. (Podrez, 2015) and the companion report (RTP Environmental Associates, Inc., 2013), though
focused on the development of ARM2, also performed extensive testing of OLM and PVMRM
with the more recent versions of AERMOD. The analysis completed model-to-monitor
comparisons for the field studies from Alaska, New Mexico, and Hawaii, reran the model
sensitivity tests initially done by (MACTEC, 2004), and added a number of additional sensitivity
tests to include other typical NO2 sources. The results of the study showed that OLM and
PVMRM tended to slightly overestimate the highest NO2 as compared to the measurement
data. All methods tended to show relatively similar maximum NO2 concentrations when NOx
concentrations were lower, as there was sufficient ozone relative to NOx for high conversion
rates, while the lower emission rates were sensitive to the ISRs and the ambient ozone.
3. Regulatory status of ARM, ARM2, OLM, and PVMRM
EPA is proposing to adopt ARM2, OLM, and PVMRM as non-default options in AERMOD, such that the
alternative model approval currently needed for these options is no longer required. However, these
methods are still considered screening techniques, and as such, will still require the approval of the
appropriate review authority (paragraph 3.0(b) of the proposed Appendix W). This section discusses
several aspects of this proposal.
3.1 Regulatory requirements
Under EPA's Guideline on Air Quality Models published as Appendix W to 40 CFR Part 51, AERMOD is the
preferred model for satisfying New Source Review (NSR) requirements, including Prevention of
Significant Deterioration (PSD) for inert pollutants. For inert pollutants, AERMOD is considered to be a
refined model, such that within the regulatory framework of a NSR or PSD demonstration, the model is
considered not to be biased high or low. NO and NO2, however, are not inert and conversion from NO
to NO2 must be estimated. At present, under Appendix W, the EPA recommends a 3-Tiered screening
approach. The three Tiers should be most conservative at Tier 1 and least conservative at Tier 3, though
the selection of any option or setting at any Tier should strive towards the most accurate estimates,
rather than the most conservative. While all three Tiers available for NO2 modeling in AERMOD are
considered screening techniques, each Tier has increasing complexity, which includes more refined (and
presumably more accurate) estimates of NO and NO2 speciation, but also increasing data input
requirements. Screening models and techniques should not be biased to under predict, i.e., should give
conservative modeled estimates. Thus, AERMOD is considered unbiased in its estimates of NOx
concentrations, but estimates of NO2 concentrations should be either unbiased or conservative. For the
Tier 2 and 3, the additional data inputs required may not be available for all facilities, necessitating the
-------
determination of default values to insure that modeled concentrations do not underestimate facility
impacts.
3.2 ARM versus ARM2
The original ARM was developed exclusively for long-term average concentrations of NO2 and the intent
was to use national default only when nearby monitoring was not available or at longer transport
distances from a source, when a plume would have mixed towards background. The historical
implementation of ARM in the NSR and PSD programs, however, has been to use the national default at
all distances and all locations, regardless of the availability of nearby monitoring. Fundamentally, this
approach for long-term averages is likely appropriately conservative for a Tier 2 method. However, ARM
was not intended to estimate short-term average concentrations of NO2. On the other hand, ARM2 was
developed exclusively for short-term average concentrations of NO2, specifically 1-hour averages.
Assuming the implementation of ARM2 is appropriately conservative for estimates of 1-hour NO2
concentrations, then ARM2 would be preferred to any modifications of ARM for short-term averages. It
is also logical that if the 1-hour estimates of NO2 derived from ARM2 are appropriate, then any long-
term averages calculated from ARM2-adjusted 1-hour concentrations should also be appropriate. Given
this fact, there is a question as to what method should be used for long term estimates of NO2 or
whether both ARM and ARM2 should be viable options. There are two arguments in favor of ARM2 over
ARM. Firstly, ARM has not been implemented according to the recommendation of (Chu & Meyer,
1991), which would a case-by-case analysis of nearby monitoring data to determine the equilibrium
ratio rather than using the national default for all cases. Secondly, long-term averages based on ARM
and ARM2 would not be identical. Thus, if ARM were used for long-term averages and ARM2 were used
for short-term, there would be a discrepancy between the underlying data in hourly values and the long-
term averages that could be calculated. This is demonstrated in Table 1 - Comparison of ARM and ARM2
estimated NO2 concentrationswhich has an average NOx concentration of 235 ppb. Using the ARM
national default of 0.75, the estimated NO2 concentration would be 176 ppb, while the average
concentration from ARM2 would be 76.3 ppb. Therefore, if ARM2 is an appropriate tool for estimating
1-hour concentrations, it would be preferred to use the hourly estimates from ARM2 to compute longer-
term averages over the application of ARM on average NOx concentrations.
NOx
50
100
150
200
250
400
500
Average
NO2/NOxARM2
curve
96%
70%
50%
38%
31%
22%
19%
Average
ARM2 default
applied
0.90
0.70
0.50
0.38
0.31
0.22
0.20
Average
NO2
45.00
69.74
75.37
76.24
78.64
89.37
100.00
Average
ARM2 EPA
applied
0.90
0.70
0.50
0.50
0.50
0.50
0.50
Average
NO2
45.00
69.74
75.37
100.00
125.00
200.00
250.00
Average
-------
| 235.71
0.47
0.46
76.34
0.59
123.59 |
Table 1 - Comparison of ARM and ARM2 estimated NO2 concentrations
3.3 Status of OLM and PVMRM
Currently, OLM and PVMRM are only available for regulatory usage as alternative models, requiring an
formal request to the appropriate reviewing authority and technical demonstration for their usage. (U.
S. EPA, 2010) and (U. S. EPA, 2011) provided the bulk of the technical demonstration, showing that both
models were appropriate as "alternative models", when applied to the appropriate source
configurations. As a result, the usage of OLM and PVMRM has mainly required an administrative request
for approval from the appropriate reviewing authority. However, there is a reasonably strong case that
both models also meet the requirements of preferred models. Both models are written in a common
programming language; the models are well documented; test datasets available for model evaluation;
the model is useful to typical users; there are robust model-to-monitor intercomparisons; and the
source code is freely available. Given that the EPA has stated that the technical demonstrations
supporting the usage of OLM and PVMRM are broadly acceptable for regulatory applications, there is
notable scientific merit behind OLM and PVMRM. Additionally, the available databases to test the model
indicate that both perform reasonably well and are not biased to underpredict. Therefore, it stands to
reason that both methods should be included in AERMOD as regulatory, non-beta preferred options for
modeling NO2.
3.4 The 3-Tiered approach
Despite the fact that OLM and PVMRM are clearly more refined and accurate than full conversion, ARM,
and ARM2, both OLM and PVMRM are still considered screening rather than refined models at this
point. Thus, a case could be made to eliminate the Tier 1 and 2 methods in preference of OLM and
PVMRM. However, the intent in the tiered screening process is to reduce the burden of performing an
NO2 modeling demonstration by providing easier to implement methods. As a result, the EPA intends to
maintain the 3-tiered screening for NO2 modeling but changing ARM2, OLM, and PVMRM to non-beta
options.
4. Proposed regulatory options for NO2 modeling in AERMOD
The Tier 2 and Tier 3 options for NO2 available within AERMOD have several settings that can affect
their results. For some of these settings, facility or area-specific data is not always available, requiring
considerations that are appropriate for default regulatory applications. This section reviews these
technical details within the context of regulatory requirements.
4.1 ARM2 options
The equation that was derived for ARM2 will indicate NO2/NOx ratios above 1 for lower NOx
concentrations and less than 0 for higher NOx concentrations. As a result, maximum and minimum
NO2/NOx ratios were selected for each end of the distribution. These maximum and minimum values
are discussed further here.
(Podrez, 2015) and (RTP Environmental Associates, Inc., 2013) recommended 0.9 as the maximum ratio,
as this ratio has generally been considered an appropriate equilibrium ratio for background air and
-------
diluted plumes. The ambient data shown in Figure 2 - NO2/NOx ratios from AQS for 2012indicates that
NO2/NOx ratios do exceed 0.9 and achieve a maximum of 1.0. Many of these high NO2/NOx ratios occur
at lower NOx levels, where instrument uncertainty can be a significant source of measurement error.
Additionally, the PSS analysis shown in Figure 1 - NO2/NOx equilibrium ratios based on the PSS
assumption from Equation 1. also indicates that when solar radiation levels are relatively low (2 cases
with 95% cloud cover) that the equilibrium ratio can approach 1.0. However, the PSS values shown in
Figure 1 - NO2/NOx equilibrium ratios based on the PSS assumption from Equation 1. assume that
ozone concentrations remain constant, i.e., ozone depletion is not accounted for in the calculations. As a
result, the NO2/NOx ratios shown in Figure 1 - NO2/NOx equilibrium ratios based on the PSS
assumption from Equation 1. overestimate what would happen in a plume, with ozone titration
occurring while converting NO to NO2.
In addition, the EPA has established that the equilibrium ratio of 0.9 is appropriate for Tier 3 methods
based on the recommendations for PVMRM given by (Hanrahan, 1999a). Though there is evidence that
equilibrium ratios can be above 0.9, these cases are generally at lower NOx concentrations and only
occur when ozone levels are very high or when ambient light is very low. Since ozone has a daytime
maximum and a nighttime minimum, the NO2/NOx ratios for higher ozone levels for even moderate
light conditions maximize at 0.9. Again, this is an overestimate, as ozone is assumed to be held constant
rather than titrated by NO). Likewise, in lower light levels and at nighttime, the ambient ozone will be
low and titrated quickly. The resulting NO2/NOx ratios for the lower-light scenarios are shown in Figure
1. These ratios would be associated with the lower ozone concentrations only, resulting in a
conservative estimate of NO2/NOx ratios around 0.9. Thus, based on the information from the PSS
analysis, the maximum NO2/NOx ratio of 0.9 is reasonable and appropriate as a maximum ratio for
ARM2.
• PVMRM, 0.2 ISR max
• PVMRM, 0.5 ISR max
• PVMRM, 0.2 ISR DV
PVMRM, 0.5 ISR DV
100%
90%
Figure 3 - Comparison of modeled NO2 concentrations from ARM2 and PVMRM
-------
(Podrez, 2015) and (RTP Environmental Associates, Inc., 2013) recommended a minimum ratio of 0.2, as
this value seemed to be the maximum ratio from the ambient data at the higher NOx concentrations.
The minimum ratio of 0.2 appears to be fairly well supported by the ambient data in that there are few
data points above the 0.2 NO2/NOx ratio at higher NOx concentrations. However, the regulatory
requirements are important to consider when determining a national default value for the minimum
ratio. (Owen & Brode, 2014) discuss at length the regulatory implications of the minimum NO2/NOx
ratio, concluding that the default 0.2 cannot appropriately account for all sources. While there is general
agreement that most sources have relatively low ISRs, there are sources with higher ISRs and it would be
inappropriate to allow sources to base their modeling on lower ISRs than they actually have. As
discussed in (Owen & Brode, 2014), the minimum ambient NO2/NOx ratios in a plume are first and
foremost driven by the ISR and that the NO2/NOx ratio can only increase from the ISR as mixing and
conversions occurs in the atmosphere. (U. S. EPA, 2011) established a national default ISR of 0.5 for
sources that did not have site-specific data (which is always preferred over default values) for the Tier 3
methods. This value was considered appropriately conservative, given that most sources are likely to
have much lower ISR. (Owen & Brode, 2014) showed that a minimum NO2/NOx ratio in ARM2 of 0.5
produced appropriately conservative results for a Tier 2 method relative to sources modeled with
PVMRM and an IRS of 0.5, as shown in Figure 3 - Comparison of modeled NO2 concentrations from
ARM2 and PVMRM. Based on the assumption that the 0.5 ISR is appropriate for OLM and PVMRM, then
a minimum NO2/NOx ratio of 0.5 should also be appropriate as a minimum ratio for ARM2 in evaluating
sources with an unknown ISR.
4.2 Default options for OLM and PVMRM
There are two additional inputs required for OLM and PVMRMR that are not required for the other NO2
options. In addition to the equilibrium ratio discussed above, the source's ISR and background ozone
concentrations are also required model inputs.
The NO2/NOx equilibrium ratio of 0.9 recommended for ARM2 is also recommended for OLM and
PVMRM, based on the information from the PSS analysis and ambient data considerations presented in
Section 4.1.
-------
•All ratios
Non-zero ratios
0.1
0.2
0.3 0.4 0.5 0.6
NOZ/NOx In-stack Ratio
0.7
0.8
0.9
Figure 4 - Distribution of NO2 ISRs from EPA's database as of June, 2015.
When modeling with OLM and PVMRM, each source requires an ISR be specified. As detailed above, the
ISR is the initial driving factor in the ambient NO2/NOx ratios and thus can have a major impact on
estimated NO2 concentrations nearest each source. The EPA recently initiated the collection of
NO2/NOx ISR measurements for compilation in a publically available database for reference in
regulatory applications (http://www.epa.gov/scram001/no2 isr database.htm). The database
represents one of the larger collection of samples of NO2/NOx ISRs and provides an opportunity to
evaluate the relative conservativeness of any particular ISR. The database has over 2300 entries, though
only 1580 are non-zero entries. The entries with an ISR of 0 indicate that there is no NO2 in the exit gas
stream, a case that is extremely unlikely. Most NO2 instruments would still report some low level of
NO2 even when no NO2 is present (e.g., less than the detection limit, commonly 1 ppb). The fact that
zero values (0 ppb of NO2) are reported for roughly 700 entries suggests that NO2 was not actually
measured (rather than measurement being collected and concentrations of 0 ppb of NO2 being
reported). Therefore, the focus here is on the remaining non-zero entries.
As shown in Figure 4, over 93% of the remaining ISR are less than the traditional default of 0.5, which
strongly indicates that this default value will be conservative for most sources and is a reasonable
default to insure model results do not underpredict potential source impacts. (Owen & Brode, 2014)
recommended that, in the absence of source-specific data, the national default of 0.5 be applied to the
new or modifying source and sources in the immediate vicinity of the new or modifying source.
However, since the ISR has a diminishing impact on ambient NO2/NOx ratios as a plume is transported
farther downwind due to mixing and reaction towards background ambient NO2/NOx ratios, (Owen &
Brode, 2014) recommended that more distant sources could use a default ISR of 0.2 in lieu of source
specific data. According to the EPA ISR database, 67% of the tests indicated ISRs less than of 0.2. Given
that the greatest impact of the NO2/NOx in-stack ratio is typically on the closest/closer receptors to a
source and that the highest modeled impact also generally is close to a source, it is reasonable to set 0.2
as a default for the more distant nearby sources. (Owen & Brode, 2014) suggested sources that use a
-------
default of 0.2 will generally be greater than 1 to 3 km away from the primary source, with the distance
dependent upon the relative strength of the primary and background sources as well as the relative
location of the background sources with respect to the prominent wind direction and location of
expected maximum impacts from the primary source.
The third required input for OLM and PVMRM are background ozone concentrations. Traditionally, there
have not been default inputs for background ozone. Instead, hourly data from a nearby ozone monitor is
used (e.g., ozone data taken from AQS (AQS, 2014)). Data gaps from a particular monitor can be filled
with: 1) data from another monitor, 2) interpolation between existing data, or 3) substitution with
maximum data from the same monitor for similar time periods (e.g., maximum ozone concentrations
from nighttime winter might be used to fill a missing hour from 3 am in January). It would be possible to
set a conservative national default ozone value (e.g., 90 ppb would likely be conservative in most cases).
However, ozone monitors are ubiquitous throughout the nation and ozone is a regional pollutant. Thus,
ambient monitoring data from a more distant monitor could easily be appropriate. These facts make it
relatively straight-forward to find reasonably representative background ozone data. Given this
situation, no default background ozone concentration is suggested for implementation in the model and
hourly, site-specific data is recommended.
4.3 Appropriate applications of OLM and PVRMR
The Tier 3 methods of OLM and PVMRM each have strengths and weaknesses that should be considered
when applied to a particular modeling scenario. (U. S. EPA, 2010), (U. S. EPA, 2011), and (Owen & Brode,
2014) discussed these aspects of the two models at length. In summary, these memos recommend the
following applications for each model:
• PVMRM is recommended for relatively isolated, elevated sources.
• PVMRM is not recommended for area or line sources, near-surface releases, or groups of
sources with moderate distances between them due to the potential to overestimate plume
volumes in these cases.
• OLM is recommended for all sources where PVMRM is anticipated to overestimate plume
volumes, including roadway emissions from mobile sources.
• For most cases, the OLMGROUP ALL option is recommended when using OLM.
4. PVMRM2
In addition to changing the status of PVMRM from an alternative model to an accepted Tier 3 screening
technique for NO2, we are proposing modifications to the current PVMRM formulation. The new
PVMRM, called PVMRM2, has 3 major modifications from the current PVMRM formulation:
1. The current formulation of PVMRM uses relative dispersion coefficients, which are most
appropriate for convective conditions but are likely to overestimate the plume volume in stable
conditions, and therefore the NO2 concentrations during stable conditions. The PVMRM2 option
utilizes the original relative coefficients for convective conditions, but uses a new formulation
based on total dispersion coefficients for stable conditions, with appropriate adjustments to the
number of sigmas used to define the plume volume.
2. The original PVMRM formulation uses a radial distance from the source to receptor to define
the plume volume, which determines the amount of O3 available for NO-NO2 conversion.
-------
However, due to the meander component, plume impacts can occur upwind of the source
resulting from upwind dispersion. These impacts can occur at receptors that are far off the
plume centerline. Using the radial distance does not take into account the plume width
associated with the concentrations far from the plume centerline. PVMRM2 includes
modifications to more accurately account for plume volumes in these situations.
3. In the process of implementing and testing the PVMRM2 option, a number of issues associated
with the treatment of penetrated plumes were address that also affected the PVMRM option.
4.1 Comparison of PVMRM vs. PVMRM2 Options in AERMOD
DISTDOM - distance from source to receptor used to calculate the plume volume and the moles of NOx
contained in the plume:
PVMRM:
For POINT & VOLUME sources: DISTDOM = radial distance from dominant source to
receptor (using center for VOLUMEs);
For AREA/LINE/OPENPIT source: DISTDOM = radial distance from upwind corner of
dominant source to receptor.
PVMRM2:
For POINT & VOLUME sources: DISTDOM = downwind distance from dominant
source (center for VOLUMEs) to
receptor, with minimum of 1m;
For AREA/LINE/OPENPIT source: DISTDOM = downwind distance from center of
dominant source to receptor, with
minimum of 1m.
Rationale: DISTDOM determines the volume of the plume and therefore the moles of O3 within
the plume. DISTDOM is also used to calculate the moles of NOx contained in the plume based
on the wind speed. Applying the original approach for receptors located upwind of the source
may produce physically unrealistic estimates of NO to NO2 conversion, and using the downwind
distance provides a more realistic estimate of NOx conversion consistent with a straight-line,
steady-state plume model.
domNOxmoles - moles of NOx in the plume from source to receptor for dominant source based on
emission rate, distance and wind speed:
PVMRM:
For all source types: domNOxmoles = Q * DISTDOM/UDOM
PVMRM2:
-------
For all source types: domNOxmoles = Q * (l-NO2Ratio)*DISTDOM/UDOM
Rationale: By using the total NOx emission rate (Q), the original PVMRM formulation effectively
double counts a portion of the NO2 emissions by applying the domO3moles/domNOxmoles
ratio to determine the conversion of NO to NO2, and adding that to the in-stack NO2/NOx ratio.
A similar approach is used for sumNOxmoles (moles of NOx in the "combined" plume) except
that AVE_NO2RATIO is used to account for the combined in-stack ratios of NO2 across all
contributing plumes.
CWMIN/CWMAX- Minimum and maximum crosswind distances of major contributing sources used to
determine the "box" within which emissions from all sources are included:
PVMRM:
For POINT & VOLUME sources: CWMIN/CWMAX = range of crosswind distances for
contributing sources (using center
for VOLUMES);
For AREA/LINE/OPENPIT sources: CWMIN/CWMAX = range of crosswind distances for
contributing sources (using center of
source), including the projected width
for AREA, LINE and OPENPIT sources.
PVMRM2:
For POINT & VOLUME sources: CWMIN/CWMAX = range of crosswind distances for
contributing sources (using center for
VOLUME sources);
For AREA/LINE/OPENPIT sources: CWMIN/CWMAX = range of crosswind distances for
contributing sources (using center of
source), without including projected
width for AREA, LINE, and OPENPIT
sources.
Rationale: Including the projected width of AREA/LINE/OPENPIT source to determine the "box" of
major contributing sources could significantly increase the moles of O3 within the combined
plume, and therefore increase the conversion of NO to NO2.
DWMIN/DWMAX - Minimum and maximum downwind distances of major contributing sources used to
determine the "box" within which emissions from all sources are included:
PVMRM:
For POINT & VOLUME sources: DWMIN/DWMAX = range of downwind distances for
-------
contributing sources (using center
for VOLUME);
For AREA/LINE/OPENPIT sources: DWMIN/DWMAX = range of downwind distances for
contributing sources (using center of
source), including the projected length
for AREA, LINE and OPENPIT sources.
PVMRM2:
For POINT & VOLUME sources: DWMIN/DWMAX = range of downwind distances for
contributing sources (using center for
VOLUME sources);
For AREA/LINE/OPENPIT sources: DWMIN/DWMAX = range of crosswind distances for
contributing sources (using center of
source), without including projected
length for AREA, LINE, and OPENPIT
sources.
Rationale: Including the projected length of AREA/LINE/OPENPIT source to determine the "box" of
major contributing sources could significantly increase the moles of O3 within the combined
plume, and therefore increase the conversion of NO to NO2.
Dispersion coefficients used to calculate the volume of the plume:
PVMRM:
For all source types: Relative dispersion coefficients are used to calculate the plume
volume for stable and convective conditions; however, these
coefficients are likely to overestimate the plume volume under
stable conditions, and therefore overestimate NO2
concentrations.
PVMRM2:
For all source types: Relative dispersion coefficients are used to calculate the plume
volume for convective conditions, but the "standard" total
dispersion coefficients are used to calculate plume volume for
stable conditions. However, since total dispersion coefficients
are bimodal (i.e., separate lateral and vertical coefficients) the
stable dispersion coefficients are calculated as SQRT(Oy*Oz).
-------
Rationale: The relative dispersion coefficients used for the original PVMRM option are not
appropriate for stable conditions, and will tend to overestimate plume volume and NO2
concentrations under stable conditions.
Defining the extent of the plume used to calculate the plume volume:
PVMRM:
For all source types:
PVMRM2:
For all source types:
The plume volume calculation is based on a radius of four (4)
times the relative dispersion coefficient, which accounts for
about 99.994% of the plume. A minimum value of 5m is
assumed for the radius of the plume to account for near-source
effects (e.g., downwash), which is consistent with the initial
plume volume assumed by Hanrahan (1999),i.e.,
(1.282/4.0)*15 = 4.8 (rounded to 5)
The radius used to define the plume volume varies based on
stability. For stable conditions, the radius of the plume is
assumed to be 1.282 times the total dispersion coefficient,
consistent with the original PVMRM approach proposed by
Hanrahan (1999). This accounts for about 80% of the plume.
For unstable conditions, the radius of the plume is defined as
2.58 times the relative dispersion coefficient, which accounts
for about 99% of the plume. A minimum value of 9m is assumed
for the radius of the plume to account for near-source effects
(e.g., downwash).
In addition, since AERMOD incorporates a horizontal meander
meander algorithm that increases lateral plume spread beyond
that accounted for based on dispersion coefficients, the number
of sigmas used to define the plume volume is adjusted to
account for meander, i.e.,
NSUBZ = MIN( 2.15, 1.282 * (SYEFF/SY),
where SYEFF is the effective sigma-y value that replicates the
plume centerline concentration associated with meander, but
based on a standard Gaussian plume calculation. An upper
bound of 2.15 is applied, which corresponds with the point at
which the concentration is 10 percent of the centerline value,
and is used to define the extent of the plume in other contexts
-------
(e.g., initial dispersion coefficients for volume sources). The
minimum value of the dispersion coefficient is also adjusted
based on the ratio 1.282/NSUBZ, where NSUBZ is the adjusted
value.
Since the portion of the plume that penetrates above the mixing
height during unstable conditions is treated as a stable plume,
the minimum relative dispersion coefficient for penetrated
plumes (SRMIN3) is a weighted average of the stable and
unstable values based on plume penetration factor (PPF), i.e.,
SRMIN3 = PPF*SRMINS + (1-PPF)*SRMINU
In addition, the value of NSUBZ for penetrated plumes is
adjusted based on PPF-weighted mean of stable and unstable
values of NSUBZ.
Rationale: The approach for the original PVMRM option of using 4 times the dispersion coefficients
to define the volume of the plume is excessive. It accounts for about 99.994% of the plume, and
corresponds to the point at which the concentration is about 0.00005 times the centerline
concentration. That approach effectively assumes full and instant entrainment of ozone within
the plume which is physically unrealistic. The PVMRM2 option provides a more reasonable
account of the plume volume for purposes of NOx conversion.
4.2 Comparisons of PVMRM and PVMRM2 for select field studies
Comparisons of PVMRM and PVMRM2 were conducted for the Empire Abo gas plant in New Mexico and
the Palaau generating station on the island of Molokai. The Empire Abo data set consists of data from
two monitors located near the Empire Abo gas plant, with one monitor located 1.6 km north of the
plant, and the other located 2.5 km south of the plant. Data collection included ozone, NOx, and
meteorological data. Ambient data is available for a two year period, from June, 1993, through June,
1995. Emissions from the 20 stacks at the facility were assumed to be constant. The Palaau generating
station data set consists of data from one monitor 240 meters from the generating stations. Data
collection included ozone, NOx, and meteorological data. Ambient data is available for a one year
period, from January, 1993, through December, 1993.
-------
AERMOD Model - NM N Monitor w/S-OS - 1hr NO2 Q-Q Plot - PVMRM2 w/Adj SBulkRn
1000
100
Observed Cone (|Jg/m3)
1000
Figure 5 - Comparison of PVMRM and PVMRM2 for the Empire Abo north monitor
-------
AERMOD - NM S Monitor w/N-OS - 1hr NO2 Q-Q Plot - PVMRM2 wAdj & BulkRn
1000
O 100
I
T3
10
10
X
X
100
Observed Cone (|jg/m3)
X
X
1000
Figure 6 - Comparison of PVMRM and PVMRM2 for the Empire Abo south monitor
1000
AERMOD - Palaau, HI - PVMRM1 Opt v15DFT w/Adj
10 100
Observed Cone (|jg/m3)
1000
Figure 7 - QQ plot of PVMRM for the Palaau monitor
-------
1000
AERMOD - Palaau, HI - PVMRM2 Opt v15DFT w/Adj
10 100
Observed Cone (|jg/m3)
1000
Figure 8 - QQ plot of PVMRM2 for the Palaau monitor
There is clearly a substantial improvement in the results for the Empire Abo field study. Both the north
and south monitor show PVMRM2 significantly outperforms PVMRM and is the best performing model
from the comparison. For the south monitor, PVMRM is outside the factor of 2 line for highest
concentrations, while the PVMRM2 results are within 20% of the measurements. For Palaau, PVMRM2
performs slightly better for the highest few concentrations, but tends to underpredict slightly for much
of the upper end of the distribution. Overall, the PVMRM2 performance for Palaau seems to be slightly
worse, but the difference is marginal, particularly when compared to the model performance found in
the Empire Abo data set. Based on these findings, along with the technical bases for these changes, the
EPA believes that it is appropriate to make PVMRM2 the new default, replacing the current
implementation of PVMRM.
-------
References
AQS. (2014). Download Detailed AQS Data. Retrieved from
http://www.epa.gov/ttn/airs/airsaqs/detaildata/downloadaqsdata.htm
Atkinson, R., Baulch, D. L, Cox, R. A., Crowley, J. N., Hampson, R. P., Hynes, R. G.,. . . Troe, J. (2004).
Evaluated kinetic and photochemical data for atmospheric chemistry: Volume I - gas phase
reactions of Ox, HOx, NOx and SOx species. Atmos. Chem. Phys, 4, 1461-1738.
Builtjes, P. J. (1981). Chemically reacting plume experiments for the wind tunnel. Netherlands
Organization for Applied Scientific Research. Division of Technology for Society Ref. No. 81-
013563, File No. 8710-20330.
Chu, S., & Meyer, E. (1991). Use of Ambient Ratios to Estimate Impact of NOx Sources on Annual N02
Concentrations. Proceedings of the 84th Annual Meeting & Exhibition, (pp. 91-180.6). Pittsburgh,
PA: Air & Waste Management Association.
Cole, H. S., & Summerhays, J. E. (1979). A Review of Techniques Available for Estimating Short-Term NO2
concentrations. J. Air Poll. Cont. Assoc., 29:8, 812-817. doi:10.1080/00022470.1979.10470866
Dickerson, R., Stedman, D., & Delany, A. (1982). Direct Measurements of Ozone and Nitrogen Dioxide
Photolysis Rates in the Troposphere. J. Geophys. Res., 87(C7), 4933-4946.
Hanrahan, P. L. (1999a). The Plume Volume Molar Ratio Method for Determining NO2/NOx Ratios in
Modeling—Part I: Methodology. J. Air & Waste Manage. Assoc., 1324-1331.
Hanrahan, P. L. (1999b). The Plume Volume Molar Ratio Method for Determining NO2/NOx Ratios in
Modeling—Part II: Evaluation Studies. J. Air & Waste Manage. Assoc., 1332-1338.
Hendrick, E., Tino, V., Hanna, S., & Egan, B. (2012). Review and Evaluation of the Plume Volume Molar
Ratio Method (PVMRM) and Ozone Limiting Method (OLM)for short-term (1-hour average) NO2
Impacts. Washington, DC: The American Petroleum Institute.
Hendrick, E., Tino, V., Hanna, S., & Egan, B. (2013). Evaluation of NO2 predictions by the plume volume
molar ratio method (PVMRM) and ozone limiting method (OLM) in AERMOD using new field
observations. J. Air & Waste Mgt. Assoc., 844-854. doi: 10.1080/10962247.2013.798599
Janssen, L, Wakeren, J. V., Duuren, H. V., & Elshout, A. (1988). A classification of no oxidation rates in
power plant plumes based on atmospheric conditions. Atm. Env., 43-53.
Keifer, W. S. (1977). The generation of ozone in plumes from large point sources dealing with power
plants and other rural point sources. PhD dissertation, University of Maryland.
MACTEC. (2004). Sensitivity Analysis of PVMRM and OLM in AERMOD. Contract No. 18-8018-04: Alaska
DEC.
Owen, R. C, & Brode, R. (2014, Sept. 30). Clarification on the Use of AERMOD Dispersion Modeling for
Demonstrating Compliance with the N02 National Ambient Air Quality Standard.
Podrez, M. (2015). An update to the ambient ratio method for 1-h NO2 air quality standards dispersion
modeling. Atm. Env., 163-170.
-------
RTP Environmental Associates, Inc. (2013, March 3). Ambient Ratio Method Version 2 (ARM2) for use
with AERMODfor 1-hr NO2 Modeling. Retrieved from
http://www.epa.gov/ttn/scram/models/aermod/ARM2_Development_and_Evaluation_Report-
September_20_2013.pdf
Schroeder, A. J. (2012). AERMOD tiering approach case study for 1-hour NO2.105th Air and Waste
Management Association Annual Conference and Exhibition, (p. Paper No 481). San Antonio, TX .
Seinfeld, J., & Pandis, S. (2006). Atmospheric Chemistry and Physics: From Air Pollution to Climate
Change. New York: John Wiley & Sons.
U. S. EPA. (2010, June 28). Applicability of Appendix W Modeling Guidance for the 1-hour NO2 National
Ambient Air Quality Standard.
U. S. EPA. (2011, March 1). Additional Clarification Regarding Application of Appendix W Modeling
Guidance for the 1-hour NO2, National Ambient Air Quality Standard.
Zel'dovich, Y., Sadonikov, P. Y., & Frank-Kamenetskii, D. A. (1946). The Oxidation of Nitrogen in
Combustion Explosions. Acta Physicochimica U.S.S.R. 21, 577-628.
-------
-------
United States Office of Air Quality Planning and Standards Publication No. EPA- 454/B-15-004
Environmental Protection Air Quality Analysis Division [June, 2015]
Agency Research Triangle Park, NC
------- |