* \
*1 PROt^
Bayesian Space-time Downscaling Fusion
Model (Downsealer) - Derived Estimates of Air
Quality for 2017
-------
-------
EPA-454/R-20-005
August 2020
Bayesian Space-time Downscaling Fusion Model (Downscaler) - Derived Estimates of Air
Quality for 2017
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Air Quality Assessment Division
Research Triangle Park, NC
-------
Authors:
Adam Reff (EPA/OAR)
Sharon Phillips (EPA/OAR)
Alison Eyth (EPA/OAR)
David Mintz (EPA/OAR)
Acknowledgements
The following people served as reviewers of this document: Liz Naess (EPA/OAR) and David
Mintz (EPA/OAR).
-------
Contents
Contents 1
1.0 Introduction 2
2.0 Air Quality Data 5
2.1 Introduction to Air Quality Impacts in the United States 5
2.2 Ambient Air Quality Monitoring in the United States 7
2.3 Air Quality Indicators Developed for the EPHT Network 11
3.0 Emissions Data 13
3.1 Introduction to Emissions Data Development 13
3.2 Emission Inventories and Approaches 13
3.3 Emissions Modeling Summary 40
3.4 Emissions References 76
4.0 CMAQ Air Quality Model Estimates 80
4.1 Introduction to the CMAQ Modeling Platform 80
4.2 CMAQ Model Version, Inputs and Configuration 81
5.0 Bayesian space-time downscaling fusion model (downscaler) -Derived Air Quality Estimates 90
5.1 Introduction 110
5.2 Downscaler Model 110
5.3 Downscaler Concentration Predictions Ill
5.4 Downscaler Uncertainties 116
5.5 Summary and Conclusions 118
Appendix A - Acronyms 119
1
-------
1.0 Introduction
This report describes estimates of daily ozone (maximum 8-hour average) and PM2.5 (24-hour average)
concentrations throughout the contiguous United States during the 2017 calendar year generated by
EPA's recently developed data fusion method termed the "downscaler model" (DS). Air quality
monitoring data from the State and Local Air Monitoring Stations (SLAMS) and numerical output from
the Community Multiscale Air Quality (CMAQ) model were both input to DS to predict concentrations
at the 2010 US census tract centroids encompassed by the CMAQ modeling domain. Information on
EPA's air quality monitors, CMAQ model, and DS is included to provide the background and context for
understanding the data output presented in this report. These estimates are intended for use by
statisticians and environmental scientists interested in the daily spatial distribution of ozone and PM2.5.
DS essentially operates by calibrating CMAQ data to the observational data, and then uses the resulting
relationship to predict "observed" concentrations at new spatial points in the domain. Although similar
in principle to a linear regression, spatial modeling aspects have been incorporated for improving the
model fit, and a Bayesian1 approach to fitting is used to generate an uncertainty value associated with
each concentration prediction. The uncertainties that DS produces are a major distinguishing feature
from earlier fusion methods previously used by EPA such as the "Hierarchical Bayesian" (HB) model
(McMillan et al, 2009). The term "downscaler" refers to the fact that DS takes grid-averaged data
(CMAQ) for input and produces point-based estimates, thus "scaling down" the area of data
representation. Although this allows air pollution concentration estimates to be made at points where no
observations exist, caution is needed when interpreting any within-gridcell spatial gradients generated by
DS since they may not exist in the input datasets. The theory, development, and initial evaluation of DS
can be found in the earlier papers of Berrocal, Gelfand, and Holland (2009, 2010, and 2011).
EPA's Office of Air and Radiation's (OAR) Office of Air Quality Planning and Standards (OAQPS)
provides air quality monitoring data and model estimates to the Centers for Disease Control and
Prevention (CDC) for use in their Environmental Public Health Tracking (EPHT) Network. CDC's
EPHT Network supports linkage of air quality data with human health outcome data for use by various
public health agencies throughout the U.S. The EPHT Network Program is a multidisciplinary
collaboration that involves the ongoing collection, integration, analysis, interpretation, and dissemination
of data from: environmental hazard monitoring activities; human exposure assessment information; and
surveillance of noninfectious health conditions. As part of the National EPHT Program efforts, the CDC
led the initiative to build the National EPHT Network (http:// www.cdc.gov/nceh/tracking/default.htm).
The National EPHT Program, with the EPHT Network as its cornerstone, is the CDC's response to
requests calling for improved understanding of how the environment affects human health. The EPHT
Network is designed to provide the means to identify, access, and organize hazard, exposure, and health
data from a variety of sources and to examine, analyze and interpret those data based on their spatial and
temporal characteristics.
Since 2002, EPA has collaborated with the CDC on the development of the EPHT Network. On
September 30, 2003, the Secretary of Health and Human Services (HHS) and the Administrator of EPA
1 Bayesian statistical modeling refers to methods that are based on Bayes' theorem, and model the world in terms of
probabilities based on previously acquired knowledge.
2
-------
signed a joint Memorandum of Understanding (MOU) with the objective of advancing efforts to
achieve mutual environmental public health goals2. HHS, acting through the CDC and the Agency for
Toxic Substances and Disease Registry (ATSDR), and EPA agreed to expand their cooperative
activities in support of the CDC EPHT Network and EPA's Central Data Exchange Node on the
Environmental Information Exchange Network in the following areas:
Collecting, analyzing and interpreting environmental and health data from both agencies (HHS
and EPA).
Collaborating on emerging information technology practices related to building, supporting,
and operating the CDC EPHT Network and the Environmental Information Exchange
Network.
Developing and validating additional environmental public health indicators.
Sharing reliable environmental and public health data between their respective networks in an
efficient and effective manner.
Consulting and informing each other about dissemination of results obtained through work
carried out under the MOU and the associated Interagency Agreement (IAG) between EPA and
CDC.
The best available statistical fusion model, air quality data, and CMAQ numerical model output were
used to develop the estimates. Fusion results can vary with different inputs and fusion modeling
approaches. As new and improved statistical models become available, EPA will provide updates.
Although these data have been processed on a computer system at the EPA, no warranty expressed or
implied is made regarding the accuracy or utility of the data on any other system or for general or
scientific purposes, nor shall the act of distribution of the data constitute any such warranty. It is also
strongly recommended that careful attention be paid to the contents of the metadata file associated with
these data to evaluate data set limitations, restrictions or intended use. The EPA shall not be held liable
for improper or incorrect use of the data described and/or contained herein.
The four remaining sections and appendix in the report are as follows:
Section 2 describes the air quality data obtained from EPA's nationwide monitoring network
and the importance of the monitoring data in determining health potential health risks.
Section 3 details the emissions inventory data, how it is obtained and its role as a key input into
the CMAQ air quality computer model.
Section 4 describes the CMAQ computer model and its role in providing estimates of pollutant
2The original HHS and EPA MOU is available at https://www.cdc.gov/nceh/tracking/pdfs/epa mou 2007.pdf.
3
-------
concentrations across the U.S. based on 12-km grid cells over the contiguous U.S.
Section 5 explains the downscaler model used to statistically combine air quality monitoring
data and air quality estimates from the CMAQ model to provide daily air quality estimates for
the 2010 U.S. census tract centroid locations within the contiguous U.S.
Appendix A provides a description of acronyms used in this report.
Appendix B is a spreadsheet that shows emissions totals for the modeling domain and for each
emissions modeling sector (see Section 3 for more details).
4
-------
Air Quality Data
To compare health outcomes with air quality measures, it is important to understand the origins of those
measures and the methods for obtaining them. This section provides a brief overview of the origins and
process of air quality regulation in this country. It provides a detailed discussion of ozone (O3) and
particulate matter (PM). The EPHT program has focused on these two pollutants, since numerous studies
have found them to be most pervasive and harmful to public health and the environment, and there are
extensive monitoring and modeling data available.
2.1 Introduction to Air Quality Impacts in the United States
2.1.1 The Clean Air Act
In 1970, the Clean Air Act (CAA) was signed into law. Under this law, EPA sets limits on how much of
a pollutant can be in the air anywhere in the United States. This ensures that all Americans have the same
basic health and environmental protections. The CAA has been amended several times to keep pace with
new information. For more information on the CAA, go to https://www.epa.gov/clean-air-act-overview.
Under the CAA, the EPA has established standards, or limits, for six air pollutants known as the criteria
air pollutants: carbon monoxide (CO), lead (Pb), nitrogen dioxide (NO2), sulfur dioxide (SO2), ozone
(O3), and particulate matter (PM). These standards, called the National Ambient Air Quality Standards
(NAAQS), are designed to protect public health and the environment. The CAA established two types of
air quality standards. Primary standards set limits to protect public health, including the health of
"sensitive" populations such as asthmatics, children, and the elderly. Secondary standards set limits to
protect public welfare, including protection against decreased visibility, damage to animals, crops,
vegetation, and buildings. The CAA requires EPA to review these standards at least every five years. For
more specific information on the NAAQS, go to https://www.epa.gov/criteria-air-pollutants/naaqs-table.
For general information on the criteria pollutants, go to https://www.epa.gov/criteria-air-pollutants.
When these standards are not met, the area is designated as a nonattainment area. States must develop
state implementation plans (SIPs) that explain the regulations and controls it will use to clean up the
nonattainment areas. States with an EPA-approved SIP can request that the area be designated from
nonattainment to attainment by providing three consecutive years of data showing NAAQS compliance.
The state must also provide a maintenance plan to demonstrate how it will continue to comply with the
NAAQS and demonstrate compliance over a 10-year period, and what corrective actions it will take
should a NAAQS violation occur after designation. EPA must review and approve the NAAQS
compliance data and the maintenance plan before designating the area; thus, a person may live in an area
designated as nonattainment even though no NAAQS violation has been observed for quite some time.
For more information on designations, go to https://www.epa.gov/ozone-designations and
https://www.epa.gov/particle-pollution-designations.
2.1.2 Ozone
Ozone is a colorless gas composed of three oxygen atoms. Ground level ozone is formed when pollutants
released from cars, power plants, and other sources react in the presence of heat and sunlight. It is the
prime ingredient of what is commonly called "smog." When inhaled, ozone can cause acute respiratory
problems, aggravate asthma, cause inflammation of lung tissue, and even temporarily decrease the lung
5
-------
capacity of healthy adults. Repeated exposure may permanently scar lung tissue. EPA's Integrated
Science Assessments and Risk and Exposure documents are available at
https://www.epa.gov/naaqs/ozone-o3-air-qualitv-standards. The current NAAQS for ozone (last revised
in 2015) is a daily maximum 8-hour average of 0.070 parts per million [ppm] (for details, see
https://www.epa.gOv/ozone-pollution/setting-and-reviewing-standards-control-ozone-pollution#standards.
The CAA requires EPA to review the NAAQS at least every five years and revise them as appropriate in
accordance with Section 108 and Section 109 of the Act. The standards for ozone are shown in Table 2-1.
Table 21. Ozone National Ambient Air Quality Standards
Form of the Standard (parts per million, ppm)
1997
2008
2015
Annual 4th highest daily max 8-hour average, averaged over
three years
0.08
0.075
0.070
2.1.3 Particulate Matter
PM air pollution is a complex mixture of small and large particles of varying origin that can contain
hundreds of different chemicals, including cancer-causing agents like polycyclic aromatic hydrocarbons
(PAH), as well as heavy metals such as arsenic and cadmium. PM air pollution results from direct
emissions of particles as well as particles formed through chemical transformations of gaseous air
pollutants. The characteristics, sources, and potential health effects of particulate matter depend on its
source, the season, and atmospheric conditions.
As practical convention, PM is divided by sizes into classes with differing health concerns and potential
sources3. Particles less than 10 micrometers in diameter (PMio) pose a health concern because they can be
inhaled into and accumulate in the respiratory system. Particles less than 2.5 micrometers in diameter
(PM2.5) are referred to as "fine" particles. Because of their small size, fine particles can lodge deeply into
the lungs. Sources of fine particles include all types of combustion (motor vehicles, power plants, wood
burning, etc.) and some industrial processes. Particles with diameters between 2.5 and 10 micrometers
(PM10-2.5) are referred to as "coarse" or PMc. Sources of PMc include crushing or grinding operations and
dust from paved or unpaved roads. The distribution of PM10, PM2.5 and PMc varies from the Eastern U.S.
to arid western areas.
Particle pollution - especially fine particles - contains microscopic solids and liquid droplets that are so
small that they can get deep into the lungs and cause serious health problems. Numerous scientific
studies have linked particle pollution exposure to a variety of problems, including premature death in
people with heart or lung disease, nonfatal heart attacks, irregular heartbeat, aggravated asthma, decreased
lung function, and increased respiratory symptoms, such as irritation of airways, coughing or difficulty
breathing. Additional information on the health effects of particle pollution and other technical
documents related to PM standards are available at https://www.epa.gov/pm-pollution.
3 The measure used to classify PM into sizes is the aerodynamic diameter. The measurement instruments used for PM are
designed and operated to separate large particles from the smaller particles. For example, the PM2 5 instrument only captures
and thus measures particles with an aerodynamic diameter less than 2.5 micrometers. The EPA method to measure PMc is
designed around taking the mathematical difference between measurements for PMi0and PM2 5
6
-------
The current NAAQS for PM2.5 (last revised in 2012) includes both a 24-hour standard to protect against
short-term effects, and an annual standard to protect against long-term effects. The annual average PM2.5
concentration must not exceed 12.0 micrograms per cubic meter (ug/m ) based on the annual mean
concentration averaged over three years, and the 24-hr average concentration must not exceed 35 ug/m
based on the 98th percentile 24-hour average concentration averaged over three years. More information is
available at https://www.epa.gov/pm-pollution/setting-and-reviewing-standards-control-particulate-
matter-pm-pollution#standards. The standards for PM2.5 are shown in Table 2-2.
Table 22. PM2.5 National Ambient Air Quality Standards
Form of the Standard
(micrograms per cubic meter, jig/m3)
1997
2006
2012
Annual mean of 24-hour averages, averaged over 3 years
15.0
15.0
12.0
98th percentile of 24-hour averages, averaged over 3 years
65
35
35
2.2 Ambient Air Quality Monitoring in the United States
2.2.1 Monitoring Networks
The CAA (Section 319) requires establishment of an air quality monitoring system throughout the U.S.
The monitoring stations in this network have been called the State and Local Air Monitoring Stations
(SLAMS). The SLAMS network consists of approximately 4,000 monitoring sites set up and operated by
state and local air pollution agencies according to specifications prescribed by EPA for monitoring
methods and network design. All ambient monitoring networks selected for use in SLAMS are tested
periodically to assess the quality of the SLAMS data being produced. Measurement accuracy and
precision are estimated for both automated and manual methods. The individual results of these tests for
each method or analyzer are reported to EPA. Then, EPA calculates quarterly integrated estimates of
precision and accuracy for the SLAMS data.
The SLAMS network experienced accelerated growth throughout the 1970s. The networks were further
expanded in 1999 based on the establishment of separate NAAQS for fine particles (PM2.5) in 1997. The
NAAQS for PM2.5 were established based on their link to serious health problems ranging from increased
symptoms, hospital admissions, and emergency room visits, to premature death in people with heart or
lung disease. While most of the monitors in these networks are located in populated areas of the country,
"background" and rural monitors are an important part of these networks. For more information on
SLAMS, as well as EPA's other air monitoring networks go to https://www.epa.gov/amtic.
In 2019, approximately 40 percent of the U.S. population was living within 10 kilometers of ozone and
PM2.5 monitoring sites. Highly populated areas in the eastern U.S. and California are well covered by both
ozone and PM2.5 monitoring network (Figure 2-1).
7
-------
.v*. .& lom ^ ' *~. ^
Lf /»! * ;.:. *Ur 5*r ''.¦*. i' * Jf J$
>./*Oma t- .4«U»«.' .. A ,- ¦ i ' ,* ~
K 'r «T' !> » ~ :'* V * arjf i ' .rV .
Distance to the Nearest Ozone Monitor
< 10 km 954 trar.ts/137 mill
arest Ozone Monitor \ ^ 7 ^***# |
(30,224 tracts/132 million people) ; * 4,
:
(25,065 tracts/120 million people) -- ^
* »r » *
*¦¦ W$r ..><
v
10 km - 25 km
(10,369 tracts/45.7 million people)
25 km - 50 km
(3,935 tracts /14.6 million people)
50 km - 75 km
/o Km - iuu Km (1,864 tracts / 6.9 million people)
100 km-150 km (1512 tracts/5.5 million people)
150 km < (712 tracts / 2.7 million DeoDle^
75 km -100 km
(712 tracts / 2.7 million people)
150 km
& W_ Z W 1 " ^ 4 - . »-. *
1 t«. 'ifc' ¦ I'.'.' j? . *«[»»?*«;? Jk JA "
*v»*»F*4<4«* »
ce to the Nearest PM 2.5 Monitor * *
%" W f ^ v *
<10 km (30,904 tracts /130 million people) # ^ » * \ . *^*S/ ~ ^
0 km-25 km (20,910 tracts /102.5 million people) J * J3t It
WF Jk
Distance to the Nearest PM 2.5 Monitor
< 10 km (30,904 tracts /130 million people)
10 km-25 km (20,910 tracts /102.5 million people)
25 km - 50 km (11,643 tracts / 53.6 million people)
50 km - 75 km (5,638 tracts / 23.5 million)
75 km -100 km (2,309 tracts / 9.1 million people)
100 km -150 km (1,771 tracts/6.9 million people)
150 km < (507 tracts /1.8 million people)
. <
Figure 2-1. Distances from U.S. Census Tract centroids to the nearest monitoring site, 2019.
8
-------
In summary, state and local agencies and tribes implement a quality-assured monitoring network to
measure air quality across the U.S. EPA provides guidance to ensure a thorough understanding of the
quality of the data produced by these networks. These monitoring data have been used to characterize the
status of the nation's air quality and the trends across the U.S. (see https://www.epa.gov/air-trends).
2.2.2 Air Quality System Database
EPA's Air Quality System (AQS) database contains ambient air monitoring data collected by EPA, state,
local, and tribal air pollution control agencies from thousands of monitoring stations. AQS also contains
meteorological data, descriptive information about each monitoring station (including its geographic
location and its operator), and data quality assurance and quality control information. State and local
agencies are required to submit their air quality monitoring data into AQS within 90 days following the
end of the quarter in which the data were collected. This ensures timely submission of these data for use
by state, local, and tribal agencies, EPA, and the public. EPA's OAQPS and other AQS users rely upon
the data in AQS to assess air quality, assist in compliance with the NAAQS, evaluate SIPs, perform
modeling for permit review analysis, and perform other air quality management functions. For more
details, including how to retrieve data, go to https://www.epa.gov/aqs.
2.2.3 Advantages and Limitations of the Air Quality Monitoring and Reporting System
Air quality data is required to assess public health outcomes that are affected by poor air quality. The
challenge is to get surrogates for air quality on time and spatial scales that are useful for EPHT activities.
The advantage of using ambient data from EPA monitoring networks for comparing with health outcomes
is that these measurements of pollution concentrations are the best characterization of the concentration of
a given pollutant at a given time and location. Furthermore, the data are supported by a comprehensive
quality assurance program, ensuring data of known quality. One disadvantage of using the ambient data
is that it is usually out of spatial and temporal alignment with health outcomes. This spatial and temporal
'misalignment' between air quality monitoring data and health outcomes is influenced by the following
key factors: the living and/or working locations (microenvironments) where a person spends their time not
being co-located with an air quality monitor; time(s)/date(s) when a patient experiences a health
outcome/symptom (e.g., asthma attack) not coinciding with time(s)/date(s) when an air quality monitor
records ambient concentrations of a pollutant high enough to affect the symptom (e.g., asthma attack
either during or shortly after a high PM2.5 day).
To compare/correlate ambient concentrations with acute health effects, daily local air quality data is
needed4. Spatial gaps exist in the air quality monitoring network, especially in rural areas, since the air
quality monitoring network is designed to focus on measurement of pollutant concentrations in high
population density areas. Temporal limits also exist. Hourly ozone measurements are aggregated to daily
values (the daily max 8-hour average is relevant to the ozone standard). Ozone is typically monitored
during the ozone season (the warmer months, approximately April through October). However, year-long
data is available in many areas and is extremely useful to evaluate whether ozone is a factor in health
outcomes during the non-ozone seasons. PM2.5 is generally measured year-round. Most Federal Reference
Method (FRM) PM2.5 monitors collect data one day in every three days, due in part to the time and costs
involved in collecting and analyzing the samples. However, in recent years, continuous monitors have
become available which can automatically collect, analyze, and report PM2.5 measurements on an hourly
4 EPA uses exposure models to evaluate the health risks and environmental effects associated with exposure. These models
are limited by the availability of air quality estimates, https://www.epa.gov/technical-air-pollution-resources.
9
-------
basis. These monitors are available in most of the major metropolitan areas. Some of these continuous
monitors have been determined to be equivalent to the FRM monitors for regulatory purposes and are
called FEM (Federal Equivalent Methods).
2.2.4 Use of Air Quality Monitoring Data
Air quality monitoring data has been used to provide the information for the following situations:
(1) Assessing effectiveness of SIPs in addressing NAAQS nonattainment areas
(2) Characterizing local, state, and national air quality status and trends
(3) Associating health and environmental damage with air quality levels/concentrations
For the EPHT effort, EPA is providing air quality data to support efforts associated with (2), and (3) above.
Data supporting (3) is generated by EPA through the use of its air quality data and its downscaler model.
Most studies that associate air quality with health outcomes use air monitoring as a surrogate for exposure
to the air pollutants being investigated. Many studies have used the monitoring networks operated by
state and federal agencies. Some studies perform special monitoring that can better represent exposure to
the air pollutants: community monitoring, near residences, in-house or workplace monitoring, and
personal monitoring. For the EPHT program, special monitoring is generally not supported, though it
could be used on a case-by-case basis.
From proximity-based exposure estimates to statistical interpolation, many approaches are developed for
estimating exposures to air pollutants using ambient monitoring data (Jerrett et al., 2005). Depending
upon the approach and the spatial and temporal distribution of ambient monitoring data, exposure
estimates to air pollutants may vary greatly in areas further apart from monitors (Bravo et al., 2012).
Factors like limited temporal coverage (i.e., PM2.5 monitors do not operate continuously such as recording
every third day or ozone monitors operate only certain part of the year) and limited spatial coverage (i. e.,
most monitors are located in urban areas and rural coverage is limited) hinder the ability of most of the
interpolation techniques that use monitoring data alone as the input. If we look at the example of Voronoi
Neighbor Averaging (VNA) (referred as the Nearest Neighbor Averaging in most literature), rural
estimates would be biased towards the urban estimates. To further explain this point, assume the scenario
of two cities with monitors and no monitors in the rural areas between, which is very plausible. Since
exposure estimates are guaranteed to be within the range of monitors in VNA, estimates for the rural areas
would be higher according to this scenario.
Air quality models may overcome some of the limitations that monitoring networks possess. Models such
as CMAQ can estimate concentrations in reasonable temporal and spatial resolutions. However, these
sophisticated air quality models are prone to systematic biases since they depend upon so many variables
(i.e., metrological models and emission models) and complex chemical and physical process simulations.
10
-------
Combining monitoring data with air quality models (via fusion or regression) may provide the best results
in terms of estimating ambient air concentrations in space and time. EPA's eVNA5 is an example of an
earlier approach for merging air quality monitor data with CMAQ model predictions. DS attempts to
address some of the shortcomings in these earlier attempts to statistically combine monitor and model
predicted data, see published paper referenced in section 1 for more information about DS. As discussed
in the next section, there are two methods used in EPHT to provide estimates of ambient concentrations of
air pollutants: air quality monitoring data and the downscaler model estimate, which is a statistical
'combination' of air quality monitor data and photochemical air quality model predictions (e.g., CMAQ).
2.3 Air Quality Indicators Developed for the EPHT Network
Air quality indicators have been developed for use in the Environmental Public Health Tracking Network
by CDC using the ozone and PM2.5 data from EPA. The approach used divides "indicators" into two
categories. First, basic air quality measures were developed to compare air quality levels over space and
time within a public health context (e.g., using the NAAQS as a benchmark). Next, indicators were
developed that mathematically link air quality data to public health tracking data (e.g., daily PM2 5 levels
and hospitalization data for acute myocardial infarction). Table 2-3 and Table 2-4 describe the issues
impacting calculation of basic air quality indicators.
Table 23. Public Health Surveillance Goals and Current Status
Goal
Status
Air data sets and metadata required for air quality
indicators are available to EPHT state Grantees.
Data are available through state agencies and EPA's
AQS. EPA and CDC developed an interagency
agreement, where EPA provides air quality data along
with statistically combined AQS and CMAQ data,
associated metadata, and technical reports that are
delivered to CDC.
Estimate the linkage or association of PM2.5 and ozone on
health to: Identify populations that may have higher risk
of adverse health effects due to PM2.5 and ozone,
Generate hypothesis for further research, and
Provide information to support prevention and pollution
control strategies.
Regular discussions have been held on health-air linked
indicators and CDC/HFI/EPA convened a workshop
January 2008. CDC has collaborated on a health impact
assessment (HIA) with Emory University, EPA, and
state grantees that can be used to facilitate greater
understanding of these linkages.
Produce and disseminate basic indicators and other
findings in electronic and print formats to provide the
public, environmental health professionals, and
policymakers, with current and easy-to-use information
about air pollution and the impact on public health.
Templates and "how to" guides for PM2.5 and ozone
have been developed for routine indicators. Calculation
techniques and presentations for the indicators have been
developed.
5 eVNA is described in the "Regulatory Impact Analysis for the Final Clean Air Interstate Rule", EPA-452/R-05-002, March
2005, Appendix F.
11
-------
Table 24. Basic Air Quality Indicators used in EPHT, derived from the EPA data delivered to
CDC
Ozone (daily 8-hr period with maximum concentration, ppm. by FRM)
Number of days with maximum ozone concentration over the NAAQS (or other relevant benchmarks (by county
and MSA)
Number of person-days with maximum 8-hr average ozone concentration over the NAAQS & other relevant
benchmarks (by county and MSA)
PM2.5 (daily 24-hr integrated samples, ufi/m3 by FRM)
Average ambient concentrations of particulate matter (< 2.5 microns in diameter) and compared to annual
PM2.5 NAAQS (by state).
% population exceeding annual PM2.5 NAAQS (by state).
% of days with PM2.5 concentration over the daily NAAQS (or other relevant benchmarks (by county and MSA)
Number of person-days with PM2.5 concentration over the daily NAAQS & other relevant benchmarks (by
county and MSA)
2.3.1 Rationale for the Air Quality Indicators
The CDC EPHT Network is initially focusing on ozone and PM2.5. These air quality indicators are based
mainly around the NAAQS health findings and program-based measures (measurement, data and analysis
methodologies). The indicators will allow comparisons across space and time for EPHT actions. They are
in the context of health-based benchmarks. By bringing population into the measures, they roughly
distinguish between potential exposures (at broad scale).
2.3.2 Air Quality Data Sources
The air quality data will be available in the EPA's AQS database based on the state/federal air program's
data collection and processing. The AQS database contains ambient air pollution data collected by EPA,
state, local, and tribal air pollution control agencies from thousands of monitoring stations (SLAMS).
2.3.3 Use of Air Quality Indicators for Public Health Practice
The basic indicators will be used to inform policymakers and the public regarding the degree of hazard
within a state and across states (national). For example, the number of days per year that ozone is above
the NAAQS can be used to communicate to sensitive populations (such as asthmatics) the number of days
that they may be exposed to unhealthy levels of ozone. This is the same level used in the Air Quality
Alerts that inform these sensitive populations when and how to reduce their exposure. These indicators,
however, are not a surrogate measure of exposure and therefore will not be linked with health data.
12
-------
,... 7. ,' <\
3.1 Introduction to Emissions Data Development
The U.S. Environmental Protection Agency (EPA), developed an air quality modeling platform
representing the year 2017 for this project that was based on the April 2020 version of the 2017 National
Emissions Inventory (NEI). This section provides a summary of the emissions inventory and emissions
modeling techniques applied to Criteria Air Pollutants (CAPs) and the following select Hazardous Air
Pollutants (HAPs) included in the modeling platform: chlorine (CI), hydrogen chloride (HC1), benzene,
acetaldehyde, formaldehyde, napthalene, and methanol. This section also describes the approach and data
used to produce emissions inputs to the air quality model. The air quality modeling, meteorological inputs
and boundary conditions are described in a separate section.
The Community Multi scale Air Quality (CMAQ) model (https://www.epa.gov/cmaq) was used to model
ozone (O3) and particulate matter (PM) for this project. CMAQ requires hourly and gridded emissions of
the following inventory pollutants: carbon monoxide (CO), nitrogen oxides (NO\), volatile organic
compounds (VOC), sulfur dioxide (SO:), ammonia (NH3), particulate matter less than or equal to 10
microns (PMin), and individual component species for particulate matter less than or equal to 2.5 microns
(PM: 5) In addition, the Carbon Bond mechanism version 6 (CB6) with chlorine chemistry within CMAQ
allows for explicit treatment of the VOC H APs naphthalene, benzene, acetaldehyde, formaldehyde and
methanol (NBAFM) and includes anthropogenic HAP emissions of HC1 and CI.
The effort to create the 2017 emission inputs for this study included development of emission inventories
for input to a 2017 modeling case, along with application of emissions modeling tools to convert the
inventories into the format and resolution needed by CMAQ. Year-specific emissions were used for all
source categories, including fires and continuous emission monitoring system (CEMS) data for electric
generating units (EGUs). The primary emissions modeling tool used to create the CMAQ model-ready
emissions was the Sparse Matrix Operator Kernel Emissions (SMOKE) modeling system. SMOKE
version 4.7 was used to create CMAQ-ready emissions files for a 12-ktn national grid. Additional
information about SMOKE is available from http://www.cmascenter.org/smoke.
This chapter contains two additional sections. Section 3.2 describes the inventories input to SMOKE and
the ancillary files used along with the emission inventories. Section 3.3 describes the emissions modeling
performed to convert the inventories into the format and resolution needed by CM AQ.
3.2 Emission Inventories and Approaches
This section describes the emissions inventories created for input to SMOKE, which are based on the
April, 2020 version of the 2017 NEI. The NEI includes five main data categories: a) nonpoint (formerly
called "stationary area") sources; b) point sources; c) nonroad mobile sources; d) onroad mobile sources;
and e) fires. For CAPs, the NEI data are largely compiled from data submitted by state, local and tribal
(S/L/T) agencies. HAP emissions data are often augmented by EPA when they are not voluntarily
submitted to the NEI by S/L/T agencies. The NEI was compiled using the Emissions Inventory System
(EIS). EIS includes hundreds of automated QA checks to improve data quality, and it also supports
release point (stack) coordinates separately from facility coordinates. EPA collaboration with S/L/T
agencies helped prevent duplication between point and nonpoint source categories such as industrial
boilers. The 2017 NEI Technical Support Document describes in detail the development of the 2017
13
-------
emission inventories and is available at https://www.epa.gov/air-emissions-inventories/2017-national-
emissions-inventory-nei-technical-support-document-tsd (EPA, 2020a).
Point source data from the 2017 NEI, including data submitted to EIS by S/L/T agencies, were used for
this study. EPA used the SMARTFIRE2 system and the BlueSky emissions modeling framework to
develop year 2017 fire emissions. SM AR.TFIR.E2 categorizes all fires as either prescribed burning or
wildfire categories, and the Bluesky framework includes emission factor estimates for both types of fires.
Onroad and nonroad mobile source emissions for year 2017 were developed by running MOVES2014b
(https://www.epa.gov/moves). Canadian emissions interpolated to the year 2017 from 2015 and 2023
were used, and Mexican emissions were for the year 2016.
The emissions modeling process, performed using SMOKE v4.7, apportions the emissions inventories
into the grid cells used by CMAQ and temporalizes the emissions into hourly values. In addition, the
pollutants in the inventories (e.g., NOx, PM and VOC) are split into the chemical species needed by
CMAQ. For the purposes of preparing the CMAQ- ready emissions, the NEI emissions inventories by
data category are split into emissions modeling platform "sectors"; and emissions from sources other than
the NEI are added, such as the Canadian, Mexican, and offshore inventories. Emissions within the
emissions modeling platform are separated into sectors for groups of related emissions source categories
that are run through all of the SMOKE programs, except the final merge, independently from emissions
categories in the other sectors. The final merge program called Mrggrid combines low-level sector-
specific gridded, speciated and temporalized emissions to create the final CMAQ-ready emissions inputs.
For biogenic emissions, the CMAQ model allows for biogenic emissions to be included in the CM AQ-
ready emissions inputs, or for biogenic emissions to be computed within CM AQ itself (the "inline"
option). This study uses the inline biogenic emissions option.
Table 3-1 presents the sectors in the emissions modeling platform used to develop the year 2017
emissions for this project. The sector abbreviations are provided in italics; these abbreviations are used in
the SMOKE modeling scripts, the inventory file names, and throughout the remainder of this section.
Annual emission summaries for the U.S. anthropogenic sectors are shown in Table 3-2 (i.e., biogenic
emissions are excluded). Table 3-3 provides a summary of emissions for the anthropogenic sectors
containing Canadian, Mexican, and offshore sources. State total emissions for each sector are provided in
Appendix B, a workbook entitled "Appendix_B_2017_emissions_totals_by_sector.xlsx".
14
-------
Table 31. Platform Sectors Used in the Emissions Modeling Process
2014 Platform Sector (Abbrev)
NEI Category
Description and resolution of the data input to SMOKE
Point source aircraft (airports)
Point
2017 NEI point source emissions from airports, including
aircraft and airport ground support emissions. Annual
resolution.
EGUs (ptegu)
Point
2017 NEI point source EGUs, replaced with hourly 2017
Continuous Emissions Monitoring System (CEMS) values
for NOx and SO2 where the units are matched to the NEI.
Emissions for all sources not matched to CEMS data come
from 2017 NEI point inventory. Annual resolution for
sources not matched to CEMS data, hourly for CEMS
sources.
Point source oil and gas
(pt oilgcts)
Point
2017 NEI point sources that include oil and gas production
emissions processes for facilities with North American
Industry Classification System (NAICS) codes related to
Oil and Gas Extraction, Natural Gas Distribution, Drilling
Oil and Gas Wells, Support Activities for Oil and Gas
Operations, Pipeline Transportation of Crude Oil, and
Pipeline Transportation ofNatural Gas. Includes U.S.
offshore oil production. Annual resolution.
Remaining non-EGU point
(ptnonipm)
Point
All 2017 NEI point source records not matched to the
airports, ptegu, or pt_oilgas sectors. Includes 2017-
specific rail yard emissions. Annual resolution.
Point source fire (ptfire)
Fires
Point source day-specific wildfires and prescribed fires for
2017 computed using SMARTFIRE 2 and BlueSky.
Point Source agricultural fires
(ptagfire)
Nonpoint
Agricultural fire sources for year 2017 that were developed
by EPA as point and day-specific emissions6. Agricultural
fires are in the nonpoint data category of the NEI, but in
the modeling platform, they are treated as day-specific
point sources.
Agricultural (cig)
Nonpoint
2017 NEI nonpoint livestock and fertilizer application
emissions. Livestock includes ammonia and other
pollutants (except PM2.5). Fertilizer includes only
ammonia. County and annual resolution.
Area fugitive dust (afdiist adj)
Nonpoint
PM10 and PM2.5 fugitive dust sources from the 2017 NEI
nonpoint inventory; including building construction, road
construction, agricultural dust, and paved and unpaved
road dust. The emissions modeling system applies a
transport fraction reduction and a zero-out based on 2017
gridded hourly meteorology (precipitation and snow/ice
cover). Emissions are county and annual resolution.
Biogenic (beis)
Nonpoint
Year 2017 emissions from biogenic sources. These were
left out of the CMAQ-ready merged emissions, in favor of
inline biogenic emissions produced during the CMAQ
model run itself.
6 Only EPA-developed agricultural fire data were included in this study; data submitted by states to the NEI were excluded.
15
-------
2014 Platform Sector (Abbrev)
NEI Category
Description and resolution of the data input to SMOKE
CI and C2 commercial marine
(cmv clc2)
Nonpoint
2017 NEI Category 1 (CI) and Category 2 (C2),
commercial marine vessel (CMV) emissions based on
Automatic Identification System (AIS) data. Point and
hourly resolution.
C3 commercial marine
(cmv c3)
Nonpoint
Within state and federal waters, 2017 NEI Category 3
commercial marine vessel (CMV) emissions based on AIS
data. Outside of state and federal waters, emissions are
based on AIS data in selected areas, and are gapfilled with
emissions from the Emissions Control Area (ECA)
inventory. Point and hourly resolution.
Remaining nonpoint (nonpt)
Nonpoint
2017 NEI nonpoint sources not included in other platform
sectors. County and annual resolution.
Nonpoint source oil and gas
(np oilgcts)
Nonpoint
Nonpoint 2017 NEI sources from oil and gas-related
processes. County and annual resolution.
Locomotive (rail)
Nonpoint
Line haul rail locomotives emissions for year 2017.
County and annual resolution.
Residential Wood Combustion
(rwc)
Nonpoint
2017 NEI nonpoint sources with residential wood
combustion (RWC) processes. County and annual
resolution.
Nonroad (nonroctd)
Nonroad
2017 nonroad equipment emissions developed with
MOVES2014b. MOVES was used for all states except
California, which submitted their own emissions for the
2017 NEI. County and monthly resolution.
Onroad (onroctd)
Onroad
2017 onroad mobile source gasoline and diesel vehicles
from parking lots and moving vehicles. Includes the
following emission processes: exhaust, extended idle,
auxiliary power units, evaporative, permeation, refueling,
and brake and tire wear. For all states except California,
developed using winter and summer MOVES emission
factors tables produced by MOVES2014b.
Onroad California
(onroctd cct ctdj)
Onroad
California-provided 2017 CAP and metal HAP onroad
mobile source gasoline and diesel vehicles from parking
lots and moving vehicles based on Emission Factor
(EMFAC) 2017, gridded and temporalized based on
outputs from MOVES2014b. Volatile organic compound
(VOC) HAP emissions derived from California-provided
VOC emissions and MOVES-based speciation.
Onroad Canada (onroctd cctn)
Non-US
Monthly onroad mobile inventory for Canada interpolated
to 2017 between 2015 and 2023 (province resolution).
Onroad Mexico (onroctd mex)
Non-US
Monthly onroad mobile inventory from MOVES-Mexico
(municipio resolution) for 2017.
16
-------
2014 Platform Sector (Abbrev)
NEI Category
Description and resolution of the data input to SMOKE
Other area fugitive dust sources
(othafdust)
Non-US
Area fugitive dust sources from Canada interpolated to
2017 between 2015 and 2023, with transport fraction and
snow/ice adjustments based on 2017 meteorological data.
Annual and province resolution.
Other nonpoint and nonroad
(othctr)
Non-US
Year 2017 Canada interpolated between 2015 and 2023
(province resolution) and projected year 2016 Mexico
(municipio resolution, provided by SEMARNAT)
nonpoint and nonroad mobile inventories, annual
resolution.
Other point fugitive dust
sources (othptdnst)
Non-US
2017 point source fugitive dust sources from Canada
interpolated between 2015 and 2023, with transport
fraction and snow/ice adjustments based on 2016
meteorological data. Annual and province resolution.
Other point sources not from
the NEI (othpt)
Non-US
2017 Canada point source emissions interpolated between
2015 and 2023, and Mexico point source emissions for
2016 (provided by SEMARNAT). Annual and monthly
resolution.
Point fires in Mexico and
Canada (ptfire othnct)
Non-US
Point source day-specific wildfires and agricultural fires
outside of the U.S. for 2017 from vl.5 of the Fire
INventory (FINN) from National Center for Atmospheric
Research (NCAR, 2017 and Wiedinmyer, C., 2011) for
Canada, Mexico, Caribbean, Central American, and other
international fires.
17
-------
Table 3-2. 2017 Continental United States Emissions by Sector (tons/yr in 48 states + D.C.)
Sector
CO
NH3
NOx
PMio
PM25
SO2
voc
afdust_adj
6,313,419
880,170
ag
3,457,964
227,853
cmv_clc2
24,150
0
166,625
4,569
4,428
648
6,596
cmv_c3
14,721
0
116,250
2,325
2,139
4,780
9,100
nonpt
1,928,006
102,968
741,729
573,265
475,768
166,650
3,593,114
np_oilgas
604,340
13
599,138
10,530
10,384
34,438
2,413,140
nonroad
10,478,152
1,890
1,050,224
102,809
97,089
2,105
1,066,205
onroad
19,254,046
99,546
3,459,049
237,003
112,964
25,338
1,786,147
ptagfire
383,474
85,468
15,085
53,381
32,240
6,171
49,072
ptfire
23,817,819
389,528
357,178
2,460,819
2,087,231
186,731
5,555,642
airports
669,550
0
194,479
10,952
9,717
26,122
85,797
ptegu
608,601
22,755
1,151,253
136,070
111,638
1,383,548
31,994
ptnonipm
1,352,011
60,663
871,767
380,127
242,546
584,632
739,442
pt_oilgas
159,713
2,175
332,011
11,762
11,478
38,812
125,336
rail
109,882
342
557,384
16,021
15,515
682
25,704
rwc
2,160,536
16,409
34,095
300,141
299,280
7,989
323,683
Continental U.S.
61,565,002 4,239,722
9,646,266
10,613,193
4,392,588
2,468,645
16,038,825
Table 3-3. 2016 Non-US Emissions by Sector within the 12US2 Modeling Domain (tons/yr for
Canada, Mexico, Offshore)
Sector
CO
NHj
NOx
PMio
PM25
SO2
VOC
Canada othafdust*
2,310,150
403,705
Canada othptdust*
212,444
82,776
Canada othar
2,087,684
3,560
250,144
229,107
187,530
13,657
631,367
Canada onroad_can
1,140,575
5,035
244,637
17,925
9,065
1,054
91,216
Canada othpt
520,623
378,366
232,062
61,028
25,367
480,317
358,163
Canada ptfire_othna
760,326
13,076
33,162
103,810
83,986
6,273
271,715
Canada CMV
5,958
24
50,552
979
912
1,614
2,788
Canada Subtotal
4,515,165
400,061
810,557
2,935,443
793,340
502,916
1,355,249
Mexico othar
241,571
201,994
220,491
115,460
54,294
7,717
522,236
Mexico onroad_mex
1,828,101
2,789
442,410
15,151
10,836
6,247
158,812
Mexico othpt
205,083
5,049
447,675
73,256
57,440
476,079
71,031
Mexico ptfire_othna
384,764
7,466
16,665
45,198
38,354
2,798
131,980
Mexico Subtotal
2,659,519
217,300
1,127,242
249,066
160,923
492,841
884,059
Offshore pt_oilgas
51,872
8
49,962
636
635
462
38,833
Offshore CMV to EEZ 30,811
103
268,495
5,772
5,354
18,472
15,104
Non-ECA CMV
12,967
245
140,943
13,845
12,742
99,838
6,283
2016 Total Non-U.S.
6,777,842
506,286
1,904,482
3,152,521
925,922
959,392
2,018,302
* The Canada othafdust and othptdust emissions shown are before meteorological and transport fraction reductions.
18
-------
3.2.1 Point Sources (ptegu, ptoilgas ptnonipm, and airports)
Point sources are sources of emissions for which specific geographic coordinates (e.g., latitude/longitude)
are specified, as in the case of an individual facility. A facility may have multiple emission release points
that may be characterized as units such as boilers, reactors, spray booths, kilns, etc. A unit may have
multiple processes (e.g., a boiler that sometimes burns residual oil and sometimes burns natural gas).
With a couple of minor exceptions, this section describes only NEI point sources within the contiguous
U.S. The offshore oil platform (pt oilgas sector) and CMV emissions (cmv c 1 c2 and cmv_c3 sectors)
are processed by SMOKE as point source inventories and are discussed later in this section. A complete
NEI is developed every three years, with 2017 being the most recently finished complete NEI. A
comprehensive description about the development of the 2017 NEI is available in the 2017 NEI TSD
(EPA, 2020a). Point inventories are also available in EIS for intermediate years such as 2018. In the
intermediate point inventories, states are required to update larger sources with emissions for the interim
year, while sources not updated by states are either carried forward from the most recent NEI or marked
as closed and removed.
In preparation for modeling, the complete set of point sources in the NEI was exported from EIS for the
year 2017 into the Flat File 2010 (FF10) format that is compatible with SMOKE (see
https://www.cmascenter.Org/smoke/documentation/4.7/html/ch08s02s08.htmn and was then split into
several sectors for modeling. After dropping sources without specific locations (i.e., the FIPS code ends in
777), initial versions of inventories for the other three point source sectors were created from the
remaining point sources. The point sectors are: EGUs (ptegu), point source oil and gas extraction-related
sources (pt oilgas), airport emissions (airports), and the remaining non-EGUs (ptnonipm). The EGU
emissions are split out from the other sources to facilitate the use of distinct SMOKE temporal processing
and future-year projection techniques. The oil and gas sector emissions (pt oilgas) and airport emissions
(airports) were processed separately for summary tracking purposes and distinct projection techniques
from the remaining non-EGU emissions (ptnonipm).
The inventory pollutants processed through SMOKE for the ptegu, pt oilgas, ptnonipm, and airports
sectors were: CO, NOx, VOC, SO:, NBb, PMio, and PM2.5 and the following HAPs: HQ (pollutant code
= 7647010), and CI (code = 7782505). NBAFM pollutants from the point sectors were not utilized
because VOC was speciated without the use (i.e., integration) of VOC HAP pollutants from the inventory.
The ptnonipm, pt oilgas, and airports sector emissions were provided to SMOKE as annual emissions.
For sources in the ptegu sector that could be matched to 2017 CEMS data, hourly CEMS NOx and SO2
emissions for 2017 from EPA's Acid Rain Program were used rather than annual inventory emissions.
For all other pollutants (e.g., VOC, PM2.5, HQ), annual emissions were used as-is from the annual
inventory, but were allocated to hourly values using heat input from the CEMS data. For the unmatched
units in the ptegu sector, annual emissions were allocated to daily values using IPM region- and pollutant-
specific profiles, and similarly, region- and pollutant-specific diurnal profiles were applied to create
hourly emissions.
The non-EGU stationary point source (ptnonipm) emissions were input to SMOKE as annual emissions.
The full description of how the NEI emissions were developed is provided in the NEI documentation, but
a brief summary of their development follows:
a. CAP and HAP data were provided by States, locals and tribes under the Air Emissions Reporting
Rule (AERR) [the reporting size threshold is larger for inventory years between the triennial inventory
19
-------
years of 2011,2014, 2017, ...].
b. EPA corrected known issues and filled PM data gaps.
c. EPA added HAP data from the Toxic Release Inventory (TRI) where corresponding data was not
already provided by states/locals.
d. EPA stored and applied matches of the point source units to units with CEMS data and also for all
EGU units modeled by EPA's Integrated Planning Model (IPM).
e. Data for airports and rail yards were incorporated.
f. Off-shore platform data were added from the Bureau of Ocean Energy Management (BOEM).
The changes made to the NEI point sources prior to modeling with SMOKE are as follows:
The tribal data, which do not use state/county Federal Information Processing Standards (FIPS)
codes in the NEI, but rather use the tribal code, were assigned a state/county FIPS code of 88XXX,
where XXX is the 3-digit tribal code in the NEI. This change was made because SMOKE requires
all sources to have a state/county FIPS code.
Sources that did not have specific counties assigned (i.e., the county code ends in 777) were not
included in the modeling because it was only possible to know the state in which the sources
resided, but no more specific details related to the location of the sources were available.
Each of the point sectors is processed separately through SMOKE as described in the following
subsections.
3.2.1.1 EGU sector (ptegu)
The ptegu sector contains emissions from EGUs in the 2017 point source inventory that could be matched
to units found in the National Electric Energy Database System (NEEDS) v6 that is used by the Integrated
Planning Model (IPM) to develop future year EGU emissions. It was necessary to put these EGlJs into a
separate sector in the platform because EGUs use different temporal profiles than other sources in the
point sector and it is useful to segregate these emissions from the rest of the point sources to facilitate
summaries of the data. Sources not matched to units found in NEEDS are placed into the ptoilgas or
ptnonipm sectors. For studies with future year cases, the sources in the ptegu sector are fully replaced
with the emissions output from IPM. It is therefore important that the matching between the NEI and
NEEDS database be as complete as possible because there can be double-counting of emissions in future
year modeling scenarios if emissions for units are projected by IPM are not properly matched to the units
in the point source inventory.
The ptegu emissions inventory is a subset of the point source flat file exported from the Emissions
Inventory System (EIS). The 2017 point source emissions were selected from the 2017 NEIFinal dataset
on April 12, 2020 which included submissions from states up through that time. In the point source flat
file, emission records for sources that have been matched to the NEEDS database have a value filled into
the IPM YN column based on the matches stored within EIS. Thus, unit-level emissions were split into a
separate EGU flat file for units that have a populated (non-null) ipm_yn field. A populated ipm_yn field
indicates that a match was found for the EIS unit in the NEEDS v6 database. Updates were made to the
20
-------
flat file output from EIS as described in the list below:
ORIS facility and unit identifiers were updated based on additional matches in a cross-platform
spreadsheet, based on state comments, and using the EIS alternate identifiers table as described
later in this section.
Some units in the ptegu sector are matched to Continuous Emissions Monitoring System (CEMS) data via
Office of Regulatory Information System (ORIS) facility codes and boiler IDs. For the matched units, the
annual emissions of NOx and SO2 in the flat file are replaced with the hourly CEMS emissions in base
year modeling. For other pollutants at matched units, the hourly CEMS heat input data are used to
allocate the NEI annual emissions to hourly values. All stack parameters, stack locations, and Source
Classification Codes (SCC) for these sources come from the flat file. If CEMS data exists for a unit, but
the unit is not matched to the NEI, the CEMS data for that unit are not used in the modeling platform.
However, if the source exists in the NEI and is not matched to a CEMS unit, the emissions from that
source are still modeled using the annual emission value in the NEI temporally allocated to hourly values.
EIS stores many matches from NEI units to the ORIS facility codes and boiler IDs used to reference the
CEMS data. In the flat file, emission records for point sources matched to CEMS data have values filled
into the ORISFAdLITYCODE and ORIS BOILER ID columns. The CEMS data are available at
http://ampd.epa.gov/ampd near the bottom of the "Prepackaged Data" tab. Many smaller emitters in the
CEMS program cannot be matched to the NEI due to inconsistencies in the way a unit is defined between
the NEI and CEMS datasets, or due to uncertainties in source identification such as inconsistent plant
names in the two data systems. In addition, the NEEDS database of units modeled by IPM includes many
smaller emitting EGUs that do not have CEMS. Therefore, there will be more units in the ptegu sector
than have CEMS data.
For the 2016 platform, matches from the NEI to ORIS codes and the NEEDS database were improved.
The matching improvements were carried forward to the 2017 platform where applicable. In some cases,
NEI units in EIS match to many CAMD units. In these cases, a new entry was made in the flat file with a
"_M_" in the ipm_yn field of the flat file to indicate that there are "multiple" ORIS IDs that match that
unit. This helps facilitate appropriate temporal allocation of the emissions by SMOKE. Temporal
allocation for EGUs is discussed in more detail in the Ancillary Data section below. A cross reference
between NEEDS, NEI, and ERTAC is available here:
ftp://newftp.epa.gov/Air/emismod/2016/vl/reports/EGU/egu 2016 2023 2030 NEEDS NEI ERTAC x
ref 13iun2019.xlsx.
The EGU flat file was split into two flat files: those that have unit-level matches to CEMS data using the
oris facility code and oris boiler id fields and those that do not so that different temporal profiles could
be applied. In addition, the hourly CEMS data were processed through v2.1 of the CEMCorrect tool to
mitigate the impact of unmeasured values in the data.
3.2.1.2 Point Oil and Gas Sector (ptoilgas)
The pt oilgas sector was separated from the ptnonipm sector by selecting sources with specific North
American Industry Classification System (NAICS) codes shown in Table 3-4. The emissions and other
source characteristics in the pt oilgas sector are submitted by states, while EPA developed a dataset of
21
-------
nonpoint oil and gas emissions for each county in the U.S. with oil and gas activity that was available for
states to use. Nonpoint oil and gas emissions can be found in the np oilgas sector. More information on
the development of the 2017 oil and gas emissions can be found in Section 4.17 of the 2017 NEITSD.
The ptoilgas sector includes emissions from offshore oil platforms.
Table 3-4. Point source oil and gas sector NAICS Codes
NAICS
NAICS description
2111
Oil and Gas Extraction
211111
Crude Petroleum and Natural Gas Extraction
211112
Natural Gas Liquid Extraction
21112
Crude Petroleum Extraction
211120
Crude Petroleum Extraction
21113
Natural Gas Extraction
211130
Natural Gas Extraction
213111
Drilling Oil and Gas Wells
213112
Support Activities for Oil and Gas Operations
2212
Natural Gas Distribution
22121
Natural Gas Distribution
221210
Natural Gas Distribution
237120
Oil and Gas Pipeline and Related Structures Construction
4861
Pipeline Transportation of Crude Oil
48611
Pipeline Transportation of Crude Oil
486110
Pipeline Transportation of Crude Oil
4862
Pipeline Transportation of Natural Gas
48621
Pipeline Transportation of Natural Gas
486210
Pipeline Transportation of Natural Gas
3.2.1.3 A irports Sector (airports)
Emissions at airports were separated from other sources in the point inventory based on sources that have
the facility source type of 100 (airports). The airports sector includes all aircraft types used for public,
private, and military purposes and aircraft ground support equipment. The Federal Aviation
Administration's (FAA) Aviation Environmental Design Tool (AEDT) is used to estimate emissions for
this sector. For 2017, Texas and California submitted aircraft emissions. Additional information about
aircraft emission estimates can be found in section 3.2.2 of the 2017 NEI TSD.
3.2.1.4 Non-IPM Sector (ptnonipm)
With some exceptions, the non-IPM (ptnonipm) sector contains the point sources that are not in the ptegu,
pt oilgas, or airports sectors. For the most part, the ptnonipm sector reflects the non-EGU sources of the
2017 NEI point inventory; however, it is likely that some low-emitting EGUs not matched to the NEEDS
database or to CEMS data may be found in the ptnonipm sector.
22
-------
The ptnonipm sector contains a small amount of fugitive dust PM emissions from vehicular traffic on
paved or unpaved roads at industrial facilities, coal handling at coal mines, and grain elevators. Sources
with state/county FIPS code ending with "777" are in the NEI but are not included in any modeling
sectors. These sources typically represent mobile (temporary) asphalt plants that are only reported for
some states, and are generally in a fixed location for only a part of the year and are therefore difficult to
allocate to specific places and days as is needed for modeling. Therefore, these sources are dropped from
the point-based sectors in the modeling platform.
The ptnonipm sources (i.e., not EGUs and non -oil and gas sources) were used as-is from the 2017 NEI
point inventory. Emissions from rail yards are included in the ptnonipm sector.
3.2.2 Day-Specific Point Source Fires (ptfire)
Wildfire and prescribed burning emissions are contained in the ptfire sector. The ptfire sector has emissions
provided at geographic coordinates (point locations) and has daily emissions values. The ptfire sector excludes
agricultural burning and other open burning sources that are included in the ptagfire sector. Emissions are day-
specific and include satellite-derived latitude/longitude of the fire's origin and other parameters associated with the
emissions such as acres burned and fuel load, which allow estimation of plume rise.
Figure 3-1 shows the processing stream for wildfire and prescribed burn sources. The emissions estimate
methodology consists of two tools or systems. The first system is called Satellite Mapping Automated
Reanalysis Tool for Fire Incident Reconciliation version 2 (SMARTFIRE2). SMARTFIRE2 is an
algorithm and database system that operate within a geographic information system (GIS) framework.
SMARTFIRE combines multiple sources of fire information and reconciles them into a unified GIS
database. It reconciles fire data from space-borne sensors and ground-based reports, thus drawing on the
strengths of both data types while avoiding double-counting. At its core, SMARTFIRE2 is an association
engine that links reports covering the same fire in any number of multiple databases. In this process, all
input information is preserved, and no attempt is made to reconcile conflicting or potentially contradictory
information (for example, the existence of a fire in one database but not another). In this study, the
national and S/L/T fire information is input into SMARTFIRE2 and then all information is merged and
associated together based on user-defined weights for each fire information dataset. The output from
SMARTFIRE2 is daily acres burned and latitude-longitude coordinates for each fire.
23
-------
Input Data Sets
(state/local/tribal and national data sets)
Fuel Moisture and
Fuel Loading Data
Smoke Modeling (BlueSky Framework)
Daily smoke emissions
for each fire
Emissions Post-Processing
Final Wildland Fire Emissions Inventory
Figure 3-1. Processing flow for fire emission estimates
Inputs to SMARTFIRE for 2017 include (see Section 7 of the 2017 NEI TSD for more details):
The National Oceanic and Atmospheric Administration's (NOAA's) Hazard Mapping System
(HMS) fire location information
GeoMAC (Geospatial Multi-Agency Coordination), an online wildfire mapping application
designed for fire managers to access maps of current fire locations and perimeters in the
United States
The Incident Status Summary, also known as the "ICS-209", used for reporting specific
information on fire incidents of significance
Incident reports including dates of fire activity, acres burned, and fire locations from the
National Association of State Foresters (NASF)
Hazardous fuel treatment reduction polygons for prescribed burns from the Forest Service
Activity Tracking System (FACTS)
Fire activity on federal lands from the United States Fish and Wildlife Service (USFWS)
Wildfire and prescribed date, location, and locations from S/L/T activity submitters
The second system used to estimate emissions is the BlueSky Modeling Framework version 3.5 (revision
#38169). The framework supports the calculation of fuel loading and consumption, and emissions using
* O &
Data Preparation
Data Aggregation and Reconciliation
(SmartFire2)
Daily fire locations
with fire size and type
24
-------
various models depending on the available inputs as well as the desired results. The contiguous United
States and Alaska, where Fuel Characteristic Classification System (FCCS) fuel loading data are
available, were processed using the modeling chain described in Figure 3-2. The Fire Emissions
Production Simulator (FEPS) in the Bluesky Framework generates all the CAP emission factors for
wildland fires used in the 2017 inventory. HAP emission factors were obtained from Urbanski's(2014)
work and applied by region and by fire type.
Bluesky Framework v3.5.0
SrnartFire 2
Dates
Emissions
Location
Size
Type
Fuels
(FCCS v3;
LF v1.4)
Emission
Factors
(FEPS v2)
Consumption
(Consume v4)
Figure 3-2. Blue Sky Modeling Framework
The FCCSv3 cross-reference was implemented along with the LANDFIREvl.4 (at 200 meter resolution)
to provide better fuel bed information for the BlueSky Framework (BSF). The LANDFIREvl .4 was
aggregated from the native resolution and projection to 200 meter using a nearest-neighbor methodology.
Aggregation and reprojection was required for the proper function on BSF.
The final products from this process are annual and daily FFlO-formatted emissions inventories. These
SMOKE-ready inventory files contain both CAPs and HAPs. The BAFM HAP emissions from the
inventory were used directly in modeling and were not overwritten with VOC speciation profiles (i.e., an
"integrate HAP" use case).
3.2.3 Nonpoint Sources (afdust, ag, nonpt, npoilgas, rwc)
Several modeling platform sectors were created from the 2017 NEI nonpoint inventory. This section
describes the stationary nonpoint sources. Locomotives, CI and C2 CMV, and C3 CMV are also
included the NEI nonpoint data category, but are mobile sources and are described in a later section. The
2017 NEI TSD includes documentation for the nonpoint data.
The nonpoint tribal-submitted emissions are dropped during spatial processing with SMOKE due to the
configuration of the spatial surrogates, which are available by county, but not at the tribal level. In
addition, possible double-counting with county-level emissions is prevented. These omissions are not
expected to have an impact on the results of the air quality modeling at the 12-km scales used for this
platform.
25
-------
In the rest of this section, each of the platform sectors into which the sources in the nonpoint NEI data
category were divided is described, along with any data that were updated or replaced with non-NEI data.
3.2.3.1 Area Fugitive Dust Sector (afdust)
The area-source fugitive dust (afdust) sector contains PMio and PM2.5 emission estimates for nonpoint
SCCs identified by EPA staff as dust sources. Categories included in the afdust sector are paved roads,
unpaved roads and airstrips, construction (residential, industrial, road and total), agriculture production,
and mining and quarrying. It does not include fugitive dust from grain elevators, coal handling at coal
mines, or vehicular traffic on paved or unpaved roads at industrial facilities because these are treated as
point sources so they are properly located.
The afdust sector is separated from other nonpoint sectors to allow for the application of a "transport
fraction," and meteorological/precipitation reductions based on the meteorological data for the year being
modeled. These adjustments are applied with a script that applies land use-based gridded transport
fractions followed by another script that zeroes out emissions for days on which at least 0.01 inches of
precipitation occurs or there is snow cover on the ground. The land use data used to reduce the NEI
emissions determines the amount of emissions that are subject to transport. This methodology is
discussed in (Pouliot, et al., 2010),
https://www3.epa.gov/ttn/chief/conference/eil9/session9/pouliot pres.pdf. and in "Fugitive Dust
Modeling for the 2008 Emissions Modeling Platform" (Adelman, 2012). Both the transport fraction and
meteorological adjustments are based on the gridded resolution of the platform (e.g., 12km grid cells);
therefore, different emissions will result if the process were applied to different grid resolutions. A
limitation of the transport fraction approach is the lack of monthly variability that would be expected with
seasonal changes in vegetative cover. While wind speed and direction are not accounted for in the
emissions processing, the hourly variability due to soil moisture, snow cover and precipitation are
accounted for in the subsequent meteorological adjustment.
For the data compiled into the NEI, meteorological adjustments are applied to paved and unpaved road
SCCs but not transport adjustments. This is because the modeling platform applies meteorological
adjustments and transport adjustments based on unadjusted NEI values. For the 2017 NEI, the
meteorological adjustments that were applied (to paved and unpaved road SCCs) were backed out in order
reapply them in SMOKE. The FF10 that is run through SMOKE consists of 100% unadjusted emissions,
and after SMOKE all afdust sources have both transport and meteorological adjustments applied
according to year 2017 meteorology.
For categories other than paved and unpaved roads, where states submitted afdust data, it was assumed
that the state-submitted data were not met-adjusted and therefore the meteorological adjustments were
applied. Thus, if states submitted data that were met-adjusted for sources other than paved and unpaved
roads, these sources would have been adjusted for meteorology twice. Even with that possibility, air
quality modeling shows that, in general, dust is frequently overestimated in the air quality modeling
results.
26
-------
3.2.3.2
Agricultural Ammonia Sector (ag)
The agricultural (ag) sector includes livestock and fertilizer application emissions. Livestock emissions
consist of emissions from the 2017 NEI nonpoint inventory. Fertilizer emissions for 2017 are based on
the FEST-C model (https://www.cmascenter.org/fest-c/). The livestock and fertilizer emissions in this
sector are based only on the SCCs starting with 2805 and 2801. The livestock SCCs are related to beef
and dairy cattle, poultry production and waste, swine production, waste from horses and ponies, and
production and waste for sheep, lambs, and goats. The "ag" sector includes all of the NH3 emissions from
fertilizer. However, the "ag" sector does not include all of the livestock NFb emissions, as there is a very
small amount of NFb emissions from livestock in the ptnonipm inventory (as point sources). In addition
to NFb, the "ag" sector also includes livestock emissions from all pollutants other than PM2.5. PM2.5 from
livestock are in the afdust sector.
Agricultural livestock emissions in the platform are based on the 2017 NEI, which is a mix of state-
submitted data and EPA estimates. Livestock emissions utilized improved animal population data. VOC
livestock emissions, new for this sector, were estimated by multiplying a national VOC/NH3 emissions
ratio by the county NFb emissions. The 2017 NEI approach for livestock utilizes daily emission factors
by animal and county from a model developed by Carnegie Mellon University (CMU) (Pinder, 2004,
McQuilling, 2015) and 2017 U.S. Department of Agriculture (USDA) National Agricultural Statistics
Service (NASS) survey. Details on the approach are provided in Section 4.5 of the 2017 NEI TSD.
As described in the 2017 NEI TSD, fertilizer emissions for 2017 are based on the FEST-C model
(https://www.cmascenter.org/fest-c/). The bidirectional version of CMAQ (v5.3) and the Fertilizer
Emissions Scenario Tool for CMAQ FEST-C (vl.3) were used to estimate ammonia (NFb) emissions
from agricultural soils. The approach to estimate year-specific fertilizer emissions consists of these steps:
Run FEST-C and CMAQ model with bidirectional ("bidi") NFb exchange to produce nitrate
(NO3), Ammonium (NH4+, including Urea), and organic (manure) nitrogen (N) fertilizer usage
estimates, and gaseous ammonia NFb emission estimates respectively.
Calculate county-level emission factors as the ratio of bidirectional CMAQ NFb fertilizer
emissions to FEST-C total N fertilizer application.
Assign the NFb emissions to one SCC: ".. .Miscellaneous Fertilizers" (2801700099).
For livestock and fertilizer, meteorological-based temporalization (described in Section 3.3.5.3) is used
for month-to-day and day-to-hour temporalization. Monthly profiles for livestock are based on the daily
data underlying the EPA estimates from 2014NEIv2. The fertilizer inventory includes monthly emissions
from FEST-C, and uses the same meteorological-based month-to-hour profiles as livestock in the same
way as was done for other recent platforms.
3.2.3.3 Agricultural fires (ptagfire)
In the NEI, agricultural fires are stored as county-annual emissions and are part of the nonpoint data
category. For this study agricultural fires are modeled as day specific fires derived from satellite data for
the year 2017 NEI in a similar way to the emissions in ptfire. State-provided agricultural fire data from
the 2014NEIv2 are not used in this study.
Daily year-specific agricultural burning emissions are derived from HMS fire activity data, which
contains the date and location of remote-sensed anomalies. The activity is filtered using the 2017 USDA
27
-------
cropland data layer (CDL). Satellite fire detects over agricultural lands are assumed to be agricultural
burns and assigned a crop type. Detects that are not over agricultural lands are output to a separate file for
use in the ptfire sector. Each detect is assigned an average size of between 40 and 80 acres based on crop
type. The assumed field sizes can be found at http://www.epa.gov/sites/production/files/2015-
06/draft 2014 ag grasspasture emissions nei mav62015.xlsx. Grassland/pasture fires were moved to
the ptfire sector for this 2016 modeling platform.
Another feature of the database is that the satellite detections for 2017 were filtered out to exclude areas
covered by snow during the winter months. To do this, the daily snow cover fraction per grid cell was
extracted from a 2017 meteorological simulation (WRF). The location of fire detections was then
compared with this daily snow cover file. For any day in which a grid cell has snow cover, that fire
detection was excluded. Due to the inconsistent reporting of fire detections from the Visible Infrared
Imaging Radiometer Suite (VIIRS) platform, any fire detections in the HMS dataset that were flagged as
VIIRS or SUOMI were excluded. In addition, certain crop types (corn and soybeans) have been excluded
from these specific midwestern states: Iowa, Kansas, Indiana, Illinois, Michigan, Missouri, Minnesota,
Wisconsin, and Ohio.
Emissions factors were applied to each daily fire to calculate criteria and hazardous pollutant values.
These factors vary by crop type. In all prior NEIs for this sector, the HAP emission factors and the VOC
emission factors were known to be inconsistent. Corrections have been made to the application of the
HAP emissions factors for the 2017 NEI. Please see section 7.4.4 in the 2017 NEI TSD for the details of
the corrections.
Heat flux for plume rise was calculated using the size and assumed fuel loading of each daily agricultural
fire. This information is needed for a plume rise calculation within a chemical transport modeling system.
The daily agricultural and open burning emissions were converted from a tabular format into the
SMOKE-ready daily point flat file format. The daily emissions were also aggregated into annual values
by location and converted into the annual point flat file format.
The state of Georgia provided their own estimates of agricultural crop residue burning and completely
replaced the emission estimates by the EPA. The HMS information was replaced with the state-supplied
activity data and the emissions were recomputed for this state. See section 4.12 in 2017 NEI TSD for
more details on agricultural crop residue burning.
For this modeling platform, a SMOKE update allows the use of HAP integration for speciation for
PTDAY inventories. The 2017 agricultural fire inventories include emissions for HAPs, so HAP
integration was used for this study.
3.2.3.4 Nonpoint Oil-gas Sector (npoilgas)
The nonpoint oil and gas (np oilgas) sector contains onshore and offshore oil and gas emissions. The
EPA estimated emissions for all counties with 2017 oil and gas activity data with the Oil and Gas Tool,
and many S/L/T agencies also submitted nonpoint oil and gas data. Where S/L/T submitted nonpoint
CAPs but no HAPs, the EPA augmented the HAPs using HAP augmentation factors (county and SCC
level) created from the Oil and Gas Tool. The types of sources covered include drill rigs, workover rigs,
artificial lift, hydraulic fracturing engines, pneumatic pumps and other devices, storage tanks, flares, truck
loading, compressor engines, and dehydrators.
28
-------
EPA has updated the Nonpoint Oil and Gas Emission Estimation Tool (i.e., the "tool") to estimate
emissions for 2017. Year 2017 oil and gas activity data was supplied to EPA by state air agencies and
where state data is not supplied to EPA, EPA populates the 2017 inventory with the best available data.
The tool is an Access database that utilizes county-level activity data (e.g., oil production and well
counts), operational characteristics (types and sizes of equipment), and emission factors to estimate
emissions. The tool creates a CSV-formatted emissions dataset covering all national nonpoint oil and
gas emissions. This dataset is then converted to FF10 format for use in SMOKE modeling. More details
on the inputs for and running of the tool are provided in the 2017 NEI TSD.
3.2.3.5 Residential Wood Combustion Sector (rwc)
The residential wood combustion (rwc) sector includes residential wood burning devices such as
fireplaces, fireplaces with inserts (inserts), free standing woodstoves, pellet stoves, outdoor hydronic
heaters (also known as outdoor wood boilers), indoor furnaces, and outdoor burning in firepots and
chimneas. Free standing woodstoves and inserts are further differentiated into three categories:
1) conventional (not EPA certified); 2) EPA certified, catalytic; and 3) EPA certified, noncatalytic.
Generally speaking, the conventional units were constructed prior to 1988. Units constructed after 1988
have to meet EPA emission standards and they are either catalytic or non-catalytic. As with the other
nonpoint categories, a mix of S/L and EPA estimates were used. The EPA's estimates use updated
methodologies for activity data and some changes to emission factors.
Some improvements to RWC emissions estimates were made for the 2017 NEI and were included in this
study. The EPA, along with the Commission on Environmental Cooperation (CEC), the Northeast States
for Coordinated Air Use Management (NESCAUM), and Abt Associates, conducted a national survey of
wood-burning activity in 2018. The results of this survey were used to estimate county-level burning
activity data. The activity data for RWC processes is the amount of wood burned in each county, which is
based on data from the CEC survey on the fraction of homes in each county that use each wood-burning
appliance and the average amount of wood burned in each appliance. These assumptions are used with the
number of occupied homes in each county to estimate the total amount of wood burned in each county, in
cords for cordwood appliances and tons for pellet appliances. Cords of wood are converted to tons using
county-level density factors from the U.S. Forest Service. RWC missions were calculated by multiplying
the tons of wood burned by emissions factors. For more information on the development of the residential
wood combustion emissions, see Section 4.15 of the 2017 NEI TSD.
3.2.3.6 Other Nonpoint Sources (nonpt)
Stationary nonpoint sources that were not subdivided into the afdust, ag, np oilgas, or rwc sectors were
assigned to the "nonpt" sector. Locomotives and CMV mobile sources from the 2017 NEI nonpoint
inventory are described with the mobile sources. The types of sources in the nonpt sector include:
stationary source fuel combustion, including industrial, commercial, and residential and orchard
heaters;
chemical manufacturing;
industrial processes such as commercial cooking, metal production, mineral processes, petroleum
refining, wood products, fabricated metals, and refrigeration;
29
-------
solvent utilization for surface coatings such as architectural coatings, auto refinishing, traffic
marking, textile production, furniture finishing, and coating of paper, plastic, metal, appliances,
and motor vehicles;
solvent utilization for degreasing of furniture, metals, auto repair, electronics, and manufacturing;
solvent utilization for dry cleaning, graphic arts, plastics, industrial processes, personal care
products, household products, adhesives and sealants;
solvent utilization for asphalt application and roofing, and pesticide application;
storage and transport of petroleum for uses such as portable gas cans, bulk terminals, gasoline
service stations, aviation, and marine vessels;
storage and transport of chemicals;
waste disposal, treatment, and recovery via incineration, open burning, landfills, and composting;
and
miscellaneous area sources such as cremation, hospitals, lamp breakage, and automotive repair
shops.
The nonpt sector includes emission estimates for Portable Fuel Containers (PFCs), also known as "gas
cans." The PFC inventory consists of five distinct sources of PFC emissions, further distinguished by
residential or commercial use. The five sources are: (1) displacement of the vapor within the can; (2)
spillage of gasoline while filling the can; (3) spillage of gasoline during transport; (4) emissions due to
evaporation (i.e., diurnal emissions); and (5) emissions due to permeation. Note that spillage and vapor
displacement associated with using PFCs to refuel nonroad equipment are included in the nonroad
inventory.
3.2.4 Biogenic Sources (beis)
Biogenic emissions were computed based on the 17j version of the 2017 meteorology data used for the air
quality modeling and were developed using the Biogenic Emission Inventory System version 3.7
(BEIS3.7) within CMAQ. The BEIS3.7 creates gridded, hourly, model-species emissions from vegetation
and soils. It estimates CO, VOC (most notably isoprene, terpene, and sesquiterpene), and NO emissions
for the contiguous U.S. and for portions of Mexico and Canada. In the BEIS 3.7 two-layer canopy model,
the layer structure varies with light intensity and solar zenith angle (Pouliot and Bash, 2015). Both layers
include estimates of sunlit and shaded leaf area based on solar zenith angle and light intensity, direct and
diffuse solar radiation, and leaf temperature (Bash et al., 2015). The new algorithm requires additional
meteorological variables over previous versions of BEIS. The variables output from the Meteorology-
Chemistry Interface Processor (MCIP) that are used to convert WRF outputs to CMAQ inputs are shown
in Table 3-5.
Table 3-5. Meteorological variables required by BEIS 3.7
Variable
Description
LAI
leaf-area index
PRSFC
surface pressure
Q2
mixing ratio at 2 m
RC
convective precipitation per met TSTEP
RGRND
solar rad reaching sfc
RN
nonconvective precipitation per met TSTEP
30
-------
Variable
Description
RSTOMI
inverse of bulk stomatal resistance
SLYTP
soil texture type by USD A category
SOIM1
volumetric soil moisture in top cm
SOIT1
soil temperature in top cm
TEMPG
skin temperature at ground
USTAR
cell averaged friction velocity
RADYNI
inverse of aerodynamic resistance
TEMP2
temperature at 2 m
BEIS3.7 was used in conjunction with Version 5 of the Biogenic Emissions Landuse Database (BELD5).
The BELD5 is based on an updated version of the USDA-USFS Forest Inventory and Analysis (FIA)
vegetation speciation-based data from 2001 to 2017 from the FIA version 8.0. Canopy coverage is based
on the Global Moderate Resolution Imaging Spectroradiometer (MODIS) 20 category data with enhanced
lakes and Fraction of Photosynthetically Active Radiation (FPAR) for vegetation coverage from National
Center for Atmospheric Research (NCAR). The FIA includes approximately 250,000 representative plots
of species fraction data that are within approximately 75 km of one another in areas identified as forest by
the MODIS canopy coverage. For land areas outside the conterminous United States, 500 meter grid
spacing land cover data from the Moderate Resolution Imaging Spectroradiometer (MODIS) is used.
BELDv5 also incorporates the following:
Canadian BELD land use, Updates to Version 4 of the Biogenic Emissions Landuse Database
(BELD4) for Canada and Impacts on Biogenic VOC Emissions
(https://www.epa.gov/sites/production/files/2019-08/documents/8Q0am zhang 2 O.pdf)
2017 30 meter USD A Cropland Data Layer (CDL) data
(http://www.nass.usda.gov/research/Cropland/Release/).
Biogenic emissions computed with BEIS were left out of the CMAQ-ready merged emissions, in favor of
inline biogenics produced during the CMAQ model run itself using the same algorithm described above
but with finer time steps within the air quality model.
3.2.5 Mobile Sources (onroad, ottroadcaadj, nottroad, cmv_clc2, cmv_c3, rail)
Mobile sources are emissions from vehicles that move and include several sectors. Onroad mobile source
emissions result from motorized vehicles that are normally operated on public roadways. These include
passenger cars, motorcycles, minivans, sport-utility vehicles, light-duty trucks, heavy-duty trucks, and
buses. Nonroad mobile source emissions are from vehicles that do not operate on roads such as tractors,
construction equipment, lawnmowers, and recreational marine vessels. All nonroad emissions are treated
as county-specific low-level emissions (i.e., they are released into model layer 1).
Commercial marine vessel (CMV) emissions are split into two sectors: emissions from Category 1 and
Category 2 vessels are in the cmv c 1 c2 sector, and emissions from the larger ocean-going Category 3
vessels are in the cmv_c3 sector. Both CMV sectors are treated as point sources with plume rise.
Locomotive emissions are in the rail sector. Having the emissions split into these sectors facilitates
separating them in summaries and also allows for CMV to be modeled with plume rise.
31
-------
3.2.5.1 Onroad (onroad)
Onroad mobile sources include emissions from motorized vehicles that normally operate on public
roadways. These include passenger cars, motorcycles, minivans, sport-utility vehicles, light-duty trucks,
heavy-duty trucks, and buses. The sources are further divided between diesel, gasoline, E-85, and
compressed natural gas (CNG) vehicles. The sector characterizes emissions from parked vehicle
processes (e.g., starts, hot soak, and extended idle) as well as from on-network processes (i.e., from
vehicles moving along the roads).
The onroad SCCs in the modeling platform are more finely resolved than those in the NEI, because the
NEI SCCs distinguish vehicles and fuels, but in the platform they also distinguish between emissions on
roadways, off-network, extended idle, and the various MOVES road-types. For more details on the
approach and for a summary of the MOVES inputs submitted by states, see section 6.5.1 of the 2017 NEI
TSD. In addition, a number of states submitted 2017-specific activity data for incorporation into this
platform.
Except for California, onroad emissions are generated using the SMOKE-MOVES interface that leverages
MOVES generated emission factors (https://www.epa.gov/moves). county and SCC-specific activity data,
and hourly meteorological data. SMOKE-MOVES takes into account the temperature sensitivity of the
on-road emissions. Specifically, EPA used MOVES inputs for representative counties, vehicle miles
traveled (VMT), vehicle population (VPOP), and hoteling hours data for all counties, along with tools that
integrated the MOVES model with SMOKE. In this way, it was possible to take advantage of the gridded
hourly temperature data available from meteorological modeling that are also used for air quality
modeling.
SMOKE-MOVES makes use of emission rate "lookup" tables generated by MOVES that differentiate
emissions by process (i.e., running, start, vapor venting, etc.), vehicle type, road type, temperature, speed,
hour of day, etc. To generate the MOVES emission rates that could be applied across the U.S., EPA used
an automated process to run MOVES to produce year 2017-specific emission factors by temperature and
speed for a series of "representative counties," to which every other county was mapped. The
representative counties for which emission factors are generated are selected according to their state,
elevation, fuels, age distribution, ramp fraction, and inspection and maintenance programs. Each county
is then mapped to a representative county based on its similarity to the representative county with respect
to those attributes. For this study, there are 296 representative counties in the continental U.S. A detailed
discussion of the representative counties is in the 2017 NEI TSD, Section 6.8.2.
Once representative counties have been identified, emission factors are generated with MOVES for each
representative county and for two "fuel months" - January to represent winter months, and July to
represent summer months - due to the different types of fuels used. SMOKE selects the appropriate
MOVES emissions rates for each county, hourly temperature, SCC, and speed bin and then multiplies the
emission rate by appropriate activity data. For on-roadway emissions, vehicle miles travelled (VMT) is
the activity data, vehicle population (VPOP) is used for many off-network processes, and hoteling hours
are used to develop emissions for extended idling of combination long-haul trucks. These calculations are
done for every county and grid cell in the continental U.S. for each hour of the year.
The SMOKE-MOVES process for creating the model-ready emissions consists of the following steps:
1) Determine which counties will be used to represent other counties in the MOVES runs.
32
-------
2) Determine which months will be used to represent other month's fuel characteristics.
3) Create inputs needed only by MOVES. MOVES requires county-specific information on
vehicle populations, age distributions, and inspection-maintenance programs for each of the
representative counties.
4) Create inputs needed both by MOVES and by SMOKE, including temperatures and activity
data.
5) Run MOVES to create emission factor tables for the temperatures found in each county.
6) Run SMOKE to apply the emission factors to activity data (VMT, VPOP, and HOTELING) to
calculate emissions based on the gridded hourly temperatures in the meteorological data.
7) Aggregate the results to the county-SCC level for summaries and quality assurance.
The onroad emissions are processed in four processing streams that are merged together into the onroad
sector emissions after each of the four streams have been processed:
rate-per-distance (RPD) uses VMT as the activity data plus speed and speed profile information to
compute on-network emissions from exhaust, evaporative, permeation, refueling, and brake and
tire wear processes;
rate-per-vehicle (RPV) uses VPOP activity data to compute off-network emissions from exhaust,
evaporative, permeation, and refueling processes;
rate-per-profile (RPP) uses VPOP activity data to compute off-network emissions from
evaporative fuel vapor venting, including hot soak (immediately after a trip) and diurnal (vehicle
parked for a long period) emissions; and
rate-per-hour (RPH) uses hoteling hours activity data to compute off-network emissions for idling
of long-haul trucks from extended idling and auxiliary power unit process.
The onroad emissions inputs for the platform are based on the 2017 NEI, described in more detail in
Section 6 of the 2017 NEI TSD. These inputs include:
MOVES County databases (CDBs) including Low Emission Vehicle (LEV) table
Representative counties
Fuel months
Meteorology
Activity data (VMT, VPOP, speed, HOTELING)
Representative counties, fuel months and other inputs were consistent with those in the 2017 NEI.
States that submitted activity data and development of the EPA default activity data sets for VMT, VPOP,
and hoteling hours are described in detail in the 2017 NEI TSD and supporting documents. Hoteling hours
activity are used to calculate emissions from extended idling and auxiliary power units (APUs) by
combination long-haul trucks.
Hoteling hours were capped by county at a theoretical maximum and any excess hours of the maximum
were reduced. For calculating reductions, a dataset of truck stop parking space availability was used,
which includes a total number of parking spaces per county. This same dataset is used to develop the
spatial surrogate for allocating county-total hoteling emissions to model grid cells. The parking space
dataset includes several recent updates based on new truck stops opening and other new information.
There are 8,760 hours in the year 2017; therefore, the maximum number of possible hoteling hours in a
33
-------
particular county is equal to 8,760 * the number of parking spaces in that county. Hoteling hours were
capped at that theoretical maximum value for 2017 in all counties, with some exceptions.
Because the truck stop parking space dataset may be incomplete in some areas, and trucks may sometimes
idle in areas other than designated spaces, it was assumed that every county has at least 12 parking spaces,
even if fewer parking spaces are found in the parking space dataset. Therefore, hoteling hours are never
reduced below 105,408 hours for the year in any county. If the unreduced hoteling hours were already
below that maximum, the hours were left unchanged; in other words, hoteling activity are never increased
as a result of this analysis. Four states requested that no reductions be applied to the hoteling activity
based on parking space availability: CO, ME, NJ, and NY. For these states, reductions based on parking
space availability were not applied.
The final step related to hoteling activity is to split county totals into separate values for extended idling
(SCC 2202620153) and Auxiliary Power Units (APUs) (SCC 2202620191). New Jersey's submittal of
hoteling activity specified a 30% APU split, and this was used throughout NJ. For the rest of the country,
a 12.4% APU split was used, meaning that during 12.4% of the hoteling hours auxiliary power units are
assumed to be running.
The last pieces of activity data needed for SMOKE-MOVES are related to the average speed of vehicles,
which affects the selection of MOVES emission factors for on-network emissions. One such dataset is the
SPEED inventory read by the SMOKE program Smkinven, which includes a single overall average speed
for each county, SCC, and month. The second dataset is the SPDIST dataset read by the SMOKE program
Movesmrg, which specifies the amount of time spent in each MOVES speed bin for each county, vehicle
(aka source) type, road type, weekday/weekend, and hour of day. SMOKE still requires the SPEED
dataset exist even when hourly speed data is available, even though only the hourly speed data affects the
selection of emission factors. The SPEED and SPDIST datasets are from the 2017 NEI with some of the
data carried over from the Coordinating Research Council A-100 study (CRC, 2017).
MOVES2014b was run in emission factor mode to create emission factor tables using CB6 speciation for
the year 2016, for all representative counties and fuel months. The county databases used to run MOVES
to develop the emission factor tables were the same as those in the 2017 NEI, including the state-specific
control measures such as the California LEV program, and fuels represented the year 2017. In addition,
the range of temperatures run along with the average humidities used were specific to the year 2017. The
remaining settings for the CDBs are documented in the 2017 NEI TSD. To create the emission factors,
MOVES was run separately for each representative county and fuel month for each temperature bin
needed for the calendar year 2017. The MOVES results were post-processed into CSV-formatted
emission factor tables that can be read by SMOKE-MOVES.
California is the only state agency for which submitted onroad emissions were used in the 2017 NEI and
this study. California uses their own emission model, EMFAC, which uses emission inventory codes
(EICs) to characterize the emission processes instead of SCCs. The EPA and California worked together
to develop a code mapping to better match EMFAC's EICs to EPA MOVES' detailed set of SCCs that
distinguish between off-network and on-network and brake and tire wear emissions. This detail is needed
for modeling but not for the NEI. California provided their CAP and HAP emissions by county using
EPA SCCs after applying an agreed-to EIC to SCC mapping. California provided EMFAC2017-based
onroad emissions inventories for 2017 that were used for this study.
The California onroad mobile source emissions were created through a hybrid approach of combining
34
-------
state-supplied annual emissions with EPA-developed SMOKE-MOVES runs. Through this approach, the
platform was able to reflect the unique rules in California, while leveraging the more detailed SCCs and
the highly resolved spatial patterns, temporal patterns, and speciation from SMOKE-MOVES. The basic
steps involved in temporally allocating onroad emissions from California based on SMOKE-MOVES
results were:
1) Run CA using EPA inputs through SMOKE-MOVES to produce hourly emissions hereafter
known as "EPA estimates." These EPA estimates for CA are run in a separate sector called
"onroadca."
2) Calculate ratios between state-supplied emissions and EPA estimates. The ratios were calculated
for each county/SCC/pollutant combination based on the California onroad emissions inventory.
Unlike in previous platforms, the California data separated off and on-network emissions and
extended idling. However, the on-network did not provide specific road types, and California's
emissions did not include information for vehicles fueled by E-85, so these differentiations were
obtained using MOVES.
3) Create an adjustment factor file (CFPRO) that includes EPA-to-state estimate ratios.
4) Rerun CA through SMOKE-MOVES using EPA inputs and the new adjustment factor file.
Through this process, adjusted model-ready files were created that sum to annual totals from California,
but have the temporal and spatial patterns reflecting the highly resolved meteorology and SMOKE-
MOVES. After adjusting the emissions, this sector is called "onroadcaadj " Note that in emission
summaries, the emissions from the "onroad" and "onroad ca adj" sectors are summed and designated as
the emissions for the onroad sector.
3.2.5.2 MOVES-based Nonroad Mobile Sources (nottroad)
The nonroad equipment emissions in the platform and the NEI result primarily from running the
MOVES2014b model. MOVES2014b was used for all states other than California, which uses their own
model. California nonroad emissions were provided by the California Air Resources Board (CARB) for
the 2017 NEI.
MOVES2014b creates a monthly emissions inventory for criteria air pollutants (CAPs) and a full set of
HAPs, plus additional pollutants such as NONHAPTOG and ETHANOL, which are not part of the NEI
but are used for speciation. MOVES2014b provides estimates of NONHAPTOG along with the speciation
profile code for the NONHAPTOG emission source. This was accomplished by using NHTOG#### as
the pollutant code in the Flat File 2010 (FF10) inventory file that can be read into SMOKE, where #### is
a speciation profile code. For California, NHTOG####-VOC and HAP-VOC ratios from MOVES-based
emissions were applied to VOC emissions from California so that VOC emissions in California can be
speciated consistently with other states.
MOVES2014b, unlike MOVES2014a, also provides estimates of PM2.5 by speciation profile code for the
PM2.5 emission source, using PM25_#### as the pollutant code in the FF10 inventory file, where #### is a
speciation profile code. To facilitate calculation of PMC within SMOKE, and to help create emissions
summaries, an additional pollutant representing total PM2.5 called PM25TOTAL was added to the
inventory. As with VOC, PM2.5 _####-PM25TOTAL ratios were calculated and applied to PM2.5
emissions in California so that PM2.5 emissions in California can be speciated consistently with other
35
-------
states.
MOVES2014b outputs emissions data in county-specific databases, and then a post-processing script
converted the data into FF10 format. Additional post-processing steps were performed as follows:
County-specific FFlOs were combined into a single FF10 file.
To reduce the size of the inventory, HAPs that are not needed for air quality modeling, such as
dioxins and furans, were removed from the inventory.
To reduce the size of the inventory further, all emissions for sources (identified by county/SCC)
for which total CAP emissions are less than 1*10"10 were removed from the inventory. The
MOVES model attributes a very tiny amount of emissions to sources that are actually zero, for
example, snowmobile emissions in Florida. Removing these sources from the inventory reduces
the total size of the inventory by 7%.
Gas and particulate components of HAPs that come out of MOVES separately, such as
naphthalene, were combined.
VOC was renamed VOC INV so that SMOKE does not speciate both VOC and NONHAPTOG,
which would result in a double count.
PM25TOTAL, referenced above, was also created at this stage of the process.
California emissions from MOVES were deleted, in favor of the CARB data.
Emissions for airport ground support vehicles (SCCs ending in -8005), and oil field equipment
(SCCs ending in -10010), were removed from the inventory at this stage, to prevent a double
count with the airports and np oilgas sectors, respectively.
3.2.5.3 Locomotive (rail)
The rail sector includes all locomotives in the NEI nonpoint data category. This sector excludes railway
maintenance locomotives and point source yard locomotives. Railway maintenance emissions are
included in the nonroad sector. The point source yard locomotives are included in the ptnonipm sector.
Typically in the NEI, yard locomotive emissions are split between the nonpoint and point categories, but
for this study, all yard locomotive emissions are represented as point sources and included in the
ptnonipm sector.
This study uses a new 2017 rail inventory developed for the 2017 NEI by the Lake Michigan Air
Directors Consortium (LADCO) and the State of Illinois with support from various other states. Class I
railroad emissions are based on confidential link-level line-haul activity GIS data layer maintained by the
Federal Railroad Administration (FRA). In addition, the Association of American Railroads (AAR)
provided national emission tier fleet mix information. Class II and III railroad emissions are based on a
comprehensive nationwide GIS database of locations where short line and regional railroads operate.
Passenger rail (Amtrak) emissions follow a similar procedure as Class II and III, except using a database
of Amtrak rail lines. Yard locomotive emissions are based on a combination of yard data provided by
individual rail companies, and by using Google Earth and other tools to identify rail yard locations for rail
companies which did not provide yard data. Information on specific yards were combined with fuel use
data and emission factors to create an emissions inventory for rail yards. More detailed information on
36
-------
the development of the 2017 rail inventory for this study is available in the 2017 NEI TSD.
3.2.5.4 Category 1,2, and 3 commercial marine vessels (cmv_clc2 and cmv_3)
The cmv_clc2 sector contains Category 1 and 2 CMV emissions from the 2017 NEI. Category 1 and 2
vessels use diesel fuel. All emissions in this sector are annual and at county-SCC resolution; however, in
the NEI they are provided at the sub-county level (port or underway shape ids) and by SCC and emission
type (e.g., hoteling, maneuvering). This sub-county data in the NEI are used to create spatial surrogates.
For more information on CMV sources in the 2017 NEI, see Section 4.21 of the 2017 NEI TSD. CI and
C2 emissions that occur outside of state waters are not assigned to states. All CMV emissions in the
cmv_clc2 sector are treated as hourly gridded point sources with stack parameters that should result in
them being placed in layer 1.
The Category 3 CMV vessels in the cmv_c3 sector use residual oil. The emissions in the cmv_c3 sector
are comprised of primarily foreign-flagged ocean-going vessels, referred to as C3 CMV ships. The C3
portion of the CMV inventory includes these ships in several intra-port modes (i.e., cruising, hoteling,
reduced speed zone, maneuvering, and idling) and an underway mode, and includes near-port auxiliary
engine emissions. The cmv_c3 sector uses AIS derived 2017 NEI emissions in a gridded point source
format. Emissions from the Emissions Control Area-International Marine Organization (ECA-IMO)-
based C3 CMV are used for waters not covered by the AIS selection (portions of FIPS code 98001).
The 2017 NEI C3 inventory was developed as an hourly point inventory from 5 minute-resolution
Automated Identification System (AIS) transponder data which had emissions computed from it. Plume
rise calculations for C3 vessels are computed when the sources are modeled by SMOKE and CMAQ. To
develop the point inventory, the AIS emissions were aggregated to hourly values by grid cell, port
identifier, SCC, and FIPS on the 12 km modeling domain. Release point locations were assigned to the
centroid of the corresponding grid cell. Emissions where the source vessel idled in the same location for
longer than an hour (hoteling) were apportioned to the subsequent hours from the initial time based on the
duration of hoteling. Hoteling was capped at 400 hours and emissions were adjusted down proportionally
where a vessel exceeded 400 hours in a single location.
Emissions from vessels that occur within the boundaries of a 2017 NEI port polygon were given a port
SCC and assigned the appropriate port shape identifier. All other emissions were assigned an underway
SCC and not given a port shape identifier. FIPS were assigned to the 5-minute AIS data using a polygon
shapefile of US counties extending into state waters, Canadian provinces extending into provincial waters,
Mexican states, and US federal water economic exclusive zones (EEZ).
The resulting annual point emissions were converted to an annual point 2010 flat file format (FF10). A set
of standard stack parameters were assigned to each release point in the cmv_c3 inventory. The assigned
stack height was 65.62 ft, the stack diameter was 2.625 ft, the stack temperature was 539.6 °F, and the
velocity was 82.02 ft/s. Hourly emissions were converted to an hourly point 2010 FF10.
In 2014, marine vessels in the North American Emission Control Area (ECA), which extends 200 miles
from the shores of North America, Puerto Rico, and the Virgin Islands, met a fuel sulfur standard of
10,000 ppm. On January 1st, 2015, the ECA initiated a fuel sulfur standard which regulated large marine
37
-------
vessels to use fuel with 1,000 ppm sulfur or less. EPA multiplied European Union (EU)7 C3 emissions
factors that include these standards with the E&C calls of the respective years.
The EU emission factors also reflect IMO Tier 3 NOx regulations that apply to engines installed on ships
constructed (i.e., keel is laid) on or after January 1st, 2016. However, in allotting time for ship building
and engine installation, EPA does not expect Tier 3 vessels to be active by December 31st, 2016. The
2017 CMV emissions were computed taking these standards into account.
3.2.6 Emissions from Canada, Mexico (othpt, othar, othafdust, othptdust, onroadcan, onroadmex,
ptfire_othna)
The emissions from Canada and Mexico are included as part of the emissions modeling sectors: othpt,
othar, othafdust, othptdust, onroad can, and onroad mex. The "oth" refers to the fact that these emissions
are usually "other" than those in the U.S. state-county geographic FIPS, and the remaining characters
provide the SMOKE source types: "pt" for point, "ar" for area and nonroad mobile, "afdust" for area
fugitive dust (Canada only), and "ptdust" for point fugitive dust (Canada only). The onroad emissions for
Canada and Mexico are in the onroad can and onroad mex sectors, respectively. Environment and
Climate Change Canada (ECCC) provided a set of inventories for the year 2015. ECCC also provided
data for the year 2023, either in the form of brand new inventories, or projection factors to apply to the
2015 data. Those 2015 and 2023 inventories were interpolated to the year 2017 for all Canadian
inventories used in this study except for fires and CMV.
ECCC provided the following 2015 inventories for use in this study:
Ag livestock and fertilizer, point source format (othpt sector)
Ag fugitive dust, point source format (othptdust sector)
Other area source dust (othafdust sector)
Airports, point source format (othpt sector)
Onroad (onroad can sector)
- Nonroad and rail (othar sector)
CMV, provided as area sources but converted to point (othpt sector)
Other area sources (othar sector)
Other point sources, including oil and gas (othpt sector)
Future year Canadian emissions were also available, and for this study Canadian emissions interpolated to
the year 2017 were used.
ECCC provided all CMV emissions as an area source inventory. The 2017 NEI CMV included most
coastal waters of Canada and Mexico with emissions derived from AIS data, these NEI emissions were
used for all areas of Canada and Mexico in which they were available. Both the C1C2 and C3 emissions
were provided in a point source format with points at the center of the 12km grid cells. Activity and
corresponding emissions along the St. Lawrence Seaway were not included in the NEI. This region was
gapfilled with emissions provided by ECCC that were apportioned to point sources on the centroids of
12km grid cells using the Canadian commercial marine vessel surrogate (CA 945).
7 http://ec.europa.eu/environment/air/pdf/chapterl ship emissions.pdf.
38
-------
One of the Canadian point source inventories includes pre-speciated VOC emissions for the CB6
mechanism. However, this inventory did not include all species needed for the CB6 mechanism for
CMAQ; specifically, CH4, SOAALK, NAPH, and XYLMN were missing. For the NAPH species,
naphthalene emissions from a supplemental HAP inventory provided by ECCC were used. Then, XYL
was converted to XYLMN by subtracting NAPH. Finally, CH4 and SOAALK were speciated from total
VOC (also provided by ECCC) using traditional speciation profiles by SCC. There are also other sources
in Canada, such as oil and gas, for which we do not have pre-speciated VOC emissions and for which we
apply VOC speciation within SMOKE.
The Canadian inventory included fugitive dust emissions that do not incorporate either a transportable
fraction or meteorological-based adjustments. To properly account for this, a separate sector called
othafdust (for area sources) and othptdust (for point sources) were created and modeled using the same
adjustments as are done for U.S. sources. Since fugitive dust emissions were provided in both area and
point format, these emissions needed to be processed as through SMOKE two separate sectors, one for
area sources and one for point sources.
In addition to emissions inventories, the ECCC 2015 dataset also included temporal profiles, and
shapefiles for creating spatial surrogates. These updated profiles and surrogates were used for this study.
Other than the CB6 species of NBAFM present in the speciated point source data, there are no explicit
HAP emissions in these Canadian inventories.
Point sources in Mexico were compiled based on inventories projected from the the Inventario Nacional
de Emisiones de Mexico, 2016 (Secretaria de Medio Ambiente y Recursos Naturales (SEMARNAT)).
The point source emissions were converted to English units and into the FF10 format that could be read
by SMOKE, missing stack parameters were gapfilled using SCC-based defaults, and latitude and
longitude coordinates were verified and adjusted if they were not consistent with the reported
municipality. Mexican point inventories were projected from 2008 to the years 2014 and 2018, and then
those emissions values were interpolated to the year 2016 for this study. Only CAPs are covered in the
Mexico point source inventory.
For Mexican area and nonroad sources, emission projections based on Mexico's 2016 inventory were
used for area, point and nonroad sources (Secretaria de Medio Ambiente y Recursos Naturales
(SEMARNAT)). The resulting inventory was written using English units to the nonpoint FF10 format
that could be read by SMOKE. Note that unlike the U.S. inventories, there are no explicit HAPs in the
nonpoint or nonroad inventories for Canada and Mexico and, therefore, all HAPs are created from
speciation. Similar to the point inventories, Mexican area and nonroad inventories were projected from
2008 to the years 2014 and 2018, and then emissions values were interpolated to year 2016 values for this
study.
For Mexico onroad emissions, a version of the MOVES model for Mexico was run that provided the same
VOC HAPs and speciated VOCs as for the U.S. MOVES model (ERG, 2016a). This includes NBAFM
plus several other VOC HAPs such as toluene, xylene, ethylbenzene and others. Except for VOC HAPs
that are part of the speciation, no other HAPs are included in the Mexico onroad inventory (such as
particulate HAPs nor diesel particulate matter). Mexico onroad inventories were generated by MOVES
for the years 2014 and 2017.
Annual 2017 wildland emissions for Mexico, Canada, Central America, and Caribbean nations are
39
-------
included in the ptfireothna sector. Canadian fires, along with fires in Mexico, Central America, and the
Caribbean, were developed from Fire Inventory from NCAR (FINN) 2017 vl.5 daily fire emissions. For
FINN fires, listed vegetation type codes of 1 and 9 are defined as agricultural burning, all other fire
detections and assumed to be wildfires. All wildland fires that are not defined as agricultural are assumed
to be wildfires rather than prescribed. FINN fire detects of less than 50 square meters (0.012 acres) are
removed from the inventory. The locations of FINN fires are geocoded from latitude and longitude to
FIPS code.
3.2.7 SMOKE-ready non-anthropogenic chlorine inventory
The ocean chlorine gas emission estimates are based on the build-up of molecular chlorine (C12)
concentrations in oceanic air masses (Bullock and Brehme, 2002). Data at 36 km and 12 km resolution
were available and were not modified other than the name "CHLORINE" was changed to "CL2" because
that is the name required by the CMAQ model.
3.3 Emissions Modeling Summary
The CMAQ model requires hourly emissions of specific gas and particle species for the horizontal and
vertical grid cells contained within the modeled region (i.e., modeling domain). To provide emissions in
the form and format required by the model, it is necessary to "pre-process" the "raw" emissions (i.e.,
emissions input to SMOKE) for the sectors described above. In brief, the process of emissions modeling
transforms the emissions inventories from their original temporal resolution, pollutant resolution, and
spatial resolution into the hourly, speciated, gridded resolution required by the air quality model.
Emissions modeling includes temporal allocation, spatial allocation, and pollutant speciation. In some
cases, emissions modeling also includes the vertical allocation of point sources, but many air quality
models also perform this task because it greatly reduces the size of the input emissions files if the vertical
layers of the sources are not included.
As previously discussed, the temporal resolutions of the emissions inventories input to SMOKE vary
across sectors and may be hourly, daily, monthly, or annual total emissions. The spatial resolution, may
be individual point sources, county/province/municipio totals, or gridded emissions and varies by sector.
This section provides some basic information about the tools and data files used for emissions modeling
as part of the modeling platform.
3.3.1 The SMOKE Modeling System
SMOKE version 4.7 was used to pre-process the raw emissions inventories into emissions inputs for
CMAQ. SMOKE executables and source code are available from the Community Multiscale Analysis
System (CMAS) Center at http://www.cmasceiiter.ore. Additional information about SMOKE is available
from http: /'/ www, sm oke-m od el. org. For sectors that have plume rise, the in-line emissions capability of the
air quality models was used, which allows the creation of source-based and two-dimensional gridded
emissions files that are much smaller than full three-dimensional gridded emissions files. For quality
assurance of the emissions modeling steps, emissions totals by specie for the entire model domain are
output as reports that are then compared to reports generated by SMOKE on the input inventories to
ensure that mass is not lost or gained during the emissions modeling process.
3.3.2 Key Emissions Modeling Settings
When preparing emissions for the air quality model, emissions for each sector are processed separately
through SMOKE, and then the final merge program (Mrggrid) is run to combine the model-ready, sector-
40
-------
specific emissions across sectors. The SMOKE settings in the run scripts and the data in the SMOKE
ancillary files control the approaches used by the individual SMOKE programs for each sector. Table 3-6
summarizes the major processing steps of each platform sector. The "Spatial" column shows the spatial
approach used: here "point" indicates that SMOKE maps the source from a point location (i.e., latitude
and longitude) to a grid cell; "surrogates" indicates that some or all of the sources use spatial surrogates to
allocate county emissions to grid cells; and "area-to-point" indicates that some of the sources use the
SMOKE area-to-point feature to grid the emissions. The "Speciation" column indicates that all sectors
use the SMOKE speciation step, though biogenics speciation is done within the Tmpbeis3 program and
not as a separate SMOKE step. The "Inventory resolution" column shows the inventory temporal
resolution from which SMOKE needs to calculate hourly emissions. Note that for some sectors (e.g.,
onroad, beis), there is no input inventory; instead, activity data and emission factors are used in
combination with meteorological data to compute hourly emissions.
Table 3-6. Key emissions modeling steps by sector
Platform sector
Spatial
Speciation
Inventory
resolution
Plume rise
afdust adj
Surrogates
Yes
annual
ag
Surrogates
Yes
annual
airports
Point
Yes
annual
none
beis
Pre-gridded
land use
in BEIS3 .7
computed hourly
cmv clc2
Point
Yes
hourly
in-line
cmv c3
Point
Yes
hourly
in-line
nonpt
Surrogates &
area-to-point
Yes
annual
nonroad
Surrogates
Yes
monthly
np oilgas
Surrogates
Yes
annual
onroad
Surrogates
Yes
monthly activity,
computed hourly
onroadcaadj
Surrogates
Yes
monthly activity,
computed hourly
onroad can
Surrogates
Yes
monthly
onroad mex
Surrogates
Yes
monthly
othafdust ad]
Surrogates
Yes
annual
othptdust adj
Point
Yes
monthly
none
othar
Surrogates
Yes
annual &
monthly
othpt
Point
Yes
annual &
monthly
in-line
ptagfire
Point
Yes
daily
in-line
pt oilgas
Point
Yes
annual
in-line
ptegu
Point
Yes
daily & hourly
in-line
ptfire
Point
Yes
daily
in-line
ptfire othna
Point
Yes
daily
in-line
ptnonipm
Point
Yes
annual
in-line
rail
Surrogates
Yes
annual
41
-------
Platform sector
Spatial
Speciation
Inventory
resolution
Plume rise
rwc
Surrogates
Yes
annual
Biogenic emissions can be modeled two different ways in the CMAQ model. The BEIS model in SMOKE
can produce gridded biogenic emissions that are then included in the gridded CMAQ-ready emissions
inputs, or alternatively, CMAQ can be configured to create "in-line" biogenic emissions within CMAQ
itself. For this study, the in-line biogenic emissions option was used, and so biogenic emissions from
BEIS were not included in the gridded CMAQ-ready emissions.
The "plume rise" column indicates the sectors for which the "in-line" approach is used. These sectors are
the only ones with emissions in aloft layers based on plume rise. The term "in-line" means that the plume
rise calculations are done inside of the air quality model instead of being computed by SMOKE. The air
quality model computes the plume rise using stack parameters and the hourly emissions in the SMOKE
output files for each emissions sector. The height of the plume rise determines the model layer into which
the emissions are placed. All of these particular sectors only have "in-line" emissions, meaning that all of
the emissions are treated as elevated sources and there are no emissions for those sectors in the two-
dimensional, layer-1 files created by SMOKE. Day-specific point fire emissions are treated differently in
CMAQ. After plume rise is applied, there are emissions in every layer from the ground up to the top of
the plume. The emissions in the airports and othptdust sectors are all low-level emissions, and so in-line
emissions files are not created for these two sectors. Instead, all airports and othptdust emissions are
output to gridded emissions files, same as if airports and othptdust were area source sectors.
SMOKE has the option of grouping sources so that they are treated as a single stack when computing
plume rise. For this modeling case, no grouping was performed because grouping combined with "in-
line" processing will not give identical results as "offline" processing (i.e., when SMOKE creates 3-
dimensional files). This occurs when stacks with different stack parameters or lat/lons are grouped,
thereby changing the parameters of one or more sources. The most straightforward way to get the same
results between in-line and offline is to avoid the use of grouping.
3.3.3 Spatial Configuration
For this study, SMOKE was run for the larger 12-km CONtinental United States "CONUS" modeling
domain (12US1) shown in Figure 3-3, but the air quality model was run on the smaller 12-km domain
(12US2). The grid used a Lambert-Conformal projection, with Alpha = 33, Beta = 45 and Gamma = -97,
with a center of X = -97 and Y = 40. Later sections provide details on the spatial surrogates and area-to-
point data used to accomplish spatial allocation with SMOKE.
42
-------
12US1 Continental U S Domain
12US2 Continental US Domain
Figure 3-3. CMAQ Modeling Domain
3.3.4 Chemical Speciation Configuration
The emissions modeling step for chemical speciation creates the "model species" needed by the air
quality model for a specific chemical mechanism. These model species are either individual chemical
compounds (i.e.. "explicit species") or groups of species (i.e., "lumped species"). The chemical
mechanism used for the platform is the CB6 mechanism (Yarwood, 2010). We used an updated version
of CB6 that we refer to as "CB6R3AE7" which includes four new species that were not in the previous
version of CB6: AACD, FACD, APIN, and 1VOC. This mapping uses a new systematic methodology for
mapping low volatility compounds. Compounds with very low vapor pressure are mapped to model
species NVOL and intermediate volatility compounds are mapped to a species called IVOC. In previous
mappings, some of these low vapor pressure compounds were mapped to CB6 species. The mechanism
and mapping are described in more detail in a memorandum describing the mechanism files supplied with
the Speciation Tool, the software used to create the CB6 profiles used in SMOKE. It should be noted that
the onroad mobile sector does not use this newer mapping because the speciation is done within MOVES
and the mapping change was made after MOVES had been run.
This platform generates the PM2.5 model species associated with the CMAQ Aerosol Module version 7
(AE7) which has the same PM2.5 model species as version 6 (AE6). The AE7 mechanism is built on the
AE6 and identical in terms of model species and mechanism definition but requires that alpha-pinene
(APIN) be separate from all other monoterpenes (TERP) and not included in TERP to avoid double
counting.
43
-------
Table 3-7 lists the model species produced by SMOKE in the platform used for this study.
Table 3-7. Emission model species produced for CB6R3AE7 for CMAQ
Inventory Pollutant
Model Species
Model species description
Ch
CL2
Atomic gas-phase chlorine
HC1
HCL
Hydrogen Chloride (hydrochloric acid) gas
CO
CO
Carbon monoxide
NOx
NO
Nitrogen oxide
N02
Nitrogen dioxide
HONO
Nitrous acid
so2
S02
Sulfur dioxide
SULF
Sulfuric acid vapor
nh3
NH3
Ammonia
NH3 FERT
Ammonia from fertilizer
voc
AACD
Acetic acid
ACET
Acetone
ALD2
Acetaldehyde
ALDX
Propionaldehyde and higher aldehydes
APIN
alpha-Pinene
BENZ
Benzene
CH4
Methane
ETH
Ethene
ETHA
Ethane
ETHY
Ethyne
ETOH
Ethanol
FACD
Formic acid
FORM
Formaldehyde
IOLE
Internal olefin carbon bond (R-C=C-R)
ISOP
Isoprene
IVOC
Intermediate volatility organic compounds
KET
Ketone Groups
MEOH
Methanol
NAPH
Naphthalene
NVOL
Non-volatile compounds
OLE
Terminal olefin carbon bond (R-C=C)
PAR
Paraffin carbon bond
PRPA
Propane
SESQ
Sequiterpenes (from biogenics only)
SOAALK
Secondary Organic Aerosol (SOA) tracer
TERP
Terpenes (from biogenics only)
TOL
Toluene and other monoalkyl aromatics
UNR
Unreactive
XYLMN
Xylene and other polyalkyl aromatics, minus
naphthalene
Naphthalene
NAPH
Naphthalene from inventory
Benzene
BENZ
Benzene from the inventory
Acetaldehyde
ALD2
Acetaldehyde from inventory
Formaldehyde
FORM
Formaldehyde from inventory
Methanol
MEOH
Methanol from inventory
44
-------
Inventory Pollutant
Model Species
Model species description
PM10
PMC
Coarse PM >2.5 microns and <10 microns
PM2.5
PEC
Particulate elemental carbon <2.5 microns
PN03
Particulate nitrate <2.5 microns
POC
Particulate organic carbon (carbon only) <2.5 microns
PS04
Particulate Sulfate <2.5 microns
PAL
Aluminum
PCA
Calcium
PCL
Chloride
PFE
Iron
PK
Potassium
PH20
Water
PMG
Magnesium
PMN
Manganese
PMOTHR
PM2.5 not in other AE6 species
PNA
Sodium
PNCOM
Non-carbon organic matter
PNH4
Ammonium
PSI
Silica
PTI
Titanium
Sea-salt species (non -
PCL
Particulate chloride
anthropogenic) 8
PNA
Particulate sodium
The TOG and PM2.5 speciation factors that are the basis of the chemical speciation approach were
developed from the SPECIATE 5.0 database (https://www.epa.gov/air-emissions-modeling/speciate).
which is the EPA's repository of TOG and PM speciation profiles of air pollution sources. The
SPECIATE database development and maintenance is a collaboration involving the EPA's Office of
Research and Development (ORD), Office of Transportation and Air Quality (OTAQ), and the Office of
Air Quality Planning and Standards (OAQPS). The SPECIATE database contains speciation profiles for
TOG, speciated into individual chemical compounds, VOC-to-TOG conversion factors associated with
the TOG profiles, and speciation profiles for PM2.5.
Some key features and updates to speciation from previous platforms include the following (the
subsections below contain more details on the specific changes):
Use of the CBR3AE7 mechanism, as described earlier
Several new VOC and PM2.5 profiles added to SPECIATE 5.0 or slated for the next version of
SPECIATE were used.
PM2.5 speciation process for nonroad mobile has been updated. Similar to VOC, PM2.5 profiles are
now assigned within MOVES2014b which outputs the emissions with those assignments.
As with previous platforms, some Canadian point source inventories are provided from
Environment Canada as pre-speciated emissions, and not all CB6 species were provided; missing
species were supplemented by speciating total VOC.
A new GSPRO COMBO file was developed for use in Canada to account for ethanol mixes in
Canadian gasoline.
8 These emissions are created outside of SMOKE.
45
-------
Speciation profiles and cross-references for this study platform are available in the set SMOKE input files
for the platform. Emissions of VOC and PM2.5 emissions by county, sector and profile for all sectors
other than onroad mobile can be found in the sector summaries for the case. Totals of each model species
by state and sector can be found in the Appendix B state-sector totals workbook for this study.
For onroad mobile sources, speciation is done in MOVES, to allow for profiles that vary by model year,
which is not part of the SCC code, to be used. Therefore, cross-references or emissions summaries by
profile for onroad mobile sources are not provided. These profiles are documented in a MOVES technical
report on speciation (EPA, 2018).
Updates to speciation profiles for VOC and PM2.5, were largely from the update of the SPECIATE
database to version SPECIATE 5.0 in June 2019 which added numerous organic gas and PM
profiles. We also used several profiles added to SPECIATE 5.1 which will be available in July 2020. In
addition, we changed profile assignments to incorporate data provided by states or correct errors in
previous assignments.
For PM: 5 the following profile updates were made for the 2017 platform:
Corrected the wildfire and prescribed fire profile due to error in compositing (the previous profile
included creosote in the average)
Updated the profile for aircraft
Corrected several profile assignments for the petroleum industry
For VOC the following profile updates were made for the 2017 platform:
Consumer products - replaced the profiles developed from the CARB 1997 consumer products
survey with profiles developed from the CARB 2010 survey update
Oil and Gas - used additional region-specific profiles or updated assignments
o Used county-specific profiles gas for several Wyoming counties developed from data
provided by the Wyoming DEQ
o Used Willi son Basin gas composition data, separate profiles for the Montana and North
Dakota portions of the basin, based on data developed by the Western Regional Air
Partnership WRAP
o Used Central Montana Uplift area gas composition data, based on data developed by the
WRAP
o Updated Uinta basin profile assignments (based on data provided by Utah)
o Used Utah and Wyoming oil and gas produced water pond profiles
o Updated profile assignments (by county and SCC) for nonpoint oil and gas sources that
account for the portion of VOC estimated to come from flares. These were updated using
results from the Oil and Gas estimation tool run that was used for the 2017 NEI.
o Updated profile assignment for miscellaneous engines to use internal combustion engine
natural gas profile
Commercial Marine vessel - changed profile assignment to an existing Pre-Tier 1 nonroad diesel
profile because the previous profile was missing key species (aldehydes)
Livestock - updated profile assignments
Agricultural burning - updated profiles for rice straw and wheat straw burning, and used new
sugar cane burning profile
46
-------
The speciation of VOC includes HAP emissions from the emissions inventories in the speciation process.
Instead of speciating VOC to generate all of the species listed in Table 3-7, emissions of five specific
HAPs: naphthalene, benzene, acetaldehyde, formaldehyde, and methanol (collectively known as
"NBAFM") from the NEI were "integrated" with the NEI VOC. The integration combines these HAPs
with the VOC in a way that does not double count emissions and uses the HAP inventory directly in the
speciation process. The basic process is to subtract the specified HAPs emissions mass from the VOC
emissions mass, and to then use a special "integrated" profile to speciate the remainder of VOC to the
model species excluding the specific HAPs. The EPA believes that the HAP emissions in the NEI are
often more representative of emissions than HAP emissions generated via VOC speciation, although this
varies by sector.
The NBAFM HAPs were chosen for integration because they are the only explicit VOC HAPs in CMAQ
version 5.2. Explicit means that they are not lumped chemical groups like PAR, IOLE and several other
CB6 model species. These "explicit VOC HAPs" are model species that participate in the modeled
chemistry using the CB6 chemical mechanism. The use of inventory HAP emissions along with VOC is
called "HAP-CAP integration."
The integration of HAP VOC with VOC is a feature available in SMOKE for all inventory formats,
including PTDAY (the format used for the ptfire and ptagfire sectors). The ability to use integration with
the PTDAY format was made available in the version of SMOKE used for the v7.1 platform, and this new
feature is used for this particular study because the ptfire and ptagfire inventories for this study include
HAPs. SMOKE allows the user to specify both the particular HAPs to integrate via the INVTABLE.
This is done by setting the "VOC or TOG component" field to "V" for all HAP pollutants chosen for
integration. SMOKE allows the user to also choose the particular sources to integrate via the
NHAPEXCLUDE file (which actually provides the sources to be excluded from integration9). For the
"integrated" sources, SMOKE subtracts the "integrated" HAPs from the VOC (at the source level) to
compute emissions for the new pollutant "NONHAPVOC." The user provides NONHAPVOC-to-
NONHAPTOG factors and NONHAPTOG speciation profiles.10 SMOKE computes NONHAPTOG and
then applies the speciation profiles to allocate the NONHAPTOG to the other air quality model VOC
species not including the integrated HAPs. After determining if a sector is to be integrated, if all sources
have the appropriate HAP emissions, then the sector is considered fully integrated and does not need a
NHAPEXCLUDE file. If, on the other hand, certain sources do not have the necessary HAPs, then an
NHAPEXCLUDE file must be provided based on the evaluation of each source's pollutant mix. The
EPA considered CAP-HAP integration for all sectors in determining whether sectors would have full, no,
or partial integration (see Figure 3-4). For sectors with partial integration, all sources are integrated other
than those that have either the sum of NBAFM > VOC or the sum of NBAFM = 0.
In this platform, we create NBAFM species from the no-integrate source VOC emissions using speciation
profiles. Figure 3-4 illustrates the integrate and no-integrate processes for U.S. Sources. Since Canada
and Mexico inventories do not contain HAPs, we use the approach of generating the HAPs via speciation,
except for Mexico onroad mobile sources where emissions for integrate HAPs were available.
9 Since SMOKE version 3.7, the options to specify sources for integration are expanded so that a user can specify the
particular sources to include or exclude from integration, and there are settings to include or exclude all sources within a
sector. In addition, the error checking is significantly stricter for integrated sources. If a source is supposed to be integrated,
but it is missing NBAFM or VOC, SMOKE will now raise an error.
10 These ratios and profiles are typically generated from the Speciation Tool when it is run with integration of a specified list
of pollutants, for example NBAFM.
47
-------
It should be noted that even though NBAFM were removed from the SPECIATE profiles used to create
the GSPRO for both the NONHAPTOG and no-integrate TOG profiles, there still may be small fractions
for "BENZ", "FORM", "ALD2", and "MEOH" present. This is because these model species may have
come from species in SPECIATE that are mixtures. The quantity of these model species is expected to be
very small compared to the BAFM in the NEI. There are no NONHAPTOG profiles that produce
"NAPH."
In SMOKE, the INVTABLE allows the user to specify the particular HAPs to integrate. Two different
INVTABLE files are used for different sectors of the platform. For sectors that had no integration across
the entire sector (see Table 3-8), EPA created a "no HAP use" INVTABLE in which the "KEEP" flag is
set to "N" for NBAFM pollutants. Thus, any NBAFM pollutants in the inventory input into SMOKE are
automatically dropped. This approach both avoids double-counting of these species and assumes that the
VOC speciation is the best available approach for these species for sectors using this approach. The
second INVTABLE, used for sectors in which one or more sources are integrated, causes SMOKE to keep
the inventory NBAFM pollutants and indicates that they are to be integrated with VOC. This is done by
setting the "VOC or TOG component" field to "V" for all four HAP pollutants. Note for the onroad
sector, "full integration" includes the integration of benzene, 1,3 butadiene, formaldehyde, acetaldehyde,
naphthalene, acrolein, ethyl benzene, 2,2,4-Trimethylpentane, hexane, propionaldehyde, styrene, toluene,
xylene, and MTBE.
I Emissions Ready for SMOKE
List of "no-integrate" sources
(NHAPEXCLUDE)
Speciation cross
reference file (GSREF)
NONHAPVOC to NONHAPTOG
factors (GSCNV)
NONHAPTOG speciation factors (GSPRO)
TOG speciation factors for which NBAFM
compounds removed prior to GSPRO creation
Assign speciation profile to each source
Compute NONHAPVOC= VOC- (N+B+A+F+M) for each
integrate source
Retain VOC for each no-integrate source
Compute NONHAPTOG from NONHAPVOC for each integrate
source
Compute TOG from VOC for each no-integrate source
Compute CMAQ-CB6 Species:
For integrate source: Use (1) NONHAPTOG profiles applied to
NONHAPTOG and (2) N,BARM from inventory
For no-integrate source: Use (1) non-normalized TOG profiles
applied to TOG and (2) N,B,A,F,M from inventory
SMOKE
CMAQ-CB6 species
Figure 34. Process of integrating NBAFM with VOC for use in VOC Speciation
48
-------
Table 3-8. Integration status of naphthalene, benzene, acetaldehyde, formaldehyde and methanol
(NBAFM) for each platform sector
Platform Sector
Approach for Integrating NEI emissions of Naphthalene (N), Benzene (B),
Acetaldehyde (A), Formaldehyde (F) and Methanol (M)
afdust
N/A - sector contains no VOC
ag
Partial integration (NBAFM)
airports
No integration, create NBAFM from VOC speciation
beis
N/A - sector contains no inventory pollutant "VOC"; but rather specific VOC species
cmv clc2
No integration, create NBAFM from VOC speciation
cmv c3
No integration, create NBAFM from VOC speciation
nonpt
Partial integration (NBAFM)
nonroad
Full integration (internal to MOVES elsewhere)
np oilgas
Partial integration (NBAFM)
onroad
Full integration (internal to MOVES); however, MOVES2014a speciation was CB6-CAMx,
not CB6-CMAQ, so post-SMOKE emissions were converted to CB6-CMAQ
onroad can
No integration, no NBAFM in inventory, create NBAFM from speciation
onroadmex
Full integration (internal to MOVES-Mexico); however, MOVES-MEXICO speciation was
CB6-CAMx, not CB6-CMAQ, so post-SMOKE emissions were converted to CB6-CMAQ
othafdust
N/A - sector contains no VOC
othar
No integration, no NBAFM in inventory, create NBAFM from VOC speciation
othpt
No integration, no NBAFM in inventory, create NBAFM from VOC speciation
othptdust
N/A - sector contains no VOC
pt oilgas
No integration, create NBAFM from VOC speciation
ptagfire
Full integration (NBAFM)
Ptegu
No integration, create NBAFM from VOC speciation
ptfire
Partial integration (NBAFM)
ptfire othna
No integration, no NBAFM in inventory, create NBAFM from VOC speciation
ptnonipm
No integration, create NBAFM from VOC speciation
rail
Full integration (NBAFM)
rwc
Partial integration (NBAFM)
Integration for the mobile sources estimated from MOVES (onroad and nonroad sectors, other than for
California) is done differently. Briefly there are three major differences: 1) for these sources integration
is done using more than just NBAFM, 2) all sources from the MOVES model are integrated, and 3)
integration is done fully or partially within MOVES. For onroad mobile, speciation is done fully within
MOVES2014a such that the MOVES model outputs emission factors for individual VOC model species
along with the HAPs. This requires MOVES to be run for a specific chemical mechanism. MOVES was
run for the CB6-CAMx mechanism rather than CB6-CMAQ, so post-SMOKE onroad emissions were
converted to CB6-CMAQ. For nonroad mobile, speciation is partially done within MOVES such that it
does not need to be run for a specific chemical mechanism. For nonroad, MOVES outputs emissions of
HAPs and NONHAPTOG were split by speciation profile. Taking into account that integrated species
were subtracted out by MOVES already, the appropriate speciation profiles are then applied in SMOKE to
get the VOC model species. HAP integration for nonroad uses the same additional HAPs and ethanol as
for onroad.
In previous platforms, the GSPRO COMBO feature was used to speciate nonroad mobile and gasoline-
related stationary sources that use fuels with varying ethanol content. In these cases, the speciation
profiles require different combinations of gasoline profiles, e.g. EO and E10 profiles. Since the ethanol
content varies spatially (e.g., by state or county), temporally (e.g., by month), and by modeling year
49
-------
(future years have more ethanol), the GSPROCOMBO feature allows combinations to be specified at
various levels for different years. For the 2017 platform, GSPRO COMBO is still used for certain
gasoline-related stationary sources nationwide. GSPRO COMBO is no longer needed for nonroad
sources because nonroad emissions within MOVES have the speciation profiles built into the results, so
there is no need to assign them via the GSREF or GSPRO COMBO feature.
In Canada, ECCC provided estimates of ethanol mixes by Canadian province. These estimates were used
to develop a GSPRO COMBO for Canadian gasoline onroad emissions. For example, a province where
the average ethanol mix is 6% would have 60% E10 speciation and 40% E0 speciation. A 10% ethanol
mix would imply 100% E10 speciation. In Mexico, only E0 speciation profiles are used, but the
GSPRO COMBO feature is still used in Mexico for inventories where VOC emissions are not explicitly
defined by mode (e.g., exhaust versus evaporative). Here, the GSPRO COMBO specifies a mix of
exhaust and evaporative speciation profiles. Using the GSPRO COMBO to split total VOC into exhaust
and evaporative components is no longer necessary for Canadian mobile sources, whose inventories now
include the mode in the pollutant, or for Mexico onroad sources, where VOC speciation is calculated by
the MOVES model. The GSPRO COMBO is still used for Mexican nonroad sources which do not have
modes in the inventory.
The newer combo method for combining profiles that is available in SMOKE4.5 and later versions is used
in this platform. It allows multiple profiles to be combined by pollutant, state and county (i.e.,
state/county FIPS code) and SCC. This is used specifically for the np oilgas sector because SCCs include
both controlled and uncontrolled oil and gas operations which use different profiles. The underlying data
which defines the profile splits for oil and gas was updated for this 2017 study.
Speciation profiles for use with BEIS are not included in SPECIATE. BEIS includes a species (SESQ) that
was mapped to the CMAQ specie SESQT. The profile code associated with BEIS profiles for use with
CB6 was "BC6E7."
NOx can be speciated into NO, N02, and/or HONO. For the non-mobile sources, EPA used a single
profile "NHONO" to split NOx into NO and NO2. For the mobile sources except for onroad (including
nonroad, cmv_clc2, cmv_c3, rail, onroad can, onroadmex sectors) and for specific SCCs in othar and
ptnonipm, the profile "HONO" splits NOx into NO, NO2, and HONO. Table 3-9 gives the split factor for
these two profiles. The onroad sector does not use the "HONO" profile to speciate NOx. MOVES2014
produces speciated NO, NO2, and HONO by source, including emission factors for these species in the
emission factor tables used by SMOKE-MOVES. Within MOVES, the HONO fraction is a constant
0.008 of NOx. The NO fraction varies by heavy duty versus light duty, fuel type, and model year, and
equals 1 - NO - HONO. For more details on the NOx fractions within MOVES, see
https://www.epa.gOv/moves/moves-onroad-technical-reports#moves2014.
Table 39. NOx speciation profiles
Profile
pollutant
species
split factor
HONO
NOX
N02
0.092
HONO
NOX
NO
0.9
HONO
NOX
HONO
0.008
NHONO
NOX
N02
0.1
NHONO
NOX
NO
0.9
50
-------
3.3.5 Temporal Processing Configuration
Temporal allocation (i.e., temporalization) is the process of distributing aggregated emissions to a finer
temporal resolution, thereby converting annual emissions to hourly emissions. While the total emissions
are important, the timing of the occurrence of emissions is also essential for accurately simulating ozone,
PM, and other pollutant concentrations in the atmosphere. Many emissions inventories are annual or
monthly in nature. Temporalization takes these aggregated emissions and, if needed, distributes them to
the month, and then distributes the monthly emissions to the day and the daily emissions to the hours of
each day. This process is typically done by applying temporal profiles to the inventories in this order:
monthly, day of the week, and diurnal.
The temporal factors applied to the inventory are selected using some combination of country, state,
county, SCC, and pollutant. Table 3-10 summarizes the temporal aspects of emissions modeling by
comparing the key approaches used for temporal processing across the sectors. In the table, "Daily
temporal approach" refers to the temporal approach for getting daily emissions from the inventory using
the SMOKE Temporal program. The values given are the values of the SMOKE L TYPE setting. The
"Merge processing approach" refers to the days used to represent other days in the month for the merge
step. If this is not "all," then the SMOKE merge step runs only for representative days, which could
include holidays as indicated by the right-most column. The values given are those used for the SMOKE
M TYPE setting (see below for more information).
Table 3-10. Temporal Settings Used for the Platform Sectors in SMOKE
Platform sector
short name
Inventory
resolutions
Monthly
profiles
used?
Daily
temporal
approach
Merge
processing
approach
Process Holidays
as separate days
afdust adi
Annual
Yes
week
all
Yes
ag
Annual
Yes
met-based
all
Yes
airports
Annual
Yes
week
week
Yes
beis
Hourly
n/a
all
No
cmv clc2
Hourly
all
all
No
cmv c3
Hourly
all
all
No
nonpt
Annual
Yes
week
week
Yes
nonroad
Monthly
mwdss
mwdss
Yes
np oilgas
Annual
Yes
aveday
aveday
No
onroad
Annual & monthly1
all
all
Yes
onroad ca adj
Annual & monthly1
all
all
Yes
othafdust adj
Annual
Yes
week
week
No
othptdust adj
Monthly
week
week
No
othar
Annual & monthly
Yes
week
week
No
onroad can
Monthly
week
week
No
onroad mex
Monthly
week
week
No
othpt
Annual & monthly
Yes
mwdss
mwdss
No
ptagfire
Daily
all
all
No
pt oilgas
Annual
Yes
mwdss
mwdss
Yes
ptegu
Annual & hourly
Yes2
all
all
No
51
-------
Platform sector
short name
Inventory
resolutions
Monthly
profiles
used?
Daily
temporal
approach
Merge
processing
approach
Process Holidays
as separate days
ptnonipm
Annual
Yes
mwdss
mwdss
Yes
ptfire
Daily
all
all
No
ptfire othna
Daily
all
all
No
rail
Annual
Yes
aveday
aveday
No
rwc
Annual
No3
met-based
all
No3
1. Note the annual and monthly "inventory" actually refers to the activity data (VMT and VPOP) for onroad. The
actual emissions are computed on an hourly basis.
2. Only units that do not have matching hourly CEMs data use monthly temporal profiles.
3. Except for 2 SCCs that do not use met-based temporalization.
The following values are used in the above table: The value "all" means that hourly emissions are
computed for every day of the year and that emissions potentially have day-of-year variation. The value
"week" means that hourly emissions computed for all days in one "representative" week, representing all
weeks for each month. This means emissions have day-of-week variation, but not week-to-week variation
within the month. The value "mwdss" means hourly emissions for one representative Monday,
representative weekday (Tuesday through Friday), representative Saturday, and representative Sunday for
each month. This means emissions have variation between Mondays, other weekdays, Saturdays and
Sundays within the month, but not week-to-week variation within the month. The value "aveday" means
hourly emissions computed for one representative day of each month, meaning emissions for all days
within a month are the same. Special situations with respect to temporalization are described in the
following subsections.
In addition to the resolution, temporal processing includes a ramp-up period for several days prior to
January 1, 2017, which is intended to mitigate the effects of initial condition concentrations. The ramp-up
period was 10 days (December 22-31, 2016). For all anthropogenic sectors, emissions from December
2017 were used to fill in surrogate emissions for the end of December 2016. In particular, December
2017 emissions (representative days) were used for December 2016. For biogenic emissions, December
2016 emissions were processed using 2016 meteorology.
The Flat File 2010 format (FF10) inventory format for SMOKE provides a more consolidated format for
monthly, daily, and hourly emissions inventories than prior formats supported. Previously, processing
monthly inventory data required the use of 12 separate inventory files. With the FF10 format, a single
inventory file can contain emissions for all 12 months and the annual emissions in a single record. This
helps simplify the management of numerous inventories. Similarly, daily and hourly FF10 inventories
contain individual records with data for all days in a month and all hours in a day, respectively.
SMOKE prevents the application of temporal profiles on top of the "native" resolution of the inventory.
For example, a monthly inventory should not have annual-to-month temporalization applied to it; rather,
it should only have month-to-day and diurnal temporalization. This becomes particularly important when
specific sectors have a mix of annual, monthly, daily, and/or hourly inventories. The flags that control
temporalization for a mixed set of inventories are discussed in the SMOKE documentation. The
modeling platform sectors that make use of monthly values in the FF10 files are nonroad, onroad (for
activity data), onroad can, onroadmex, othar, othpt, and othptdust.
52
-------
3.3.5.1 Standard Temporal Profiles
Some sectors use straightforward temporal profiles not based on meteorology or other factors. For the
ptfire, ptagfire, and ptfire othna sectors, the inventories are in the daily point fire format, so temporal
profiles are only used to go from day-specific to hourly emissions. For all agricultural burning, the
diurnal temporal profile used reflected the fact that burning occurs during the daylight. This puts most of
the emissions during the work day and suppresses the emissions during the middle of the night. This
diurnal profile was used for each day of the week for all agricultural burning emissions in all states.
For the cmv sectors, most of the emissions are computed hourly and not dependent on temporal profiles.
For areas that did not use AIS data, CI and C2 monthly profiles are flat and C3 monthly profiles are
highest (but not significantly different from the rest of the year) in the summer.
A monthly temporal profile for freight rail was developed from AAR data for the year 2016
(https://www.aar.org/data-center/rail-traffic-data/) and continues to be used in this study, Monthly Rail
Traffic Data, Total Carloads & Intermodal. Passenger trains use a flat monthly profile. Monthly passenger
miles data are available; however, it is not known if there is a correlation between passenger miles and
actual rail emissions. This is because passenger trains often operate on a fixed schedule, independent of
actual passenger traffic. So, it was decided to not apply a monthly profile to passenger train emissions. All
sources in the rail sector use a flat profile for both day-of-week and hour-of-day temporalization.
For the ptfire and ptagfire sectors, the inventories are in the daily point fire format, so temporal profiles
are only used to go from day-specific to hourly emissions. For ptfire, state-specific hourly profiles were
used, with distinct profiles for prescribed fires and wildfires. For ptagfire, the diurnal temporal profile
used reflected the fact that burning occurs during the daylight hours. Additional details on these profiles
are available in the 2016vl platform TSD (EPA, 2020b).
For the nonroad sector, while the NEI only stores the annual totals, the modeling platform uses monthly
inventories from output from MOVES. For California, CARB's annual inventory was temporalized to
monthly using monthly temporal profiles applied in SMOKE by SCC.
Diurnal, weekly, and monthly temporal profiles for aviation-related sources were updated in the 2014v7.0
platform based on aviation metrics. Details on these new profiles are available in the 2014v7.0 TSD.
Temporal profiles for small airports (i.e., non-commercial) do not have any emissions between 10 PM and
6 AM due to a lack of tower operations. Industrial processes that are not likely to shut down on Sundays
such as those at cement plants are assigned to other more realistic profiles that included emissions on
Sundays. This also affected emissions on holidays because Sunday emissions are also used on holidays.
Monthly temporalization of np oilgas emissions is based primarily on monthly factors from the Oil and
Gas Tool (OGT). Factors were specific to each county and SCC. For use in SMOKE, each unique set of
factors was assigned a label (OG17 0001 through OG17 6272), and then a SMOKE-formatted
ATPRO MONTHLY and an ATREF were developed. This dataset of monthly temporal factors included
profiles for all counties and SCCs in the Oil and Gas Tool inventory. Because we are using non-tool
datasets in some states, this monthly temporalization dataset did not cover all counties and SCCs in the
entire inventory used for this study. To fill in the gaps in those states, state average monthly profiles for
oil, natural gas, and combination sources were calculated from EIA data and assigned to each county/SCC
combination not already covered by the OGT monthly temporal profile dataset. Coal bed methane (CBM)
and natural gas liquid sources in those four states were assigned flat monthly profiles where there was not
53
-------
already a profile assignment in the ERG dataset.
For agricultural livestock, annual-to-month profiles were developed based on daily emissions data output
from the CMU model by state and SCC. These profiles were used to temporally allocate ag livestock
emissions to monthly emissions, which are further temporally allocated to hours as described below in
section 3.3.5.3.
3.3.5.2 Temporal Profiles for EGUs
Electric generating unit (EGU) sources matched to ORIS units were temporally allocated to hourly
emissions needed for modeling using the hourly CEMS data. Those hourly data were processed through
v2.1 of the CEMCorrect tool to mitigate the impact of unmeasured values in the data. An example of
before and after the application of the tool is shown in Figure 3-5.
2017 January Unit 469_5
2000
1800
1600
^ 1400
2 1200
^ 1000
x 800
Z 600
400
200
0 i. ->..1 i ¦ .... - . . i i i.
^Hr^rocriLn^HP^rocriLn^HP^rocriLn^HP^rocriLn^HP^rocriLn^HP^rocr)
T-i^rtO(T>rM^rr--(T>rMLnr--orM
^H^H^H^HrNjrNjrNjrNjrorororo^r^r^r^rLnLnLnLntDtDtDr^r^
January 2017 Hour
Raw CEM Corrected
Figure 3-5. Eliminating unmeasured spikes in CEMS data
The region, fuel, and type (peaking or non-peaking) must be identified for each input EGU with CEMS
data that are used for generating profiles. The identification of peaking units was done using hourly heat
input data from the 2017 base year and the two previous years (2015 and 2016). The heat input was
summed for each year. Equation 1 shows how the annual heat input value is converted from heat units
(BTU/year) to power units (MW) using the NEEDS v6 derived unit-level heat rate (BTU/kWh). In
equation 2 a capacity factor is calculated by dividing the annual unit MW value by the NEEDS v6 unit
capacity value (MW) multiplied by the hours in the year. A peaking unit was defined as any unit that had
a maximum capacity factor of less than 0.2 for every year (2015, 2016, and 2017) and a 3-year average
capacity factor of less than 0.1.
54
-------
Equation 1. Annual unit power output
8760 Hourly HI (mw\
Si=o (dt//^ 1000 ( , )
Annual Unit Output (MW) = 7btU\
NEEDS Heat Rate I ff^-)
\kWhJ
Equation 2. Unit capacity factor
_ Annual Unit Output (MW)
Capacity Factor =
NEEDS
SMW\
Unit Capacity (J*8760 (h)
Input regions were determined from one of the eight EGU modeling regions based on multi-jurisdictional
planning organization (MJO) and climate regions. Regions were used to group units with similar climate-
based load demands. Region assignment is made on a state level, where all units within a state were
assigned to the appropriate region as shown in Figure 3-6. Unit fuel assignments were made using the
primary NEEDS v6 fuel. Units fueled by bituminous, subbituminous, or lignite are assigned to the coal
fuel type. Natural gas units were assigned to the gas fuel type. Distillate and residual fuel oil were
assigned to the oil fuel type. Units with any other primary fuel were assigned the "other" fuel type.
Currently there is a possible region, fuel, and type group maximum of 64 based on 8 regions, 4 fuels, and
two types (peaking and non-peaking).
The daily and diurnal profiles were calculated for each region, fuel, and peaking type group from the 2016
CEM heat input values. The heat input values were summed for each input group to the annual level at
each level of temporal resolution: monthly, month-of-day, and diurnal. The sum by temporal resolution
value is then divided by the sum of annual heat input in that group to get a set of temporalization factors.
Diurnal factors were created for both the summer and winter seasons to account for the variation in hourly
load demands between the seasons. For example, the sum of all hour 1 heat input values in the group was
divided by the sum of all heat inputs over all hours to get the hour 1 factor. Each grouping contained 12
monthly factors, up to 31 daily factors per month, and two sets of 24 hourly factors. The profiles were
weighted by unit size where the units with more heat input have a greater influence on the shape of the
profile. Composite profiles were created for each region and type across all fuels as a way to provide
profiles for a fuel type that does not have hourly CEM data in that region. Figure 3-7 shows peaking and
non-peaking daily temporal profiles for the gas fuel type in the LADCO region. Figure 3-8 shows the
diurnal profiles for the coal fuel type in the Mid-Atlantic/Northeast Visibility Union (MANE VU) region.
55
-------
Northwest
[MANE-V,U?
South
Southwest
SESARM
EGU Regions
¦ LADCO
B MANE-VU
I I Northwest
I I SESARM
I South
I I Southwest
I I West
II 1 West North Central
Figure 3-6. Small EGU Temporal Profile Regions
56
-------
0.040
0.035 -
0.030
c 0.025 H
o
Daily Small EGU Profile for LADCO gas
Nonpeaking
Peaking
u
03
0.020
ro
A 0.015
0.010 -
0.005
0.000
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
2017
Figure 3-7. Example Daily Temporal Profiles for the LADCO region and Gas Fuel Type
57
-------
Diurnal Small EGU Profile for MANE-VU coal
o.io
Summer Nonpeaking
Summer Peaking
Winter Nonpeaking
Winter Peaking
0.06 -
o
<0
3
'~
0
5
10
15
20
Hour
Figure 3-8. Example Diurnal Profile for MANE-VU Region and Coal Fuel Type
SMOKE uses a cross-reference file to select a monthly, daily, and diurnal profile for each source. For the
2017 platform, the temporal profiles were assigned in the cross-reference at the unit level to EGU sources
without hourly CEM data. An inventory of all EGU sources without CEMS data was used to identify the
region, fuel type, and type (peaking/non-peaking) of each source. As with the input unit the regions are
assigned using the state from the unit FIPS. The fuel is assigned by SCC to one of the four fuel types:
coal, gas, oil, and other. A fuel type unit assignment is made by summing the VOC, NOx, PM2.5, and SO2
for all SCCs in the unit. The SCC that contributed the highest total emissions to the unit for selected
pollutants was used to assign the unit fuel type. Peaking units were identified as any unit with an oil, gas,
or oil fuel type with a NAICS of 22111 or 221112. Some units may be assigned to a fuel type within a
region that does not have an available input unit with a matching fuel type in that region. These units
without an available profile for their group were assigned to use the regional composite profile. Municipal
waste combustors (MWC) and cogeneration units were identified using the NEEDS primary fuel type and
cogeneration flag, respectively, from the NEEDS v6 database.
3.3.5.3 Meteorological-based Temporal Profiles
There are many factors that impact the timing of when emissions occur, and for some sectors this includes
meteorology. The benefits of utilizing meteorology as a method for temporalization are: (1) a
meteorological dataset consistent with that used by the AQ model is available (e.g., outputs from WRF);
(2) the meteorological model data are highly resolved in terms of spatial resolution; and (3) the
meteorological variables vary at hourly resolution and can therefore be translated into hour-specific
temporalization.
The SMOKE program GenTPRO provides a method for developing meteorology-based temporalization.
Currently, the program can utilize three types of temporal algorithms: annual-to-day temporalization for
58
-------
residential wood combustion (RWC), month-to-hour temporalization for agricultural livestock ammonia,
and a generic meteorology-based algorithm for other situations. For this platform, meteorological-based
temporalization was used for portions of the rwc sector and for the entirety of the ag sector.
GenTPRO reads in gridded meteorological data (output from MCIP) along with spatial surrogates and
uses the specified algorithm to produce a new temporal profile that can be input into SMOKE. The
meteorological variables and the resolution of the generated temporal profile (hourly, daily, etc.) depend
on the selected algorithm and the run parameters. For more details on the development of these
algorithms and running GenTPRO, see the GenTPRO documentation and the SMOKE documentation at
http://www.cmascenter.Org/smoke/documentation/3.l/GenTPRQ Technical Summary Aug2012 Final, pd
f and https://www.cmascenter.Org/smoke/documentation/4.6/html/ch05s03s05.html. respectively.
For the RWC algorithm, GenTPRO uses the daily minimum temperature to determine the temporal
allocation of emissions to days. GenTPRO was used to create an annual-to-day temporal profile for the
RWC sources. These generated profiles distribute annual RWC emissions to the coldest days of the year.
On days where the minimum temperature does not drop below a user-defined threshold, RWC emissions
for most sources in the sector are zero. Conversely, the program temporally allocates the largest
percentage of emissions to the coldest days. Similar to other temporal allocation profiles, the total annual
emissions do not change, only the distribution of the emissions within the year is affected. The
temperature threshold for RWC emissions was 50 °F for most of the country, and 60 °F for the following
states: Alabama, Arizona, California, Florida, Georgia, Louisiana, Mississippi, South Carolina, and
Texas.
Figure 3-9 illustrates the impact of changing the temperature threshold for a warm climate county. The
plot shows the temporal fraction by day for Duval County, Florida for the first four months of 2007. The
default 50 °F threshold creates large spikes on a few days, while the 60 °F threshold dampens these spikes
and distributes a small amount of emissions to the days that have a minimum temperature between 50 and
60 °F.
RWC temporal profile, Duval County, FL, Jan - Apr
.2 0.025
60F, alternate formula
50F, default formula
5 0.015
Figure 3-9. Example of RWC temporalization using a 50 °F versus 60°F threshold
The diurnal profile for used for most RWC sources places more of the RWC emissions in the morning
and the evening when people are typically using these sources. This profile is based on a 2004 MANE-
VU survey based temporal profiles (see
http://www.marama.org/publications folder/ResWoodCombustion/Final report.pdf). This profile was
created by averaging three indoor and three RWC outdoor temporal profiles from counties in Delaware
and aggregating them into a single RWC diurnal profile. This new profile was compared to a
59
-------
concentration based analysis of aethalometer measurements in Rochester, NY (Wang et al. 2011) for
various seasons and day of the week and found that the new RWC profile generally tracked the
concentration based temporal patterns.
The temporalization for "Outdoor Hydronic Heaters" (i.e.,"OHH", SCC=2104008610), and "Outdoor
wood burning device, NEC (fire-pits, chimneas, etc.)" (i.e., "recreational RWC", SCC=21040087000) do
not use the same meteorological-based temporalization as the rest of the rwc sector, because
meteorological-based temporalization did not agree with observations for how these appliances are used.
For hydronic heaters, the annual-to-month, day-of-week and diurnal profiles were modified based on
information in the New York State Energy Research and Development Authority (NYSERDA)
"Environmental, Energy Market, and Health Characterization of Wood-Fired Hydronic Heater
Technologies, Final Report" (NYSERDA, 2012) as well as a Northeast States for Coordinated Air Use
Management (NESCAUM) report "Assessment of Outdoor Wood-fired Boilers" (NESCAUM, 2006). A
Minnesota 2008 Residential Fuelwood Assessment Survey of individual household responses (MDNR,
2008) provided additional annual-to-month, day-of-week and diurnal activity information for OHH as
well as recreational RWC usage.
The diurnal profile for OHH, shown in Figure 3-10 is based on a conventional single-stage heat load unit
burning red oak in Syracuse, New York. The NESCAUM report describes how for individual units, OHH
are highly variable day-to-day but that in the aggregate, these emissions have no day-of-week variation.
In contrast, the day-of-week profile for recreational RWC follows a typical "recreational" profile with
emissions peaked on weekends. Annual-to-month temporalization for OHH as well as recreational RWC
were computed from the MN DNR survey (MDNR, 2008) and are illustrated in Figure 3-11. OHH
emissions still exhibit strong seasonal variability, but do not drop to zero because many units operate
year-round for water and pool heating. In contrast to all other RWC appliances, recreational RWC
emissions are used far more frequently during the warm season.
The 2017 NEI includes two new hydronic heater SCCs, 2104008620 (indoor hydronic heaters) and
2104008630 (pellet-fired hydronic heaters). Both of these new SCCs use the same monthly, weekly, and
diurnal temporal profiles as OHHs.
50,000
40,000
30,000
20,000
10,000
Heat Load (BTU/hr)
aaaaaaaaaa
C3CjCjCjC3C3C3C3C303
CL(CLiCLiCLiCLiCLiCLiC1iC1iOhOhOh03
Figure 3-10. Diurnal profile for OHH, based on heat load (BTU/hr)
60
-------
Monthly Temporal Activity for OHH & Recreational RWC
Fire Pit/Chimenea
Outdoor Hydronic Heater
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
Figure 311. Annual-to-month temporal profiles for OHH and recreational RWC
For the ag sector, agricultural GenTPRO temporalization was applied to both livestock and fertilizer
emissions, and to all pollutants within the ag sector, not just NH3. This is a change from the 2014v7.0
modeling platform, in which agricultural GenTPRO temporalization was only applied to livestock NH3
sources. The GenTPRO algorithm is based on an equation derived by Jesse Bash of EPA ORD based on
the Zhu, Henze, et al. (2014) empirical equation. This equation is based on observations from the TES
satellite instrument with the GEOS-Chem model and its adjoint to estimate diurnal NH3 emission
variations from livestock as a function of ambient temperature, aerodynamic resistance, and wind speed.
The equations are:
Ea = [161500/T,/; x e(-1380/V] x ARlJl
where
PE;,/; = Euh / Sum(E, /,)
PE;,/; = Percentage of emissions in county i on hour h
Eij, = Emission rate in county i on hour h
Tij, = Ambient temperature (Kelvin) in county i on hour h
Vu, = Wind speed (meter/sec) in county i (minimum wind speed is 0.1 meter/sec)
AR;,/; = Aerodynamic resistance in county i
GenTPRO was run using the "BASHNH3" profile method to create month-to-hour temporal profiles for
these sources. Because these profiles distribute to the hour based on monthly emissions, the monthly
emissions are obtained from a monthly inventory, or from an annual inventory that has been temporalized
to the month. Figure 3-13 compares the daily emissions for Minnesota from the "old" approach (uniform
monthly profile) with the "new" approach (GenTPRO generated month-to-hour profiles). Although the
GenTPRO profiles show daily (and hourly variability), the monthly total emissions are the same between
the two approaches.
61
-------
MN ag NH3 livestock temporal profiles
0.0
1/1/2008 2/1/2008 3/1/2008 4/1/2008 5/1/2008 6/1/2008 7/1/2008 8/1/2008 9/1/2008 10/1/2008 11/1/2008 12/1/2008
-old
-new
Figure 313. Example of animal NH3 emissions temporalization approaches, summed to daily
emissions
For the afdust sector, meteorology is not used in the development of the temporal profiles, but it is used to
reduce the total emissions based on meteorological conditions. These adjustments are applied through
sector-specific scripts, beginning with the application of land use-based gridded transport fractions and
then subsequent zero-outs for hours during which precipitation occurs or there is snow cover on the
ground. The land use data used to reduce the NEI emissions explains the amount of emissions that are
subject to transport. This methodology is discussed in Pouliot, et al., 2010, and in "Fugitive Dust
Modeling for the 2008 Emissions Modeling Platform" (Adelman, 2012). The precipitation adjustment is
applied to remove all emissions for days where measurable rain occurs. Therefore, the afdust emissions
vary day-to-day based on the precipitation and/or snow cover for that grid cell and day. Both the
transport fraction and meteorological adjustments are based on the gridded resolution of the platform;
therefore, somewhat different emissions will result from different grid resolutions. Application of the
transport fraction and meteorological adjustments prevents the overestimation of fugitive dust impacts in
the grid modeling as compared to ambient samples.
Biogenic emissions in the beis sector vary by every day of the year because they are developed using
meteorological data including temperature, surface pressure, and radiation/cloud data. The emissions are
computed using appropriate emission factors according to the vegetation in each model grid cell, while
taking the meteorological data into account.
3.3.5.4 Temporal Profiles for Onroad Mobile Sources
For the onroad sector, the temporal distribution of emissions is a combination of more traditional
temporal profiles and the influence of meteorology. This section discusses both the meteorological
influences and the updates to the diurnal temporal profiles for this platform.
Meteorology is not used in the development of the temporal profiles, but rather it impacts the calculation
of the hourly emissions through the program Movesmrg. The result is that the emissions vary at the
hourly level by grid cell. More specifically, the on-network (RPD) and the off-network parked vehicle
(RPV, RPH, and RPP) processes use the gridded meteorology (MCIP) directly. Movesmrg determines
the temperature for each hour and grid cell and uses that information to select the appropriate emission
factor (EF) for the specified SCC/pollutant/mode combination. RPP uses the gridded minimum and
maximum temperature for the day. The combination of these four processes (RPD, RPV, RPH, and RPP)
is the total onroad sector emissions. The onroad sector show a strong meteorological influence on their
temporal patterns (see the 2014NEIv2 TSD for more details).
62
-------
Figure 3-14 illustrates the difference between temporalization of the onroad sector and the meteorological
influence via SMOKE-MOVES. Similar temporalization is done for the VMT in SMOKE-MOVES, but
the meteorologically varying emission factors add variation on top of the temporalization.
4
2014v2 onroad RPD hourly NOX and VMT: Wake County, NC
_ 3
3.5 -j
s 3
i
2.5
5
_oj 2.5 -j
A A
hJ\
2 g
v\
v\
/-y
\V\
1.5 1
3 VMT
I 15
r
^4
r
5 J
J
1
]
L J
1
^ / \
\
1 O NOX
V /
V
\ /
\ /
v i
0.5
XJ
xj
XJ
0
40:00
7/8/140:00
7/9/140:00
7/10/140:00
7/11/140:00 7/12/140:00 7/13/140:00 7/14/140:00 7/15/1
Date and time (GMT)
Figure 3-14. Example of SMOKE-MOVES temporal variability of NOx emissions versus activity
For the onroad sector, the "inventories" referred to in Table 3-10 actually consist of activity data, not
emissions. For RPP and RPV processes, the VPOP inventory is annual and does not need
temporalization. For RPD, the VMT inventory is annual for some sources and monthly for other sources,
depending on the source of the data. Sources without monthly VMT were temporalized from annual to
month through temporal profiles. VMT data were also temporalized from month to day of the week, and
then to hourly through temporal profiles. The RPD processes use an average speed distribution (SPDIST)
that specifies the amount of time spent in each MOVES speed bin for each county, vehicle (aka source)
type, road type, weekday/weekend, and hour of day. Unlike other sectors, the temporal profiles and
SPDIST will impact not only the distribution of emissions through time but also the total emissions.
Because SMOKE-MOVES (for RPD) calculates emissions from VMT, speed and meteorology, if one
shifted the VMT or speed to different hours, it would align with different temperatures and hence
different emission factors. In other words, two SMOKE-MOVES runs with identical annual VMT,
meteorology, and MOVES emission factors, will have different total emissions if the temporalization of
VMT changes. For RPH, the HOTELING inventory is annual and was temporalized to month, day of the
week, and hour of the day through temporal profiles. This is an analogous process to RPD except that
speed is not included in the calculation of RPH.
For this study, the temporal profiles for the onroad sector come from the 2017 NEI, which is a
compilation of state/local-provided data and nationally available datasets. VMT day-of-week and hour-of-
day temporal profiles were developed for counties across the continental U.S. as part of the effort to
update the inputs to MOVES and SMOKE-MOVES under CRC A-100 (Coordinating Research Council,
2017). CRC A-100 data includes profiles by region or county, road type, and broad vehicle category.
There are three vehicle categories: passenger vehicles (11/21/31), commercial trucks (32/52), and
combination trucks (53/61/62). CRC A-100 does not cover buses, refuse trucks, or motor homes, so those
vehicle types were mapped to other vehicle types for which CRC A-100 did provide profiles, as follows:
1) Intercity/transit buses were mapped to commercial trucks; 2) Motor homes were mapped to passenger
vehicles for day-of-week and commercial trucks for hour-of-day; 3) School buses and refuse trucks were
mapped to commercial trucks for hour-of-day and use a new custom day-of-week profile called
63
-------
LOWSATSUN that has a very low weekend allocation, since school buses and refuse trucks operate
primarily on business days. In addition to temporal profiles, CRC A-100 data were also used to develop
the average speed distribution data (SPDIST) used by SMOKE-MOVES. In areas where state-provided
data and CRC A-100 data does not exist, hourly speed data is based on MOVES county databases.
The CRC A-100 dataset includes temporal profiles for individual counties, Metropolitan Statistical Areas
(MSAs), and entire regions (e.g. West, South). Counties without temporal profiles specific to itself, or to
its MSA, are assigned to regional temporal profiles. Temporal profiles also vary between MOVES road
types, and there are distinct hour-of-day profiles for each day of the week. Plots of hour-of-day profiles
for passenger vehicles in Fulton County, GA, are shown in Figure 3-15. Separate plots are shown for
Monday, Friday, Saturday, and Sunday, and each line corresponds to a particular MOVES road type (e.g.
road type 2 = rural restricted). Figure 3-16 shows which counties have temporal profiles specific to that
county, and which counties use regional average profiles in the CRC A-100 data.
Monday
Fulton Co
passenger
Friday
Fulton Co
passenger
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Saturday
Fulton Co
passenger
Sunday
Fulton Co
passenger
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
road 2 road 3 road 4 road 5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
road 2 road 3 road 4 road 5
Figure 3-15. Sample onroad diurnal profiles for Fulton County, GA
64
-------
Group I I Individual
I I Midwest Region Average of Single County MSA Counties
I I Midwest Region non-MSA Average
I I Northeast Region Average of Single County MSA Counties
I I Northeast Region non-MSA Average
I I South Region Average of Single County MSA Counties
I I South Region non-MSA Average
I I West Region Average of Single County MSA Counties
I I West Region non-MSA Average
Midwest Region Average of Core Counties inside MSAs
Midwest Region Average of non-Core Counties inside MSAs
I Northeast Region Average of Core Counties inside MSAs
I Northeast Region Average of non-Core Counties inside MSAs
I South Region Average of Core Counties inside MSAs
I South Region Average of non-Core Counties inside MSAs
I West Region Average of Core Counties inside MSAs
I West Region Average of non-Core Counties inside MSAs
Figure 3-16. Counties for which MOVES Speeds and Temporal Profiles could be Populated
For hoteling, day-of-week profiles are the same as non-hoteling for combination trucks, while hour-of-day
non-hoteling profiles for combination trucks were inverted to create new hoteling profiles that peak
overnight instead of during the day.
State/local-provided data for the 2017 NEI were accepted for use in the 2017 NEI if they were deemed to
be at least as credible as the CRC A-100 data. The 2017 NEI TSD includes more details on which data
were used for which counties. In areas of the contiguous United States where state/local-provided data
were not provided or deemed unacceptable, the CRC A-100 temporal profiles were used, except in
California. All California temporal profiles were carried over from the 2014v7.1 platform, although
California hoteling uses CRC A-100-based profiles just like the rest of the country, since CARB didn't
have a hoteling-specific profile. Monthly profiles in all states (national profiles by broad vehicle type)
were also carried over from 2014vl and applied directly to the VMT. For California, CARB supplied
diurnal profiles that varied by vehicle type, day of the week1 \ and air basin. These C ARB-specific
profiles were used in developing EPA estimates for California. Although the EPA adjusted the total
11 California's diurnal profiles varied within the week. Monday, Friday, Saturday, and Sunday had unique profiles and
Tuesday, Wednesday, Thursday had the same profile.
65
-------
emissions to match interpolated 2016 levels based on California's submitted inventories for 2014 and
2017, the temporalization of these emissions took into account both the state-specific VMT profiles and
the SMOKE-MOVES process of incorporating meteorology. For more details on the adjustments to
California's onroad emissions, see section 3.2.5.1.
3.3.6 Vertical A llocation of Emissions
Table 3-11 specifies the sectors for which plume rise is calculated. If there is no plume rise for a sector, the
emissions are placed into layer 1 of the air quality model. Vertical plume rise was performed in-line within
CMAQ for all of the SMOKE point-source sectors (i .e., ptegu, ptnonipm, pt oilgas, ptfire, ptagfire,
ptfireothna, othpt, and cmv_c3). The in-line plume rise computed within CMAQ is nearly identical to the
plume rise that would be calculated within SMOKE using the Lay point program. The selection of point
sources for plume rise is pre-determined in SMOKE using the Elevpoint program. The calculation is done
in conjunction with the CMAQ model time steps with interpolated meteorological data and is therefore
more temporally resolved than when it is done in SMOKE. Also, the calculation of the location of the
point sources is slightly different than the one used in SMOKE and this can result in slightly different
placement of point sources near grid cell boundaries.
For point sources, the stack parameters are used as inputs to the Briggs algorithm, but point fires
do not have traditional stack parameters. However, the ptfire, ptagfire, and ptfire_othna inventories do contain
data on the acres burned (acres per day) and fuel consumption (tons fuel per acre) for each day. CMAQ
uses these additional parameters to estimate the plume rise of emissions into layers above the surface
model layer. Specifically, these data are used to calculate heat flux, which is then used to estimate plume
rise. In addition to the acres burned and fuel consumption, heat content of the fuel is needed to compute
heat flux. The heat content was assumed to be 8000 Btu/lb of fuel for all fires because specific data on the
fuels were unavailable in the inventory. The plume rise algorithm applied to the fires is a modification of
the Briggs algorithm with a stack height of zero.
CMAQ uses the Briggs algorithm to determine the plume top and bottom, and then computes the plumes"
distributions into the vertical layers that the plumes intersect. The pressure difference across each layer
divided by the pressure difference across the entire plume is used as a weighting factor to assign the
emissions to layers. This approach gives plume fractions by layer and source.
3.3.7 Emissions Modeling Spatial Allocation
The methods used to perform spatial allocation are summarized in this section. For the modeling
platform, spatial factors are typically applied by county and SCC. Spatial allocation was performed for a
national 12-km domain. To accomplish this, SMOKE used national 12-km spatial surrogates and a
SMOKE area-to-point data file. For the U.S., EPA updated surrogates to use circa 2010-2014 data
wherever possible. For Mexico, updated spatial surrogates were used as described below. For Canada,
shapefiles for generating new surrogates were provided by ECCC for use with their 2015 inventories.
The U.S., Mexican, and Canadian 12-km surrogates cover the entire CONUS domain 12US1 shown in
Figure 3-3.
3.3.7.1 Surrogates for U.S. Emissions
There are more than 70 spatial surrogates available for spatially allocating U.S. county-level emissions to
the 12-km grid cells used by the air quality model. Note that an area-to-point approach overrides the use
of surrogates for a limited set of sources. Table 3-11 lists the codes and descriptions of the surrogates.
66
-------
Surrogate names and codes listed in italics are not directly assigned to any sources for this platform, but
they are sometimes used to gapfill other surrogates, or as an input for merging two surrogates to create a
new surrogate that is used.
Many surrogates were updated or newly developed for use in the 2017 platform using recently available
data sets. They include the use of the 2011 National Land Cover Database (the previous platform used
2006) and development of various development density levels such as open, low, medium high and
various combinations of these. These landuse surrogates largely replaced the FEMA category surrogates
that were used in the 2011 platform. Additionally, onroad surrogates were developed using average
annual daily traffic counts from the highway monitoring performance system (HPMS). Onroad
surrogates for this platform do not distinguish between urban and rural road types, which prevents issues
in areas where there are inconsistent urban and rural definitions between MOVES and the surrogate data.
Surrogate updates made for this 2017 study include:
- New spatial surrogates for the np oilgas sector were developed based on known locations of oil
and gas activity for 2017.
- New onroad surrogates were generated to incorporate 2017 Average Annual Daily Traffic
(AADT).
The surrogates for the U.S. were mostly generated using the Surrogate Tool to drive the Spatial Allocator,
but a few surrogates were developed directly within ArcGIS or using scripts that manipulate spatial data
in PostgreSQL. The tool and documentation for the Surrogate Tool is available at
https://www.cmascenter.Org/sa-tools/documentation/4.2/SurrogateToolUserGuide 4 2.pdf.
Table 3-11. U.S. Surrogates available for the 2017 modeling platform
Code
Surrogate Description
Code
Surrogate Description
N/A
Area-to-point approach (see 3.6.2)
650
Refineries and Tank Farms
100
Population
670
Spud Count - CBM Wells
110
Housing ¦
671
Spud Count - Gas Wells
150
Residential Heating - Natural Gas
672
Gas Production at Oil Wells
170
Residential Heating - Distillate Oil
673
Oil Production at CBM Wells
180
Residential Heating - Coal
674
Unconventional Well Completion Counts
190
Residential Heating - LP Gas
676
Well Count - All Producing
205
Extended Idle Locations
677
Well Count - All Exploratory
239
Total Road AADT
678
Completions at Gas Wells
240
Total Road Miles
679
Completions at CBM Wells
242
All Restricted AADT
681
Spud Count - Oil Wells
244
All Unrestricted AADT
683
Produced Water at All Wells
258
Intercity Bus Terminals
6831
Produced Water at CBM Wells
259
Transit Bus Terminals
6832
Produced Water at Gas Wells
260
Total Railroad Miles ]
6833
Produced Water at Oil Wells
261
NT AD Total Railroad Density
685
Completions at Oil Wells
271
NT AD Class 12 3 Railroad Density
686
Completions at All Wells
300
NLCD Low Intensity Development
687
Feet Drilled at All Wells
304
NLCD Open + Low
691
Well Counts - CBM Wells
305
NLCD Low + Med
692
Spud Count - All Wells
306
NLCD Med + High
693
Well Count - All Wells
67
-------
Code
Surrogate Description "
1 Code
Surrogate Description
307
NLCD All Development i
694
Oil Production at Oil Wells
308
NLCD Low + Med + High
695
Well Count - Oil Wells
309
NLCD Open + Low + Med
696
Gas Production at Gas Wells
310
NLCD Total Agriculture j
697
Oil Production at Gas Wells
319
NLCD Crop Land
698
Well Count - Gas Wells
320
NLCD Forest Land
699
Gas Production at CBM Wells
321
NLCD Recreational Land
711
Airport Areas
340
s
NLCD Land ]
801
Port Areas
350
NLCD Water
805
Offshore Shipping Area
500
Commercial Land '
806
Offshore Shipping NE12014 Activity
505
Industrial Land
807
Navigable Waterway Miles
506
Education
808
2013 Shipping Density
510
Commercial plus Industrial
820
Ports NEI2014 Activity
535
Residential + Commercial + Industrial +
Institutional + Government
850
Golf Courses
560
Hospital (COM6)
860
Mines
For the onroad sector, the on-network (RPD) emissions were allocated differently from the off-network
(RPP and RPV). On-network used average annual daily traffic (AADT) data and off network used land
use surrogates as shown in Table 3-12. Emissions from the extended (i.e., overnight) idling of trucks
were assigned to surrogate 205 that is based on locations of overnight truck parking spaces. The
underlying data in this surrogate was updated for use in the 2016 platform to include additional data
sources and corrections based on comments received.
Table 3-12. Off-Network Mobile Source Surrogates
Source type
Source Type name
Surrogate ID
Description
11
Motorcycle
307
NLCD All Development
21
Passenger Car
307
NLCD All Development
31
Passenger Truck
307
NLCD All Development
NLCD Low + Med +
32
Light Commercial Truck
308
High
41
Intercity Bus
258
Intercity Bus Terminals
42
Transit Bus
259
Transit Bus Terminals
43
School Bus
506
Education
51
Refuse Truck
306
NLCD Med + High
52
Single Unit Short-haul Truck
306
NLCD Med + High
53
Single Unit Long-haul Truck
306
NLCD Med + High
54
Motor Home
304
NLCD Open + Low
61
Combination Short-haul Truck
306
NLCD Med + High
62
Combination Long-haul Truck
306
NLCD Med + High
For the oil and gas sources in the np oilgas sector, the spatial surrogates were updated to those shown in
Table 3-13 using 2017 data consistent with what was used to develop the 2017 nonpoint oil and gas
emissions. The exploration and production of oil and gas has increased in terms of quantities and
locations over the last seven years, primarily through the use of new technologies, such as hydraulic
68
-------
fracturing. ERG prepared census-tract, 2-km, and 4-km sub-county surrogate factors for use in 2017
emissions modeling. A technical memo dated by ERG provides technical details of how the gridding
surrogates were generated.
Spatial allocation of npoilgas emissions to the national 36km and 12km domains used for air quality
modeling is accomplished using the spatial surrogates described in ERG's technical memo. All spatial
surrogates for np oilgas are developed based on known locations of oil and gas activity for year 2017.
These spatial surrogates, numbered 670 through 699 and also including 6831, 6832, and 6833, were
originally processed at 4km resolution and without gapfilling. The surrogates were first gapfilled using
fallback surrogates. For each surrogate, the last two fallbacks were surrogate 693 (Well Count - All
Wells) and 304 (NLCD Open + Low). Where appropriate, other surrogates were also part of the gapfilling
procedure. For example, surrogate 670 (Spud Count - CBM Wells) was first gapfilled with 692 (Spud
Count - All Wells), and then 693 and finally 304. After gapfilling, surrogates were aggregated to 12km
and 36km resolution. All gapfilling and aggregating was performed with the Surrogate Tool.
Table 3-13. Spatial Surrogates for Oil and Gas Sources
Surrogate Code
Surrogate Description
670
Spud Count - CBM Wells
671
Spud Count - Gas Wells
672
Gas Production at Oil Wells
673
Oil Production at CBM Wells
674
Unconventional Well Completion Counts
676
Well Count - All Producing
677
Well Count - All Exploratory
678
Completions at Gas Wells
679
Completions at CBM Wells
681
Spud Count - Oil Wells
683
Produced Water at All Wells
6831
Produced Water at CBM Wells
6832
Produced Water at Gas Wells
6833
Produced Water at Oil Wells
685
Completions at Oil Wells
686
Completions at All Wells
687
Feet Drilled at All Wells
691
Well Counts - CBM Wells
692
Spud Count - All Wells
693
Well Count - All Wells
694
Oil Production at Oil Wells
695
Well Count - Oil Wells
696
Gas Production at Gas Wells
697
Oil Production at Gas Wells
698
Well Count - Gas Wells
69
-------
Surrogate Code
Surrogate Description
699
Gas Production at CBM Wells
Not all of the available surrogates are used to spatially allocate sources in the modeling platform; that is,
some surrogates shown in Table 3-11. U.S. Surrogates available for the 2017 modeling platform 3-11
were not assigned to any SCCs, although many of the "unused" surrogates are actually used to "gap fill"
other surrogates that are used. When the source data for a surrogate has no values for a particular county,
gap filling is used to provide values for the surrogate in those counties to ensure that no emissions are
dropped when the spatial surrogates are applied to the emission inventories. The U.S. CAP emissions
allocated to the various spatial surrogates are shown in Table 3-14.
Table 314. Selected 2016 CAP emissions by sector for U.S. Surrogates (CONUS domain totals)
Sector
ID
Description
MI;
NOx
PM2S
SO2
voc
afdust
240
Total Road Miles
309,376
afdust
304
NLCD Open + Low
842,116
afdust
306
NLCD Med + High
52,278
afdust
308
NLCD Low + Med + High
117,313
afdust
310
NLCD Total Agriculture
791,881
ag
310
NLCD Total Agriculture
3,457,964
227,853
nonpt
100
Population
34,304
0
0
0
1,181,307
nonpt
150
Residential Heating - Natural Gas
33,550
204,371
4,041
1,365
12,055
nonpt
170
Residential Heating - Distillate Oil
1,531
30,031
3,284
11,510
1,039
nonpt
180
Residential Heating - Coal
1
3
1
3
3
nonpt
190
Residential Heating - LP Gas
98
31,061
163
712
1,181
nonpt
239
Total Road AADT
0
22
541
0
297,798
nonpt
240
Total Road Miles
0
0
0
0
39,013
nonpt
244
All Unrestricted AADT
0
0
0
0
101,255
nonpt
271
NT AD Class 12 3 Railroad Density
0
0
0
0
2,203
nonpt
300
NLCD Low Intensity Development
4,823
19,093
94,548
2,882
72,599
nonpt
306
NLCD Med + High
23,713
274,780
246,324
131,747
878,553
nonpt
307
NLCD All Development
109
25,803
110,507
8,260
574,679
nonpt
308
NLCD Low + Med + High
886
156,239
15,825
10,081
65,798
nonpt
310
NLCD Total Agriculture
0
0
38
0
227,051
nonpt
319
NLCD Crop Land
0
0
97
72
299
nonpt
320
NLCD Forest Land
3,953
68
273
0
279
nonpt
505
Industrial Land
0
0
0
0
48
nonpt
535
Residential + Commercial + Industrial +
Institutional + Government
0
2
121
0
36
nonpt
650
Refineries and Tank Farms
0
16
0
0
106,401
nonpt
711
Airport Areas
0
0
0
0
596
nonpt
801
Port Areas
0
0
0
0
6,730
nonroad
261
NT AD Total Railroad Density
3
2,026
212
2
398
nonroad
304
NLCD Open + Low
4
1,763
152
5
2,598
70
-------
-1
,457
,482
,196
,340
,270
,536
,824
,274
,701
,617
493
152
,761
,626
,260
,743
484
,490
868
,265
,639
,937
34
2
,175
,384
,234
,817
,440
,090
,531
,360
846
,078
,774
,892
,414
67
195
,411
,343
,858
ID
Description
MI;
NOx
305
NLCD Low + Med
94
15,378
306
NLCD Med + High
315
173,548
307
NLCD All Development
100
30,701
308
NLCD Low + Med + High
532
315,346
309
NLCD Open + Low + Med
120
21,204
310
NLCD Total Agriculture
419
353,338
320
NLCD Forest Land
15
4,767
321
NLCD Recreational Land
83
12,197
350
NLCD Water
189
115,331
850
Golf Courses
13
2,039
860
Mines
2,579
670
Spud Count - CBM Wells
0
671
Spud Count - Gas Wells
672
Gas Production - Oil Wells
511
674
Unconventional Well Completion Counts
20,367
678
Completions at Gas Wells
7,843
679
Completions at CBM Wells
681
Spud Count - Oil Wells
683
Produced Water at All Wells
22
685
Completions at Oil Wells
825
687
Feet Drilled at All Wells
37,669
691
Well Counts - CBM Wells
19,926
692
Spud Count - All Wells
365
693
Well Count - All Wells
0
694
Oil Production at Oil Wells
4,193
695
Well Count - Oil Wells
141,275
696
Gas Production at Gas Wells
40,880
697
Oil Production - Gas Wells
858
698
Well Count - Gas Wells
324,370
699
Gas Production at CBM Wells
33
6831
Produced Water at CBM Wells
6832
Produced Water at Gas Wells
6833
Produced Water at Oil Wells
0
205
Extended Idle Locations
251
89,719
239
Total Road AADT
242
All Restricted AADT
34,371
1,093,396
244
All Unrestricted AADT
64,963
1,701,659
258
Intercity Bus Terminals
195
259
Transit Bus Terminals
77
304
NLCD Open + Low
1,149
306
NLCD Med + High
13,678
307
NLCD All Development
525,284
71
-------
Sector
ID
Description
MI;
NOx
PM2S
SO2
voc
onroad
308
NLCD Low + Med + High
36,174
633
51
56,651
onroad
506
Education
599
14
1
583
rail
261
NT AD Total Railroad Density
14
34,523
1,022
30
1,755
rail
271
NT AD Class 12 3 Railroad Density
328
522,862
14,493
652
23,949
rwc
300
NLCD Low Intensity Development
16,409
34,095
299,280
7,989
323,683
3.3.7.2 Allocation Methodfor Airport-Related Sources in the U.S.
There are numerous airport-related emission sources in the NEI, such as aircraft, airport ground support
equipment, and jet refueling. The modeling platform includes the aircraft and airport ground support
equipment emissions as point sources. For the modeling platform, EPA used the SMOKE "area-to-point"
approach for only jet refueling in the nonpt sector. The following SCCs use this approach: 2501080050
and 2501080100 (petroleum storage at airports), and 2810040000 (aircraft/rocket engine firing and
testing). The ARTOPNT approach is described in detail in the 2002 platform documentation:
https://www3.epa.gov/scram001/reports/Emissions%20TSD%20Voll 02-28-08.pdf. The ARTOPNT file
that lists the nonpoint sources to locate using point data were unchanged from the 2005-based platform.
3.3.7.3 Surrogates for Canada and Mexico Emission Inventories
The surrogates for Canada to spatially allocate the Canadian emissions are based on the 2015 Canadian
inventories and associated data. The spatial surrogate data came from ECCC, along with cross references.
The shapefiles they provided were used in the Surrogate Tool (previously referenced) to create spatial
surrogates. The Canadian surrogates used for this platform are listed in Table 3-15. The population
surrogate was updated for Mexico for the 2014v7.1 platform. Surrogate code 11, which uses 2015
population data at 1 km resolution, replaces the previous population surrogate code 10. The other
surrogates for Mexico are circa 1999 and 2000 and were based on data obtained from the Sistema
Municpal de Bases de Datos (SIMBAD) de INEGI and the Bases de datos del Censo Economico 1999.
Most of the CAPs allocated to the Mexico and Canada surrogates are shown in Table 3-16. The entries in
Table 3-16 are for the othar, othafdust, onroad can, and onroadmex sectors.
Table 315. Canadian Spatial Surrogates
Code
Canadian Surrogate Description
Code
Description
100
Population
921
Commercial Fuel Combustion
TOTAL INSTITUTIONAL AND
101
total dwelling
923
GOVERNEMNT
104
capped total dwelling
924
Primary Industry
106
ALL INDUST
925
Manufacturing and Assembly
113
Forestry and logging
926
Distribtution and Retail (no petroleum)
200
Urban Primary Road Miles
927
Commercial Services
210
Rural Primary Road Miles
932
CANRAIL
211
Oil and Gas Extraction
940
PAVED ROADS NEW
212
Mining except oil and gas
946
Construction and mining
220
Urban Secondary Road Miles
948
Forest
221
Total Mining
951
Wood Consumption Percentage
222
Utilities
955
UNPAVED ROADS AND TRAILS
72
-------
Code
Canadian Surrogate Description
Code
Description
230
Rural Secondary Road Miles
960
TOTBEEF
233
Total Land Development
970
TOTPOUL
240
capped population
980
TOTSWIN
308
Food manufacturing
990
TOTFERT
321
Wood product manufacturing
996
urban area
323
Printing and related support activities
1251
OFFR TOTFERT
Petroleum and coal products
324
manufacturing
1252
OFFR MINES
Plastics and rubber products
326
manufacturing
1253
OFFR Other Construction not Urban
Non-metallic mineral product
327
manufacturing
1254
OFFR Commercial Services
331
Primary Metal Manufacturing
1255
OFFR Oil Sands Mines
350
Water
1256
OFFR Wood industries CANVEC
412
Petroleum product wholesaler-
distributors
1257
OFFR UNPAVED ROADS RURAL
448
clothing and clothing accessories
stores
1258
OFFR Utilities
482
Rail transportation
1259
OFFR total dwelling
Waste management and remediation
562
services
1260
OFFR water
901
AIRPORT
1261
OFFR ALL INDUST
902
Military LTO
1262
OFFR Oil and Gas Extraction
903
Commercial LTO
1263
OFFR ALLROADS
904
General Aviation LTO
1265
OFFR CANRAIL
945
Commercial Marine Vessels
9450
Commercial Marine Vessel Ports
Table 316. 2017 CAPs Allocated to Mexican and Canadian Spatial Surrogates for 12US2
Code
Mexican or Canadian Surrogate
Description
MI;
NOx
PM25
SO2
voc
11
MEX 2015 Population
0
75,091
454
151
204,890
14
MEX Residential Heating - Wood
0
1,942
5,445
162
14,562
16
MEX Residential Heating - Distillate
Oil
1
25
0
0
1
22
MEX Total Road Miles
2,493
316,652
10,056
5,277
63,239
24
MEX Total Railroads Miles
0
20,893
467
183
816
26
MEX Total Agriculture
95,034
17,101
13,147
412
3,565
32
MEX Commercial Land
0
68
1,478
0
24,996
34
MEX Industrial Land
79
2,016
1,173
5
31,706
36
MEX Commercial plus Industrial
Land
4
6,464
327
12
92,441
40
MEX Residential (RES1-
4)+Comercial+Industrial+Institutional
+Government
0
15
66
1
18,980
73
-------
Mexican or Canadian Surrogate
Code
Description
MI;
NOx
PM2s
SO2
voc
42
MEX Personal Repair (COM3)
0
0
0
0
4,576
44
MEX Airports Area
0
3,445
51
224
1,343
48
MEX Brick Kilns
0
192
3,849
349
94
MEX Mobile sources - Border
50
Crossing
3
71
2
0
57
100
CAN Population
611
42
515
12
175
101
CAN total dwelling
0
0
0
0
115,985
104
CAN capped total dwelling
264
26,357
1,986
148
1,394
106
CAN ALL INDUST
0
0
2,824
0
0
113
CAN Forestry and logging
81
962
5,294
20
2,715
200
CAN Urban Primary Road Miles
1,162
49,767
1,767
219
5,210
210
CAN Rural Primary Road Miles
457
28,045
964
88
2,223
211
CAN Oil and Gas Extraction
0
30
26
8
366
212
CAN Mining except oil and gas
0
0
2,369
0
0
220
CAN Urban Secondary Road Miles
2,151
77,473
3,543
455
13,514
221
CAN Total Mining
0
0
32,447
0
0
222
CAN Utilities
27
1,419
12,364
346
16
230
CAN Rural Secondary Road Miles
1,236
54,510
1,953
243
6,198
240
CAN capped population
28
34,842
838
50
64,072
308
CAN Food manufacturing
0
0
15,711
0
9,438
321
CAN Wood product manufacturing
493
3,074
1,142
148
10,455
323
CAN Printing and related support
activities
0
0
0
0
10,199
CAN Petroleum and coal products
324
manufacturing
0
615
779
298
2,906
326
CAN Plastics and rubber products
manufacturing
0
0
0
0
15,320
CAN Non-metallic mineral product
327
manufacturing
0
0
5,302
0
0
331
CAN Primary Metal Manufacturing
0
145
5,442
29
72
CAN Petroleum product wholesaler-
412
distributors
0
0
0
0
35,673
448
CAN clothing and clothing accessories
stores
0
0
0
0
117
482
CAN Rail transportation
1
2,438
52
7
148
562
CAN Waste management and
remediation services
224
1,622
1,896
2,411
10,118
901
CAN AIRPORT
0
75
7
0
7
921
CAN Commercial Fuel Combustion
157
16,140
1,674
1,053
728
CAN TOTAL INSTITUTIONAL
923
AND GOVERNEMNT
0
0
0
0
11,402
924
CAN Primary Industry
0
0
0
0
28,417
925
CAN Manufacturing and Assembly
0
0
0
0
57,090
74
-------
Code
Mexican or Canadian Surrogate
Description
MI;
NOx
PM2s
SO2
voc
926
CAN Distribution and Retail (no
petroleum)
0
0
0
0
5,704
927
CAN Commercial Services
0
0
0
0
24,917
932
CAN CANRAIL
33
62,174
1,436
260
2,972
940
CAN PAVED ROADS NEW
0
0
141,178
0
0
946
CAN Construction and mining
0
0
0
0
2,676
951
CAN Wood Consumption Percentage
1,459
16,211
133,259
2,316
188,165
955
CAN
UNPAVED ROADS AND TRAILS
0
0
212,342
0
0
960
CAN TOTBEEF
0
0
792
0
0
970
CAN TOTPOUL
0
0
156
0
0
980
CAN TOTSWIN
0
0
682
0
0
990
CAN TOTFERT
33
3,199
226
6,464
120
996
CAN urban area
0
0
326
0
0
1251
CAN OFFR TOTFERT
48
42,015
3,041
34
4,082
1252
CAN OFFR MINES
1
627
46
1
95
1253
CAN OFFR Other Construction not
Urban
44
32,470
3,616
31
7,558
1254
CAN OFFR Commercial Services
32
13,624
1,990
27
33,516
1255
OFFR Oil Sands Mines
0
0
0
0
0
1256
CAN OFFR Wood industries
CANVEC
6
3,648
293
4
895
1257
CAN OFFR UNPAVED ROADS
RURAL
20
7,359
702
17
28,618
1258
CAN OFFR Utilities
6
3,876
255
5
833
1259
CAN OFFR total dwelling
13
4,585
582
11
11,645
1260
CAN OFFR water
3
785
86
4
5,445
1261
CAN OFFR ALL INDUST
3
4,823
211
2
898
1262
CAN OFFR Oil and Gas Extraction
0
56
12
0
78
1263
CAN OFFR ALLROADS
2
1,706
168
2
396
1265
CAN OFFR CANRAIL
0
69
7
0
12
75
-------
3.4 Emissions References
Adelman, Z. 2012. Memorandum: Fugitive Dust Modeling for the 2008 Emissions Modeling Platform.
UNC Institute for the Environment, Chapel Hill, NC. September, 28, 2012.
Adelman, Z., M. Omary, Q. He, J. Zhao and D. Yang, J. Boylan, 2012. "A Detailed Approach for
Improving Continuous Emissions Monitoring Data for Regulatory Air Quality Modeling."
Presented at the 2012 International Emission Inventory Conference, Tampa, Florida. Available
from https://www3.epa.gOv/ttn/chief/conference/ei20/index.html#ses-5.
Anderson, G.K.; Sandberg, D.V; Norheim, R.A., 2004. Fire Emission Production Simulator (FEPS)
User's Guide. Available at http://www.fs.fed.us/pnw/fera/feps/FEPS users guide.pdf.
2000. "Risk Reduction Plan to Reduce Particulate Matter Emissions from Diesel-Fueled Engines
and Vehicles". California Environmental Protection Agency Air Resources Board, Mobile Source
Control Division, Sacramento, CA. October, 2000. Available at:
http://www.arb.ca.gov/diesel/documents/rrpFinal.pdf.
2007. "Proposed Regulation for In-Use Off-Road Diesel Vehicles". California Environmental
Protection Agency Air Resources Board, Mobile Source Control Division, Sacramento, CA.
April, 2007. Available at: http://www.arb.ca.gov/regact/2007/ordiesl07/isor.pdf.
2010a. "Proposed Amendments to the Regulation for In-Use Off-Road Diesel-Fueled Fleets and
the Off-Road Large Spark-Ignition Fleet Requirements". California Environmental Protection
Agency Air Resources Board, Mobile Source Control Division, Sacramento, CA. October, 2010.
Available at: http://www.arb.ca.gov/regact/2010/offroadlsil0/offroadisor.pdf.
2010b. "Estimate of Premature Deaths Associated with Fine Particle Pollution (PM2.5) in
California Using a U.S. Environmental Protection Agency Methodology". California
Environmental Protection Agency Air Resources Board, Mobile Source Control Division,
Sacramento, CA. August, 2010. Available at: http://www.arb.ca.gov/research/health/pm-
mort/pm-report_2010.pdf. Adelman, Z. 2012. Memorandum: Fugitive Dust Modeling for the
2008 Emissions Modeling Platform. UNC Institute for the Environment, Chapel Hill, NC.
September, 28, 2012.
J.O., Baker, K.R., Beaver, M.R., Park, J.-H., Goldstein, A.H., 2015. Evaluation of improved land
use and canopy representation in BEIS with biogenic VOC measurements in California (in
preparation).
Bullock Jr., R, and K. A. Brehme (2002) "Atmospheric mercury simulation using the CMAQ model:
formulation description and analysis of wet deposition results." Atmospheric Environment 36, pp
2135-2146.
CRC, 2017. Improvement of Default Inputs for MOVES and SMOKE-MOVES: Final Report.
Coordinating Research Council Report No. A-100. Available from
http://crcsite.wpengine.eom/wp-content/uploads/2019/05/CRC-Project-A-115-Final-
Report_20190411.pdf
Environ Corp. 2008. Emission Profiles for EPA SPECIATE Database, Part 2: EPAct Fuels (Evaporative
Emissions). Prepared for U. S. EPA, Office of Transportation and Air Quality, September 30,
2008.
EPA, 2005. EPA 's National Inventory Model (NMIM), A Consolidated Emissions Modeling System for
76
ARB,
ARB,
ARB,
ARB,
Bash,
-------
M0BILE6 andNONROAD, U.S. Environmental Protection Agency, Office of Transportation and
Air Quality, Ann Arbor, MI 48105, EPA420-R-05-024, December 2005. Available at
HYPERLINK
"https://nepis. epa.gov/Exe/ZyPDF. cgi?Dockey=P 10023FZ.pdf'https://nepis.epa.gOv/Exe/ZyPDF.c
gi?Dockey=P 10023FZ .pdf.
EPA 2006a. SPECIATE 4.0, Speciation Database Development Document, Final Report, U.S.
Environmental Protection Agency, Office of Research and Development, National Risk
Management Research Laboratory, Research Triangle Park, NC 27711, EPA600-R-06-161,
February 2006. Available at https://www.epa.gov/air-emissions-modeling/speciate.
EPA, 2015a. 2011 Technical Support Document (TSD) Preparation of Emissions Inventories for the
Version 6.2, 2011 Emissions Modeling Platform. Office of Air Quality Planning and Standards,
Air Quality Assessment Division, Research Triangle Park, NC. Available at
https://www.epa.gov/air-emissions-modeling/2011-version-62-platform.
EPA, 2016b. SPECIATE Version 4.5 Database Development Documentation, U.S. Environmental
Protection Agency, Office of Research and Development, National Risk Management Research
Laboratory, Research Triangle Park, NC 27711, EPA/600/R-16/294, September 2016. Available
at https://www.epa.gov/sites/production/files/2016-Q9/documents/speciate 4.5.pdf.
EPA, 2018. Speciation of Total Organic Gas and Particulate Matter Emissions from On-road Vehicles in
MOVES2015b. Office of Transportation and Air Quality. Report #EPA-420-R-012, July, 2018.
Available from https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100UXME.pdf.
EPA, 2019. Technical Support Document (TSD) Preparation of Emissions Inventories for the Version 7.1,
2016 Emissions Modeling Platform for the 2014 National Air Toxics Assessment. Office of Air
Quality Planning and Standards, Air Quality Assessment Division, Research Triangle Park, NC.
Available at https://www.epa.gov/air-emissions-modeling/2016-version-71-technical-support-
document.
EPA, 2020a. April 2020 release of the 2017 National Emissions Inventory (NEI) Technical Support
Document (TSD). Office of Air Quality Planning and Standards, Air Quality Assessment
Division, Research Triangle Park, NC. Available at https://www.epa.gov/air-emissions-
inventories/2017-national-emissions-inventory-nei-technical-support-document-tsd.
EPA, 2020b. Technical Support Document (TSD) Preparation of Emissions Inventories for the 2016vl
North American Emissions Modeling Platform Office of Air Quality Planning and Standards, Air
Quality Assessment Division, Research Triangle Park, NC. Available from
https://www.epa.gov/air-emissions-modeling/2016vl-platform.
ERG, 2014a. Develop Mexico Future Year Emissions Final Report. Available at
ftp://newftp.epa.gov/air/emismod/201 l/v2platform/2011 emissions/Mexico Emissions WA%204-
09 final report 121814.pdf.
ERG, 2014b. "Technical Memorandum: Modeling Allocation Factors for the 2011 NEI".
ERG, 2016a. "Technical Report: Development of Mexico Emission Inventories for the 2014 Modeling
Platform."
ERG, 2016b. "Technical Memorandum: Modeling Allocation Factors for the 2014 Oil and Gas Nonpoint
Tool."
77
-------
Frost & Sullivan, 2010. "Project: Market Research and Report on North American Residential Wood
Heaters, Fireplaces, and Hearth Heating Products Market (P.O. # PO1-IMP403-F&S). Final
Report April 26, 2010". Prepared by Frost & Sullivan, Mountain View, CA 94041.
Joint Fire Science Program, 2009. Consume 3.0a software tool for computing fuel consumption. Fire
Science Brief. 66, June 2009. Consume 3.0 is available at:
http://www.fs.fed.us/pnw/fera/research/smoke/consume/index.shtml.
McCarty, J.L., Korontzi, S., Jutice, C.O., and T. Loboda. 2009. The spatial and temporal distribution of
crop residue burning in the contiguous United States. Science of the Total Environment, 407 (21):
5701-5712.
McKenzie, D.; Raymond, C.L.; Kellogg, L.-K.B.; Norheim, R.A; Andreu, A.G.; Bayard, A.C.; Kopper,
K.E.; Elman. E. 2007. Mapping fuels at multiple scales: landscape application of the Fuel
Characteristic Classification System. Canadian Journal of Forest Research. 37:2421-2437.
McQuilling, A. M. & Adams, P. J. Semi-empirical process-based models for ammonia emissions from
beef, swine, and poultry operations in the United States. Atmos. Environ. 120, 127-136 (2015).
NCAR, 2016. FIRE EMISSION FACTORS AND EMISSION INVENTORIES, FINN Data, downloaded
2014 SAPRC99 version from
https://www.fs.fed.us/pnw/fera/research/smoke/consume/index.shtml.
NYSERDA, 2012; "Environmental, Energy Market, and Health Characterization of Wood-Fired
Hydronic Heater Technologies, Final Report". New York State Energy Research and
Development Authority (NYSERDA). Available from:
http://www.nvserda.ny.gov/Publications/Case-Studies/-
/media/Files/Publications/Research/Environmental/Wood-Fired-Hvdronic-Heater-Tech.ashx.
Ottmar, R.D.; Sandberg, D.V.; Riccardi, C.L.; Prichard, S.J. 2007. An Overview of the Fuel Characteristic
Classification System - Quantifying, Classifying, and Creating Fuelbeds for Resource Planning.
Canadian Journal of Forest Research. 37(12): 2383-2393. FCCS is available at:
http://www.fs.fed.us/pnw/fera/fccs/index.shtml.
Pinder, R., Strader, R., Davidson, C. & Adams, P. A temporally and spatially resolved ammonia emission
inventory for dairy cows in the United States. Atmos. Environ. 38.23, 3747-3756 (2004). 2.
Pinder, R., Pekney, N., Davidson, C. & Adams, P. A process-based model of ammonia emissions
from dairy cows: improved temporal and spatial resolution. Atmos. Environ. 38.9, 1357-1365
(2004).
Pinder, R., Pekney, N., Davidson, C. & Adams, P. A process-based model of ammonia emissions from
dairy cows: improved temporal and spatial resolution. Atmos. Environ. 38.9, 1357-1365 (2004).
Pouliot, G. and J. Bash, 2015. Updates to Version 3.61 of the Biogenic Emission Inventory System
(BEIS). Presented at Air and Waste Management Association conference, Raleigh, NC, 2015.
Pouliot, G., H. Simon, P. Bhave, D. Tong, D. Mobley, T. Pace, and T. Pierce . (2010) "Assessing the
Anthropogenic Fugitive Dust Emission Inventory and Temporal Allocation Using an Updated
Speciation of Particulate Matter." International Emission Inventory Conference, San Antonio, TX.
Available at http://www.epa.gov/ttn/chief/conference/eil9/session9/pouliot.pdf.
78
-------
Raffuse, S., D. Sullivan, L. Chinkin, S. Larkin, R. Solomon, A. Soja, 2007. Integration of Satellite-
Detected and Incident Command Reported Wildfire Information into BlueSky, June 27, 2007.
Available at: https://www.airfire.org/smartfire.
Reichle. L..R. Cook. C. Yanca. D. Sonntag. 2015. "Development of organic gas exhaust speciation
profiles for nonroad spark-ignition and compression-ignition engines and equipment", Journal of
the Air & Waste Management Association, 65:10, 1185-1193, DOI:
10.1080/10962247.2015.1020118.
Secretaria de Medio Ambiente y Recursos Naturales (SEMARNAT), Gobierno De Mexico, 2019.
Inventarios Nacionales de Emisiones en el ano 2016. Documentation available from
SEMARNAT: https://www.gob.mx/semarnat/documentos/documentos-del-inventario-nacional-
de-emisiones.
Sullivan D.C., Raffuse S.M., Pryden D.A., Craig K.J., Reid S.B., Wheeler N.J.M., Chinkin L.R., Larkin
N.K., Solomon R., and Strand T. (2008) Development and applications of systems for modeling
emissions and smoke from fires: the BlueSky smoke modeling framework and SMARTFIRE: 17th
International Emissions Inventory Conference, Portland, OR, June 2-5.
Wang, Y., P. Hopke, O. V. Rattigan, X. Xia, D. C. Chalupa, M. J. Utell. (2011) "Characterization of
Residential Wood Combustion Particles Using the Two-Wavelength Aethalometer", Environ. Sci.
Technol., 45 (17), pp 7387-7393.
Wiedinmyer, C., S.K. Akagi, R.J. Yokelson, L.K. Emmons, J.A. Al-Saadi3, J. J. Orlando1, and A. J. Soja.
(2011) "The Fire INventory from NCAR (FINN): a high resolution global model to estimate the
emissions from open burning", Geosci. Model Dev., 4, 625-641. http://www.geosci-model-
dev.net/4/625/2011/ doi: 10.5194/gmd-4-625-2011.
Urbanski S.P. (2014) Wildland fire emissions, carbon, and climate: emissions factors. Forest Ecology and
Management, 317, 51-60.
Yarwood, G., S. Rao, M. Yocke, and G. Whitten, 2005: Updates to the Carbon Bond Chemical
Mechanism: CB05. Final Report to the US EPA, RT-0400675. Available at
http://www.camx.com/publ/pdfs/CB05 Final Report 120805.pdf.
79
-------
4.0 CMAQ Air Quality Model Estimates
4.1 Introduction to the CMAQ Modeling Platform
The Clean Air Act (CAA) provides a mandate to assess and manage air pollution levels to protect human
health and the environment. EPA has established National Ambient Air Quality Standards (NAAQS),
requiring the development of effective emissions control strategies for such pollutants as ozone and
particulate matter. Air quality models are used to develop these emission control strategies to achieve the
objectives of the CAA.
Historically, air quality models have addressed individual pollutant issues separately. However, many of
the same precursor chemicals are involved in both ozone and aerosol (particulate matter) chemistry;
therefore, the chemical transformation pathways are dependent. Thus, modeled abatement strategies of
pollutant precursors, such as volatile organic compounds (VOC) and NOx to reduce ozone levels, may
exacerbate other air pollutants such as particulate matter. To meet the need to address the complex
relationships between pollutants, EPA developed the Community Multi scale Air Quality (CMAQ)
modeling system12- The primary goals for CMAQ are to:
Improve the environmental management community's ability to evaluate the impact of air quality
management practices for multiple pollutants at multiple scales.
Improve the scientist's ability to better probe, understand, and simulate chemical and physical
interactions in the atmosphere.
The CMAQ modeling system brings together key physical and chemical functions associated with the
dispersion and transformations of air pollution at various scales. It was designed to approach air quality as
a whole by including state-of-the-science capabilities for modeling multiple air quality issues, including
tropospheric ozone, fine particles, toxics, acid deposition, and visibility degradation CMAQ relies on
emission estimates from various sources, including the U.S. EPA Office of Air Quality Planning and
Standards" current emission inventories, observed emission from major utility stacks, and model estimates
of natural emissions from biogenic and agricultural sources. CMAQ also relies on meteorological
predictions that include assimilation of meteorological observations as constraints. Emissions and
meteorology data are fed into CMAQ and run through various algorithms that simulate the physical and
chemical processes in the atmosphere to provide estimated concentrations of the pollutants. Traditionally,
the model has been used to predict air quality across a regional or national domain and then to simulate the
effects of various changes in emission levels for policymaking purposes. For health studies, the model can
also be used to provide supplemental information about air quality in areas where no monitors exist.
CMAQ was also designed to have multi-scale capabilities so that separate models were not needed for
urban and regional scale air quality modeling. The CMAQ simulation performed for this 2017 assessment
used a single domain that covers the entire continental U.S. (CONUS) and large portions of Canada and
Mexico using 12-km by 12-ktn horizontal grid spacing. Currently, 12-km x 12-ktn resolution is sufficient
12 Byun, D.W., and K. L. Schere, 2006: Review of the Governing Equations, Computational Algorithms, and Other
Components of the Models-3 Community Multiscale Air Quality (CMAQ) Modeling System. Applied Mechanics Reviews,
Volume 59, Number 2 (March 2006), pp. 51-77.
80
-------
as the highest resolution for most regional-scale air quality model applications and assessments.1' With the
temporal flexibility of the model, simulations can be performed to evaluate longer term (annual to multi-
year) pollutant climatologies as well as short-term (weeks to months) transport from localized sources. By
making CMAQ a modeling system that addresses multiple pollutants and different temporal and spatial
scales, CMAQ has a "one atmosphere" perspective that combines the efforts of the scientific community.
Improvements will be made to the CMAQ modeling system as the scientific community further develops
the state-of-the-science.
For more information on CMAQ, go to https://www.epa.gov/cmaq or http://www.cmascenter.ore.
4.1.1 Advantages and Limitations of the CMAQ Air Quality Model
An advantage of using the CMAQ model output for characterizing air quality for use in comparing with
health outcomes is that it provides a complete spatial and temporal coverage across the U.S. CMAQ is a
three-dimensional Eulerian photochemical air quality model that simulates the numerous physical and
chemical processes involved in the formation, transport, and destruction of ozone, particulate matter and
air toxics for given input sets of initial and boundary conditions, meteorological conditions and emissions.
The CMAQ model includes state-of-the-science capabilities for conducting urban to regional scale
simulations of multiple air quality issues, including tropospheric ozone, fine particles, toxics, acid
deposition and visibility degradation. However, CMAQ is resource intensive, requiring significant data
inputs and computing resources.
An uncertainty of using the CMAQ model includes structural uncertainties, representation of physical and
chemical processes in the model. These consist of: choice of chemical mechanism used to characterize
reactions in the atmosphere, choice of land surface model and choice of planetary boundary layer.
Another uncertainty in the CMAQ model is based on parametric uncertainties, which includes
uncertainties in the model inputs: hourly meteorological fields, hourly 3-D gridded emissions, initial
conditions, and boundary conditions. Uncertainties due to initial conditions are minimized by using a 10-
day ramp-up period from which model results are not used in the aggregation and analysis of model
outputs. Evaluations of models against observed pollutant concentrations build confidence that the model
performs with reasonable accuracy despite the uncertainties listed above. A detailed model evaluation for
ozone and PM2.5 species provided in Section 4.3 shows generally acceptable model performance which is
equivalent or better than typical state-of-the-science regional modeling simulations as summarized in
Simon et al ., 201214.
4.2 CMAQ Model Version, Inputs and Configuration
This section describes the air quality modeling platform used for the 2017 CMAQ simulation. A modeling
platform is a structured system of connected modeling-related tools and data that provide a consistent and
transparent basis for assessing the air quality response to changes in emissions and/or meteorology. A
platform typically consists of a specific air quality model, emissions estimates, a set of meteorological
inputs, and estimates of boundary conditions representing pollutant transport from source areas outside the
region modeled. We used the CMAQ modeling system as part of the 2017 Platform to provide a national
13 U.S. EPA (2018), Modeling Guidance for Demonstrating Air Quality Goals for Ozone, PM2.5, and Regional Haze, pp 205.
https ://www3. epa.gov/ttn/scram/guidance/guide/03 -PM-RH-Modeling_Guidance-2018 .pdf.
14 Simon, H., Baker, K.R., and Phillips, S. (2012) Compilation and interpretation of photochemical model performance
statistics published between 2006 and 2012. Atmospheric Environment 61, 124-139.
81
-------
scale air quality modeling analysis. The CMAQ model simulates the multiple physical and chemical
processes involved in the formation, transport, and destruction of ozone and fine particulate matter
(PM2.5).
This section provides a description of each of the main components of the 2017 CMAQ simulation along
with the results of a model performance evaluation in which the 2017 model predictions are compared to
corresponding measured ambient concentrations.
4.2.1 CMAQ Model Version
CMAQ is a non-proprietary computer model that simulates the formation and fate of photochemical
oxidants, including PM2.5 and ozone, for given input sets of meteorological conditions and emissions. As
mentioned previously, CMAQ includes numerous science modules that simulate the emission, production,
decay, deposition and transport of organic and inorganic gas-phase and particle pollutants in the
atmosphere. This 2017 analysis employed CMAQ version 5.3.1.15 The 2017 CMAQ run included CB6r3
chemical mechanism, AER07 aerosol module with non-volatile Primary Organic Aerosol (POA). The
CMAQ community model versions 5.0.2 and 5.1 were most recently peer-reviewed in September of 2016
for the U.S. EPA.16
4.2.2 Model Domain and Grid Resolution
The CMAQ modeling analyses were performed for a domain covering the continental United States, as
shown in Figure 4-1. This single domain covers the entire continental U.S. (CONUS) and large portions
of Canada and Mexico using 12-km by 12-km horizontal grid spacing. The 2017 simulation used a
Lambert Conformal map projection centered at (-97, 40) with true latitudes at 33 and 45 degrees
north. The 12-krn CMAQ domain consisted of 396 by 246 grid cells and 35 vertical layers. Table 4-1
provides some basic geographic information regarding the 12-km CMAQ domain. The model extends
vertically from the surface to 50 millibars (approximately 17,600 meters) using a sigma-pressure
coordinate system. Table 4-2 shows the vertical layer structure used in the 2017 simulation. Air quality
conditions at the outer boundary of the 12-km domain were taken from the northern hemispheric CMAQ
model (discussed in Section 4.2.4).
Table 4-1. Geographic Information for 2017 12-km Modeling Domain
15 CMAQ version 5.3.1: https://doi.org/10.5281/zenodo.3585898: https://www.epa.gov/cmaa/cmaa-models-O. CMAQ v5.3.1 is
also available from the Community Modeling and Analysis System (CMAS) at: http://www.cmascenter.org.
" Monin, M.D., Astitha, M., Barsanti. K.C.. Brown, N.J., Kaduwela, A., McKeen, S.A., Pickering, K.E. (September 28, 2015).
Final Report: Fifth Peer Review of the CMAQ Model, https://www.epa.gov/sites/production/files/2016-
11/documents/cmaa fifth review final report 2015.pdf. This peer review was focused on CMAQ v5.0.2, which was released
in May, 2014, as well as CMAQ \5.1, which was released in October 2015. It is available from the Community Modeling and
Analysis System (CMAS) as well as previous peer-review reports at: http://www.cmascenter.org.
82
-------
National 12 km CMAQ Modeling Configuration
Map Projection
Lambert Conformal Projection
Grid Resolution
12km
Coordinate Center
97 W, 40 N
True Latitudes
33 and 45 N
Dimensions
396 x 246 x 35
Vertical Extent
35 Layers: Surface to 50 mb level (see Table 4-2)
Table 4-2. Vertical layer structure for 2017 CMAQ simulation (heights are layer top).
Vertical
Layers
Sigma P
Pressure
(mb)
Approximate
Height (m)
35
0.0000
50.00
17,556
34
0.0500
97.50
14,780
33
0.1000
145.00
12,822
32
0.1500
192.50
11,282
31
0.2000
240.00
10,002
30
0.2500
287.50
8,901
29
0.3000
335.00
7,932
28
0.3500
382.50
7,064
27
0.4000
430.00
6,275
26
0.4500
477.50
5,553
25
0.5000
525.00
4,885
24
0.5500
572.50
4,264
23
0.6000
620.00
3,683
22
0.6500
667.50
3,136
21
0.7000
715.00
2,619
20
0.7400
753.00
2,226
19
0.7700
781.50
1,941
18
0.8000
810.00
1,665
17
0.8200
829.00
1,485
16
0.8400
848.00
1,308
15
0.8600
867.00
1,134
14
0.8800
886.00
964
13
0.9000
905.00
797
12
0.9100
914.50
714
11
0.9200
924.00
632
10
0.9300
933.50
551
9
0.9400
943.00
470
8
0.9500
952.50
390
83
-------
Vertical
Layers
Sigma P
Pressure
(mb)
Approximate
Height (m)
7
0.9600
962.00
311
6
0.9700
971.50
232
5
0.9800
981.00
154
4
0.9850
985.75
115
3
0.9900
990.50
77
2
0.9950
995.25
38
1
0.9975
997.63
19
0
1.0000
1000.00
0
12US2 domain \ ft ^ L
x,y origin; -2412OOOryi,Vl62Off0Otn
col: 396 row:246 ! \ V V
Figure 4-1. Map of the 2017 CMAQ Modeling Domain. The purple box denotes the 12-km national
modeling domain.
4.2.3 Modeling Period/ Ozone Episodes
The 12-km CMAQ modeling domain was modeled for the entire year of 2017. The annual simulation
included a spin-up period, comprised of 10 days before the beginning of the simulation, to mitigate the
effects of initial concentrations. All 365 model days were used in the annual average levels of PM2.5. For
the 8-hour ozone, we used modeling results from the period between May 1 and September 30. This 153-
day period generally conforms to the ozone season across most parts of the U.S. and contains the majority
of days that observed high ozone concentrations.
84
-------
4.2.4 Model Inputs: Emissions, Meteorology and Boundary Conditions
2017Emissions: The emissions inventories used in the 2017 air quality modeling are described in Section
3, above.
2017 Meteorological Input Data: The gridded meteorological data for the entire year of 2017 at the 12-
km continental United States scale domain was derived from the publicly available version 3.8 of the
Weather Research and Forecasting Model (WRF), Advanced Research WRF (ARW) core.17 The WRF
Model is a state-of-the-science mesoscale numerical weather prediction system developed for both
operational forecasting and atmospheric research applications (http://wrf-model.org ). The 12US WRF
model was initialized using the 12-km North American Model (12NAM)18 analysis product provided by
National Climatic Data Center (NCDC). Where 12NAM data was unavailable, the 40-km Eta Data
Assimilation System (EDAS) analysis (ds609.2) from the National Center for Atmospheric Research
(NCAR) was used. Analysis nudging for temperature, wind, and moisture was applied above the
boundary layer only. The model simulations were conducted continuously. The 'ipxwrf program was
used to initialize deep soil moisture at the start of the run using a 10-day spin-up period. The 2017 WRF
meteorology simulated was based on 2011 National Land Cover Database (NLCD).19 The WRF
simulation included the physics options of the Pleim-Xiu land surface model (LSM), Asymmetric
Convective Model version 2 planetary boundary layer (PBL) scheme, Morrison double moment
microphysics, Kain- Fritsch cumulus parameterization scheme utilizing the moisture-advection trigger"
and the RRTMG long-wave and shortwave radiation (LWR/SWR) scheme.21 These simulations also
incorporated observed lightning data from the National Lightning Detection Network (NLDN) to suppress
(force) deep convection where lightning is absent (present).22 In addition, the Group for High Resolution
Sea Surface Temperatures (GHRSST)23'24 1 -km SST data was used for SST information to provide more
resolved information compared to the more coarse data in the NAM analysis.
2017 Initial and Boundary Conditions: The lateral boundary and initial species concentrations were
provided by a northern hemispheric application of a CMAQ modeling platform to the year 2017. The
hemispheric-scale platform uses a polar stereographic projection at 108-km resolution to completely and
continuously cover the northern hemisphere for 2017 with meteorology, emissions, and atmospheric
processing of pollutants. Meteorology is provided by Weather Research and Forecasting model (WRF
v3.8) using 44 non-hydrostatic sigma-pressure layers between the surface and 50 hPa (~20-km
asl). Emissions were provided by the Near-Real-Time modeling project from the Office of Research and
17 Skamarock, W.C., Klemp, J.B., Dudhia, J., Gill, DO., Barker, D.M., Duda, M.G., Huang, X., Wang, W., Powers, J.G., 2008.
A Description of the Advanced Research WRF Version 3.
18 North American Model Analysis-Only, http://nomads.ncdc.noaa.gov/data.php; download from
ftp://nomads.ncdc.noaa.gov/NAM/analysis_only/.
19 National Land Cover Database 2011, http://www.mrlc.gov/nlcd201 l.php
20 Ma, L-M. and Tan, Z-M, 2009. Improving the behavior of the Cumulus Parameterization for Tropical Cyclone Prediction:
Convection Trigger. Atmospheric Research 92 Issue 2, 190-211.
http://www.sciencedirect.com/science/article/pii/S01698095080Q2585
21 Gilliam. R.C., Plcim, I.E., 2010. Performance Assessment of New Land Surface and Planetary Boundary Layer Physics in the
WRF- ARVV. Journal of Applied Meteorology and Climatology 49, 760-774.
22 Heath, Nicholas K., Pleim, J.E., Gilliam, R., Kang, D., 2016. A simple lightning assimilation technique for improving
retrospective WRF simulations. Journal of Advances in Modeling Earth Systems. 8. 10.1002/2016MS000735.
23 Stammer, D., F.J. Wentz, and C.L. Gentemann, 2003, Validation of Microwave Sea Surface Temperature Measurements for
Climate Purposes, J. Climate, 16, 73-87.
24 Global High Resolution SST (GHRSST) analysis, https://www.ghrsst.org/.
85
-------
Development. The emission platform was consistent and combines EDGAR-HTAP (v2)25, real-time fires
from the FINN system, and GEIA climatological natural emissions. The CMAQ model also included the
on-line windblown dust emission sources (excluding agricultural land), which are not always included in
the regional platform but are important for large-scale transport of dust. The Near-Real-Time system used
a combination of model version and stratospheric treatment depending on the date. The atmospheric
processing (transformation and fate) was simulated by CMAQ (<=2017-09-23: v5.2; >=2017-09-24: v5.3)
using the Carbon Bond (cb5e51, cb6r3) with linearized halogen chemistry and the aerosol model with
semi-volatile primary organic carbon (AE6svPOA). The stratosphere was simulated by the Potential
Vorticity scaling method, which was only operational after May. As a result, the hemispheric scale
simulation is expected to underpredict stratospheric contributions in spring.
4.3 CMAQ Model Performance Evaluation
An operational model performance evaluation for ozone and PM2.5 and its related speciated components
was conducted for the 2017 simulation using state/local monitoring sites data in order to estimate the
ability of the CMAQ modeling system to replicate the 2017 base year concentrations for the 12-km
continental U.S. domain.
There are various statistical metrics available and used by the science community for model performance
evaluation. For a robust evaluation, the principal evaluation statistics used to evaluate CMAQ
performance were two bias metrics, mean bias and normalized mean bias; and two error metrics, mean
error and normalized mean error.
Mean bias (MB) is used as average of the difference (predicted - observed) divided by the total number of
replicates (n). Mean bias is defined as:
MB = P ~ O) , where P = predicted and O = observed concentrations.
Mean error (ME) calculates the absolute value of the difference (predicted - observed) divided by the total
number of replicates (n). Mean error is defined as:
ME = i£;|P-0|
Normalized mean bias (NMB) is used as a normalization to facilitate a range of concentration magnitudes.
This statistic averages the difference (model - observed) over the sum of observed values. NMB is a
useful model performance indicator because it avoids overinflating the observed range of values,
especially at low concentrations. Normalized mean bias is defined as:
±{p-o)
NMB = -J *100, where P = predicted concentrations and O = observed
i(o)
25 Janssens-Maenhout, G., Dentener, F., Van Aardenne, J., Monni, S., Pagliari, V., Orlandini, L., Klimont, Z., Kurokawa, J.,
Akimoto, H., Ohara, T., others, 2012. EDGAR-HTAP: a harmonized gridded air pollution emission dataset based on national
inventories. European Commission Publications Office, Ispra (Italy). JRC68434, EUR report No EUR 25, 299-2012.
86
-------
Normalized mean error (NME) is also similar to NMB, where the performance statistic is used as a
normalization of the mean error. NME calculates the absolute value of the difference (model - observed)
over the sum of observed values. Normalized mean error is defined as
i\p-o\
NME= J *100
n
I (O)
1
The performance statistics were calculated using predicted and observed data that were paired in time and
space on an 8-hour basis. Statistics were generated for each of the nine National Oceanic and
Atmospheric Administration (NOAA) climate regions26 of the 12-km U.S. modeling domain (Figure 4-2).
The regions include the Northeast, Ohio Valley, Upper Midwest, Southeast, South, Southwest, Northern
Rockies, Northwest and West27 28 as were originally identified in Karl and Koss (1984)29.
U.S. Climate Regions
Figure 4-2. NOAA Nine Climate Regions (source: htti)://www.ncdc.noaa.gov/monitoring-references/mai)s/us-
climate-regions.i)hi)#references)
In addition to the performance statistics, regional maps which show the MB, ME, NMB, and NME were
26 NOAA, National Centers for Environmental Information scientists have identified nine climatically consistent regions within
the contiguous U.S., http://www.ncdc.noaa.gov/monitoring-references/maps/us-climate-regions.php.
r The nine climate regions are defined by States where: Northeast includes CT, DE, ME, MA, MD. NH, NJ, NY, PA, RI, and
VT; Ohio Valley includes IL, IN, KY, MO, OH. TN. and WV; Upper Mdwest includes IA, MI, MN, and WI; Southeast
includes AL, FL, GA, NC. SC, and VA; South includes AR, KS, LA, MS, OK, and TX; Southwest includes AZ. CO. NM. and
UT; Northern Rockies includes MT, NE, ND. SD, WY; Northwest includes ID. OR, and WA; and West includes CA and NV.
28 Note most monitoring sites in the West region are located in California (see Figure 4-2), therefore statistics for the West will
be mostly representative of California ozone air quality.
29 Karl, T. R. and Koss, W. J., 1984: "Regional and National Monthly, Seasonal, and Annual Temperature Weighted by Area,
1895-1983." Historical Climatology Series 4-3, National Climatic Data Center. Asheville, NC, 38 pp.
87
-------
prepared for the ozone season. May through September, at individual monitoring sites as well as on an
annual basis for PM2.5 and its component species.
Evaluation for 8-hour Daily Maximum Ozone: The operational model performance evaluation for eight-
hour daily maximum ozone was conducted using the statistics defined above. Ozone measurements in the
continental U.S. were included in the evaluation and were taken from the 2017 state/local monitoring site
data in AQS and the Clean Air Status and Trends Network (CASTNet).
The 8-hour ozone model performance bias and error statistics for each of the nine NOAA climate regions
and each season are provided in Table 4-4. Seasons were defined as: winter (December-January-
February), spring (March-Apri 1 -May), summer (June, July, August), and fall (September-October-
November). In some instances, observational data were excluded from the analysis and model evaluation
based on a completeness criterion of 75 percent. Spatial plots of the MB, ME, NMB and NME for
individual monitors are shown in Figures 4-3 through 4-6, respectively. The statistics shown in these two
figures were calculated over the ozone season. May through September, using data pairs on days with
observed 8-hour ozone of greater than or equal to 60 ppb.
In general, the model performance statistics indicate that the 8-hour daily maximum ozone concentrations
predicted by the 2017 CMAQ simulation closely reflect the corresponding 8-hour observed ozone
concentrations in space and time in each subregion of the 12-km modeling domain. As indicated by the
statistics in Table 4-4, bias and error for 8-hour daily maximum ozone are relatively low in each
subregion, not only in the summer when concentrations are highest, but also during other times of the year.
Generally, 8-hour ozone at the AQS sites in the summer and fall is over predicted with the greatest over
prediction in the Southeast, Northwest and Norther Rockies (NMB ranging between 0.3 to 23 percent).
Likewise, 8-hour ozone at the CASTNet sites in the summer and fall is typically over predicted except in
the summer at Upper Midwest, Northern Rockies and in the West where the bias shows an under
prediction (NMB ranging from -1% to -4%). 8-hour ozone is under predicted at AQS and CASTNet sites
in all of the climate regions in the winter and spring (with NMBs less than approximately 20 percent in
each subregion).
Model bias at individual sites during the ozone season is similar to that seen on a subregional basis for the
summer. Figure 4-3 shows the mean bias for 8-hour daily maximum ozone greater than 60 ppb is
generally ±10 ppb across the AQS and CASTNet sites. Likewise, the information in Figure 4-5 indicates
that the normalized mean bias for days with observed 8-hour daily maximum ozone greater than 60 ppb is
within ± 20 percent at the vast majority of monitoring sites across the U.S. domain. Model error, as seen
from Figures 4-4 and 4-6, is generally 2 to 10 ppb and 20 percent or less at most of the sites across the
U.S. modeling domain. Somewhat greater error is evident at sites in several areas most notably in the
West, Northern Rockies, Northeast, Upper Midwest, Southeast, along portions of the Gulf Coast, and
Great Lakes coastline.
88
-------
Table 4-4. Summary of CMAQ 2017 8-Hour Daily Maximum Ozone Model Performance Statistics
by NOAA climate region, by Season and Monitoring Network.
Climate
Monitor
No. of
MB
ME
NMB
NME
region
Network
Season
Obs
(PPb)
(PPb)
(%)
(%)
AQS
Winter
11,568
-3.8
7.2
-12.5
20.9
Spring
16,622
-7.5
9.0
-17.2
20.6
Summer
16,296
3.2
6.6
7.4
15.2
Northeast
Fall
13371
4.6
6.4
14.0
18.4
CASTNet
Winter
1,290
-4.4
7.3
-11.5
21.8
Spring
1,346
-7.8
9.1
-17.7
20.6
Summer
1,301
2.8
6.0
6.9
14.6
Fall
1,302
4.8
6.3
13.3
18.5
AQS
Winter
4,546
-2.0
6.2
-6.5
20.2
Spring
20,927
-5.2
7.2
-11.7
16.3
Summer
20,843
2.5
6.0
5.5
13.0
Ohio Valley
Fall
14,765
3.9
6.1
10.4
16.1
CASTNet
Winter
1,541
-2.4
6.5
-7.2
19.4
Spring
1,555
-6.2
7.8
-13.5
16.9
Summer
1,536
2.6
5.8
5.9
13.2
Fall
1,559
3.3
5.9
9.0
16.0
AQS
Winter
2,095
-3.6
6.7
-11.6
21.4
Spring
9,626
-7.1
8.4
-16.8
19.8
Summer
9,137
0.5
6.1
1.3
14.4
Fall
6,221
3.6
5.7
10.1
16.1
Upper Midwest
CASTNet
Winter
433
-4.7
7.1
-13.7
20.7
Spring
444
-9.5
10.0
-21.7
23.0
Summer
388
-1.3
5.9
-3.4
14.6
Fall
403
3.4
5.1
10.0
15.1
AQS
Winter
6,946
-4.2
6.5
-11.0
17.2
Spring
15,589
-5.4
7.6
-11.8
16.8
Summer
15,111
6.2
7.4
16.9
20.2
Fall
11,162
5.0
6.4
13.3
17.2
Southeast
CASTNet
Winter
920
-4.7
7.5
-12.4
19.3
Spring
976
-7.3
8.5
-15.5
18.2
Summer
907
4.0
5.8
10.3
15.1
Fall
902
3.7
6.1
9.6
16.2
89
-------
Climate
Monitor
No. of
MB
ME
NMB
NME
region
Network
Season
Obs
(%)
(%)
AQS
Winter
10,799
-2.6
6.3
-7.8
19.0
Spring
12,941
-3.8
7.0
-8.4
15.7
Summer
12,276
4.9
7.4
12.8
19.3
Fall
11,880
2.6
5.8
6.4
14.5
South
CASTNet
Winter
506
-2.5
6.5
-7.3
18.4
Spring
523
-6.1
8.4
-13.4
18.4
Summer
519
3.0
7.7
7.6
19.6
Fall
541
2.3
5.2
6.0
13.5
AQS
Winter
9,407
-3.6
7.7
-9.4
19.9
Spring
10,324
-10.9
11.2
-20.5
21.1
Summer
10,401
0.1
6.7
0.3
12.2
Fall
9,940
3.8
6.2
8.7
14.2
Southwest
CASTNet
Winter
756
-5.6
7.6
-12.6
17.1
Spring
783
-12.9
13.1
-23.6
24.0
Summer
786
0.6
6.0
1.1
11.3
Fall
797
3.3
5.0
7.2
10.9
AQS
Winter
4,355
-3.9
7.8
-10.1
20.2
Spring
4,732
-8.3
9.1
-18.5
20.3
Summer
4,675
-0.5
6.5
-1.0
13.2
Northern
Fall
4,531
5.3
6.6
14.4
17.8
Rockies
CASTNet
Winter
762
-4.3
7.6
-10.6
18.9
Spring
779
-10.1
10.4
-21.4
22.1
Summer
681
-1.8
6.2
-3.5
12.2
Fall
704
4.2
5.8
10.5
14.7
AQS
Winter
568
-3.3
9.4
-9.9
28.2
Spring
1,324
-6.9
8.6
-16.8
20.9
Summer
2,447
3.8
7.6
9.0
17.9
Northwest
Fall
1,067
8.1
9.4
23.1
26.6
CASTNet
Winter
Spring
Summer
Fall
90
-------
Climate Monitor No. of MB ME NMB NME
region Network Season Obs (ppb) (ppb) (%) (%)
West
AQS
Winter
13,840
-3,7
7,9
-10.5
22,1
Spring
16,419
-9.7
10.3
-20,0
21.3
Summer
17,053
-0.7
7.6
-1.3
14.3
Fail
15,656
1.1
6.6
2,4
14,5
CASTNet
Winter
515
-4.0
8.0
-9.5
19.2
Spring
526
-11.7
12.0
-22.9
23,5
Summer
540
-4.6
7.9
-7.8
13.4
Fai! 539 1,1 5.8 2.2 12.1
units = ppb
coverage limit = 75%
1
6
4
-8
| 10
-12
-14
|-16
-18
<-20
* CASTNET Daily AQS Daily
Figure 4-3. Mean Bias (ppb) of 8-hour daily maximum ozone greater than 60 ppb over the period
May-September 2017 at AQS and CASTNet monitoring sites in the continental U.S. modeling
domain.
Q3_8hrmax MB (ppb) for run CMAQ_2017gb_17j_12US2
for 20170501 to 20170930
91
-------
03_8hrmax ME {ppb) for run CMAQ_2017gb_17j_12US2 for 20170501 to 20170930
units = ppb
* CASTNET Daily AQS Daily
Figure 4-4. Mean Error (ppb) of 8-hour daily maximum ozone greater than 60 ppb over the period
May-September 2017 at AQS and CAST Net monitoring sites in the continental U.S. modeling
domain.
03_8hrmax NMB {%) for run CMAQ_2017gb_17j_12US2 for 20170501 to 20170930
coverage limit = 75%
* CASTNET Daily AQS Daily
Figure 4-5. Normalized Mean Bias (%) of 8-hour daily maximum ozone greater than 60 ppb over
the period May-September 2017 at AQS and CASTNet monitoring sites in the continental U.S.
modeling domain.
92
-------
03_8hrmax NME (%) for run CMAQ_2017gb_17j_12US2 for 20170501 to 20170930
units
coverage limit = 75%
* CASTNET Daily AQS Daily
Figure 4-6. Normalized Mean Error (%) of 8-hour daily maximum ozone greater than 60 ppb over
the period May-September 2017 at AQS and CASTNet monitoring sites in the continental U.S.
modeling domain.
Evaluation for Annual PM?.s components: The PM evaluation focuses on PM2.5 components including
sulfate (SO4), nitrate (NO3), total nitrate (TNO3 = NO3 + HNO3), ammonium (NH4), elemental carbon
(EC), and organic carbon (OC). The bias and error performance statistics were calculated on an annual
basis for each of the nine NOAA climate subregions defined above (provided in Table 4-5). PM2.5
measurements for 2017 were obtained from the following networks for model evaluation: Chemical
Speciation Network (CSN, 24-hour average), Interagency Monitoring of Protected Visual Environments
(IMPROVE, 24-hour average, and Clean Air Status and Trends Network (CASTNet, weekly average).
For PM2.5 species that are measured by more than one network, we calculated separate sets of statistics for
each network by subregion. In addition to the tabular summaries of bias and error statistics, annual spatial
maps which show the mean bias, mean error, normalized mean bias and normalized mean error by site for
each PM2.5 species are provided in Figures 4-7 through 4-30.
As indicated by the statistics in Table 4-5, annual average sulfate is consistently over predicted at
CASTNet, IMPROVE, and CSN monitoring sites across the 12-km modeling domain (with MB values
ranging from 0.0 to -0.5 |igm"3) except at CSN sites in the South and at IMPROVE sites in the Ohio
Valley and Southeast, as well as at CASTNet sites in the Ohio Valley, Southeast and South. Sulfate
performance shows moderate error in the eastern subregions (ranging from 25 to 39 percent) while
Western subregions show slightly larger error (ranging from 44 to >100 percent). Figures 4-7 through 4-
10, suggest spatial patterns vary by region. The model bias for most of the Northeast, Southeast,
Central and Southwest states are within ±30 percent. The model bias appears to be slightly greater in
the Northwest with over predictions up to 100 percent at individual monitors. Model error also shows a
spatial trend by region, where much of the Eastern states are 10 to 40 percent, the Western and Central
U.S. states are 20 to 100 percent.
93
-------
Annual average nitrate is under predicted at the urban CSN monitoring sites in the Ohio Valley,
Upper Midwest, Southeast, South, Southwest, Northern Rockies, Northwest and West (NMB
in the range of -2 to -72 percent), except in the Northeast, where nitrate is over predicted (NMB
approximately 7%). At IMPROVE rural sites, annual average nitrate is under predicted at all
subregions, except in the Northeast where nitrate is over predicted by approximately 30 percent.
Likewise, model performance of total nitrate at sub-urban CASTNet monitoring sites shows an under
prediction at all subregions (NMB in the range of -7 to -42 percent) except in the Northeast. Model
error for nitrate and total nitrate is somewhat greater for each of the nine N O A A climate
subregions as compared to sulfate. Model bias at individual sites indicates over prediction of greater
than 20 percent at most monitoring sites along the Northeast and Northwest coastline as well as in the
Southeast as indicated in Figure 4-13. The exception to this is in the Ohio Valley, South, Southwest,
Northern Rockies and Western U.S. of the modeling domain where there appears to be a greater
number of sites with under prediction of nitrate of 10 to 80 percent. Model error for annual nitrate, as
shown in Figures 4-12 and 4-15, is least at sites in portions of the Ohio Valley and Upper Midwest.
Annual average ammonium model performance as indicated in Table 4-5 has a tendency for the model
to under predict across the CASTNet sites (ranging from -32 to -66 percent). Ammonium performance
across the urban CSN sites shows an under prediction in two of the climate subregions (Southwest and
West ranging from -47 to -50 percent), except in the Northeast, Ohio Valley, Upper Midwest, Southeast,
South, Northwest, and Northern Rockies (over prediction of NMB). The spatial variation of ammonium
across the majority of individual monitoring sites in the Eastern U.S. shows bias within ±50 percent
(Figures 4-19 and 4-21). A larger bias is seen in the Northeast and in the Northern Rockies, (over
prediction bias on average 80 to 100 percent). The urban monitoring sites exhibit larger errors than at
rural sites for ammonium.
Annual average elemental carbon is under predicted in all of the nine climate regions at urban and rural
sites except in the Northeast and Northwest (overprediction between 3 to 20 percent). There is not a large
variation in error statistics from subregion to subregion or at urban versus rural sites.
Annual average organic carbon is over predicted across most subregions in rural IMPROVE areas (NMB
ranging from 8 to 74 percent), except in the Southwest and Northern Rockies where the NMB ranges from
-3 to -5 percent. The model over predicted annual average organic carbon in all subregions at urban CSN
sites except in the Northern Rockies (NMB approximately -4 percent). Similar to elemental carbon, error
model performance does not show a large variation from subregion to subregion or at urban versus rural
sites.
94
-------
Table 4-5. Summary of CMAQ 2017 Annual PM Species Model Performance Statistics by NOAA
Climate region, by Monitoring Network.
Monitor
No. of
MB
ME
NMB
NME
Pollutant Network
Subregion
Obs
(|jgm3)
(|jgm3)
(%)
(%)
CSN
Northeast
3,082
0.2
0.4
26.4
48.4
Ohio Valley
2,227
0.1
0.4
12.3
37.1
Upper Midwest
1,247
0.3
0.4
35.4
48.8
Southeast
2,029
0.2
0.4
16.0
38.5
South
1,394
0.1
0.5
-1.3
34.1
Southwest
3,796
0.1
0.3
10.3
61.3
Northern Rockies
642
0.4
0.4
63.4
76.7
Northwest
646
0.4
0.4
84.0
97.1
West
1,173
0.1
0.5
9.4
55.0
IMPROVE
Northeast
1,862
0.2
0.3
22.8
45.0
Ohio Valley
956
0.0
0.3
-3.4
28.6
Upper Midwest
819
0.2
0.3
21.3
40.8
Southeast
1,498
0.0
0.3
-2.2
30.9
Sulfate
South
1,016
0.0
0.4
7.7
39.4
Southwest
3,796
0.1
0.3
20.5
59.7
Northern Rockies
2,219
0.2
0.3
55.7
72.7
Northwest
1,915
0.3
0.3
>100
>100
West
2,365
0.2
0.4
35.1
68.8
CASTNet
Northeast
923
0.0
0.2
1.8
21.7
Ohio Valley
894
-0.2
0.3
-13.2
20.1
Upper Midwest
293
0.1
0.2
7.7
22.2
Southeast
581
-0.2
0.3
-16.1
22.7
South
395
-0.2
0.3
-17.3
23.7
Southwest
452
0.1
0.2
26.1
48.4
Northern Rockies
557
0.1
0.2
22.1
44.1
Northwest
West
304
0.0
0.3
4.4
50.6
CSN
Northeast
3,078
0.1
0.5
7.4
59.6
Ohio Valley
2,224
-0.1
0.5
-12.0
47.0
Upper Midwest
1,245
-0.1
0.5
-6.2
43.9
Southeast
2,117
0.0
0.2
-8.2
68.2
South
1,395
-0.2
0.3
-38.0
66.2
Nitrate
Southwest
973
-0.7
0.8
-71.7
82.2
Northern Rockies
642
-0.1
0.4
-14.7
70.2
Northwest
646
0.0
0.7
-1.5
89.6
West
1,173
-1.4
1.6
-66.2
72.7
IMPROVE
Northeast
1,862
0.1
0.3
30.0
>100
-------
Pollutant
Monitor
Network
Subregion
No. of
Obs
MB
(|jgm3)
3 S
3 m
CO
NMB
(%)
NME
(%)
Ohio Valley
956
-0.3
0.3
-47.3
57.2
Upper Midwest
819
-0.1
0.4
-11.5
56.6
Southeast
1,498
-0.1
0.2
-44.9
69.7
South
1,016
-0.2
0.3
-58.4
68.4
Southwest
3,795
-0.1
0.2
-81.5
89.2
Northern Rockies
2,218
0.0
0.2
-8.0
89.8
Northwest
1,915
-0.1
0.2
-31.7
97.7
West
2,364
-0.3
0.3
-62.8
76.1
CASTNet
Northeast
923
0.1
0.4
8.1
33.8
Ohio Valley
894
-0.1
0.5
-7.6
31.9
Upper Midwest
293
-0.1
0.4
-6.9
27.9
Total Nitrate
(NO3+HNO3)
Southeast
581
-0.2
0.5
-21.6
43.8
South
395
-0.4
0.5
-27.2
26.2
Southwest
452
-0.2
0.2
-23.7
35.7
Northern Rockies
557
-0.2
0.3
-27.3
38.9
Northwest
~
~
~
West
304
-0.5
0.5
-42.1
46.7
CSN
Northeast
3,086
0.2
0.3
88.8
>100
Ohio Valley
2,228
0.1
0.3
30.0
63.2
Upper Midwest
1,248
0.2
0.3
56.7
79.4
Southeast
2,024
0.0
0.2
19.7
84.1
South
1,399
0.0
0.2
7.4
68.3
Southwest
973
-0.1
0.3
-46.7
88.5
Northern Rockies
643
0.2
0.2
>100
>100
Northwest
646
0.1
0.3
39.1
>100
West
1,174
-0.3
0.5
-50.1
79.3
Ammonium
CASTNet
Northeast
923
0.0
0.1
-9.5
29.7
Ohio Valley
894
-0.1
0.2
-25.6
30.9
Upper Midwest
293
0.0
0.1
-7.8
29.7
Southeast
581
-0.1
0.1
-29.4
34.5
South
395
-0.1
0.1
-23.8
35.6
Southwest
452
0.0
0.1
-21.4
41.5
Northern Rockies
557
-0.1
0.1
-20.9
41.7
Northwest
~
~
~
West
304
-0.1
0.1
-39.6
54.9
Elemental
CSN
Northeast
3,137
0.0
0.3
-3.0
48.0
Carbon
Ohio Valley
2,221
-0.1
0.2
-17.8
37.5
96
-------
Monitor
Pollutant Network
Subregion
No. of
Obs
MB
(|jgm3)
3 S
3 m
CO
NMB
(%)
NME
(%)
Upper Midwest
1,322
0.0
0.2
-2.8
43.3
Southeast
1,748
-0.1
0.2
-18.9
42.3
South
1,201
0.0
0.2
-10.1
40.5
Southwest
956
0.0
0.3
-6.8
45.0
Northern Rockies
612
-0.1
0.2
-21.2
57.3
Northwest
615
0.2
0.4
22.1
62.4
West
1,143
-0.1
0.3
-9.6
38.3
IMPROVE
Northeast
1,844
0.0
0.1
7.3
43.4
Ohio Valley
942
0.0
0.1
-15.2
41.2
Upper Midwest
927
0.0
0.1
-11.1
39.9
Southeast
1,693
-0.1
0.1
-27.5
43.1
South
1,007
0.0
0.1
-20.3
43.9
Southwest
3,739
-0.1
0.1
-36.5
55.6
Northern Rockies
2,273
-0.1
0.1
-32.7
66.3
Northwest
1,833
0.0
0.2
3.3
78.6
West
2,310
-0.1
0.1
-23.9
62.8
CSN
Northeast
3,133
1.4
1.6
78.3
86.9
Ohio Valley
2,221
0.8
1.1
43.0
56.5
Upper Midwest
1,321
0.9
1.1
57.9
71.7
Southeast
1,746
0.9
1.2
36.7
50.9
South
1,201
0.9
1.4
48.7
72.1
Southwest
956
0.3
1.0
14.8
53.4
Northern Rockies
612
-0.1
1.0
-4.1
65.3
Northwest
615
2.0
2.4
73.7
91.1
West
1,143
0.8
1.6
27.2
51.8
Organic
Carbon IMPROVE
Northeast
1,852
0.8
0.9
74.2
85.2
Ohio Valley
942
0.9
1.1
62.4
77.8
Upper Midwest
928
0.5
0.8
48.9
70.6
Southeast
1,700
0.4
1.0
26.3
59.2
South
1,012
0.6
0.9
52.1
75.5
Southwest
3,787
0.0
0.5
-5.3
56.1
Northern Rockies
2,319
0.0
1.0
-3.0
67.9
Northwest
1,879
0.9
1.8
45.5
89.9
West
2,355
0.1
1.2
7.6
70.8
97
-------
S04 MB (ug/m3) for run CMAQ 2017gb_17j_12US2 for 20170101 to 20171231
units - ug/m3
coverage limit - 75^
* IMPROVE CSN ¦ CASTNET Weekly
Figure 4-7. Mean Bias (figm') of annual sulfate at monitoring sites in the continental U.S.
modeling domain.
units - ug/m3
coverage limit - 75%
IMPROVE CSN ¦ CASTNET Weekly
Figure 4-8. Mean Error (jignr3) of annual sulfate at monitoring sites in the continental U.S.
modeling domain.
S04 ME (ug/m3) for run CMAQ 2017gb 17i 12US2 for 20170101 to 20171231
98
-------
(%) for run CMAQ_2017gb_17j_12US2
units = %
coverage limit - 75%
>100
90
80
70
60
50
40
30
20
10
0
-10
-20
-30
-40
-50
-60
-70
-80
-90
<-100
A IMPROVE CSN ¦ CASTNET Weekly
Figure 4-9. Normalized Mean Bias (%) of annual sulfate at monitoring sites in the continental
U.S. modeling domain.
for 20170101 to 20171231
S04NMB
{%) for run CMAQ_2017gb_17j_12US2
units - %
coverage limit - 75%
> 100
90
80
70
60
50
40
30
20
10
0
Figure 4-10. Normalized Mean Error (%) of annual sulfate at monitoring sites in the continental
U.S. modeling domain.
for 20170101 to 20171231
¦» IMPROVE CSN
S04 NME
¦ CASTNET Weekly
99
-------
N03 MB (ug/m3) for run CMAQ_2017gb_17j_12US2 for 20170101 to 20171231
units = ug/m3
coverage limit - 75%
* IMPROVE CSN
Figure 4-11. Mean Bias (jignr3) of annual nitrate at monitoring sites in the continental U.S.
modeling domain.
N03 ME (ug/m3) for run CMAQ 2017gb_17j_12US2 for 20170101 to 20171231
a IMPROVE CSN
Figure 4-12. Mean Error (jigm"3) of annual nitrate at monitoring sites in the continental U.S.
modeling domain.
100
-------
{%) for run CMAQ 2017gb_17j_12US2
* IMPROVE CSN
for 20170101 to 20171231
units = %
coverage limit - 75%
>100
90
80
70
60
50
40
30
20
10
0
-10
-20
-30
-40
-50
-60
-70
-80
-90
<-100
N03NMB
Figure 4-13. Normalized Mean Bias (%) of annual nitrate at monitoring sites in the continental U.S.
modeling domain.
(%) for run CMAQ 2017gb 17j
units - %
coverage limit - 75%
>100
90
80
70
60
50
40
30
20
10
0
12US2 for 20170101 to 20171231
N03 NME
* IMPROVE CSN
Figure 4-14. Normalized Mean Error (%) of annual nitrate at monitoring sites in the continental
U.S. modeling domain.
101
-------
TN03 MB (ug/m3) for run CMAQ_2017gb_17j_12US2 for 20170101 to 20171231
units = ug/m3
coverage limit = 75%
CASTNET Weekly
Figure 4-15. Mean Bias (jigm 3) of annual total nitrate at monitoring sites in the continental U.S.
modeling domain.
TN03 ME (ug/m3) for run CMAQ_2017gb_17jJ2US2 for 20170101 to 20171231
units = ug/m3
coverage limit = 75%
a CASTNET Weekly
Figure 4-16. Mean Error (jigin 3) of annual total nitrate at monitoring sites in the continental U.S.
modeling domain.
102
-------
TN03 NMB {%) for run CMAQ_2017gb_17j_12US2 for 20170101 to 20171231
units = %
coverage limit = 75%
> 100
t
* CASTNET Weekly
Figure 4-17. Normalized Mean Bias (%) of annual total nitrate at monitoring sites in the continental
U.S. modeling domain.
TN03 NME (%) for run CMAQ_2017gb_17j_12US2 for 20170101 to 20171231
coverage limit = 75%
> 100
* CASTNET Weekly
Figure 4-18. Normalized Mean Error (%) of annual total nitrate at monitoring sites in the
continental U.S. modeling domain.
103
-------
NH4 MB (ug/m3) for run CMAQ 2017gb 17j_12US2 for 20170101 to 20171231
units - ug/m3
coverage limit = 75%
a CSN CASTNET Weekly
Figure 4-19. Mean Bias (jignr3) of annual ammonium at monitoring sites in the continental U.S.
modeling domain.
NH4 ME (ug/m3) for run CMAQ 2017gb 17j_12US2 for 20170101 to 20171231
units = ug/m3
coverage limit = 75%
a CSN CASTNET Weekly
Figure 4-20. Mean Error (ptgm 3) of annual ammonium at monitoring sites in the continental U.S.
modeling domain.
104
-------
NH4 NMB (%) for run CMAQ_2017gb_17j_12US2 tor 20170101 to 20171231
units = %
Vr4
A CSN CASTNET Weekly
Figure 4-21. Normalized Mean Bias (%) of annual ammonium at monitoring sites in the continental
U.S. modeling domain.
NH4 NME (%) for run CMAQ 2017gb17j_12US2 for 20170101 to 20171231
coverage limit = 75%
>100
90
80
,A
40
30
* CSN
CASTNET Weekly
Figure 4-22. Normalized Mean Error (%) of annual ammonium at monitoring sites in the
continental U.S. modeling domain.
105
-------
(ug/m3) for run CMAQ_2017gb_17j_12US2
for 20170101 to 20171231
EC MB
units = ug.'m3
coverage limit = 75%
A IMPROVE CSN
Figure 4-23. Mean Bias (jignr3) of annual elemental carbon at monitoring sites in the continental
U.S. modeling domain.
units - ug.'m3
coverage limit = 75%
(ug/m3) for run CMAQ_2017gb_17j_12US2
A IMPROVE CSN
for 20170101 to 20171231
EC ME
Figure 4-24. Mean Error (jigm 3) of annual elemental carbon at monitoring sites in the continental
U.S. modeling domain.
106
-------
(%) tor run CMAQ_2017gb_17j_12US2
for 20170101 to 20171231
EC NMB
units = %
coverage limit = 75%
* IMPROVE CSN
Figure 4-25. Normalized Mean Bias (%) of annual elemental carbon at monitoring sites in the
continental U.S. modeling domain.
units = %
coverage limit 75%
> 100
L
90
J
80
70
60
-
50
-
40
¦
30
1
20
B
¦
0
A IMPROVE CSN
Figure 4-26. Normalized Mean Error (%) of annual elemental carbon at monitoring sites in the
continental U.S. modeling domain.
(%) for run CMAQ_2017gb_17j_12US2
for 20170101 to 20171231
EC NME
107
-------
(ug/m3) lor run CMAQ_2017gb_17|.
units = ug/m3
coverage limit - 75%
>2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
-1
-1,2
-14
-1.6
-1.8
<-2
a IMPROVE CSN
12US2 for 20170101 to 20171231
OC MB
Figure 4-27. Mean Bias (jigm3) of annual organic carbon at monitoring sites in the continental U.S.
modeling domain.
units - ug/m3
coverage limit - 75%
B
>2
1.8
1.6
1.4
1.2
IMPROVE CSN
Figure 4-28. Mean Error (jigm3) of annual organic carbon at monitoring sites in the continental
U.S. modeling domain.
(ug/m3) for run CMAQ_2017gb_17j
OC ME
12US2 for 20170101 to 20171231
108
-------
(%) lor run CMAQ_2017gb_17|
* IMPROVE CSN
12US2 for 20170101 to 20171231
units = %
coverage limit - 75%
>100
90
80
70
60
50
40
30
20
10
0
-10
-20
-30
-40
-50
-60
-70
-80
-90
<-100
OC NMB
Figure 4-29. Normalized Mean Bias (%) of annual organic carbon at monitoring sites in the
continental U.S. modeling domain.
units - %
coverage limit _ 75%
* IMPROVE CSN
Figure 4-30. Normalized Mean Error (%) of annual organic carbon at monitoring sites in the
continental U.S. modeling domain.
OC NME (%) for run CMAQ 2017gb 17j 12US2 tor 20170101 to 20171231
109
-------
yesian space-time downs:1.. . ( .iel (downs
uality Estimates
5.1 Introduction
The need for greater spatial coverage of air pollution concentration estimates has grown in recent years as
epidemiology and exposure studies that link air pollution concentrations to health effects have become
more robust and as regulatory needs have increased. Direct measurement of concentrations is the ideal
way of generating such data, but prohibitive logistics and costs limit the possible spatial coverage and
temporal resolution of such a database. Numerical methods that extend the spatial coverage of existing
air pollution networks with a high degree of confidence are thus a topic of current investigation by
researchers. The downscaler model (DS) is the result of the latest research efforts by EPA for performing
such predictions. DS utilizes both monitoring and CMAQ data as inputs and attempts to take advantage
of the measurement data's accuracy and CMAQ's spatial coverage to produce new spatial predictions.
This chapter describes methods and results of the DS application that accompany this report, which
utilized ozone and PM2.5 data from AQS and CMAQ to produce predictions to continental U.S. 2010
census tract centroids for the year 2017.
5.2 Downscaler Model
DS develops a relationship between observed and modeled concentrations, and then uses that relationship
to spatially predict what measurements would be at new locations in the spatial domain based on the input
data. This process is separately applied for each time step (daily in this work) of data, and for each of the
pollutants under study (ozone and PM2.5). In its most general form, the model can be expressed in an
equation similar to that of linear regression:
y(s) = /?0 (s) + /?i x(s) + £(s) (Equation 1)
Where:
Y(s) is the observed concentration at point 5. Note that Y(s) could be expressed as Kt(s), where t indicates
the model being fit at time t (in this case, t=l, ...,365 would represent day of the year.)
x(s) is the point-level regressor based on the CMAQ concentration at point 5. This value is a weighted
average of both the gridcell containing the monitor and neighboring gridcells.
/?0(s) is the intercept, where /?0(s) = /?0 + /?0(s) is composed of both a global component /?0and a
local component /?0(s) that is modeled as a mean-zero Gaussian Process with exponential decay
is the global slope; local components of the slope are contained in the x(s) term.
£(s) is the model error.
DS has additional properties that differentiate it from linear regression:
1) Rather than just finding a single optimal solution to Equation 1, DS uses a Bayesian approach so that
uncertainties can be generated along with each concentration prediction. This involves drawing random
samples of model parameters from built-in "prior" distributions and assessing their fit on the data on the
order of thousands of times. After each iteration, properties of the prior distributions are adjusted to try to
improve the fit of the next iteration. The resulting collection of /?0 and /?x values at each space-time point
110
-------
are the "posterior" distributions, and the means and standard distributions of these are used to predict
concentrations and associated uncertainties at new spatial points.
2) The model is "hierarchical" in structure, meaning that the top-level parameters in Equation 1 (ie /?0(s),
x(s)) are actually defined in terms of further parameters and sub-parameters in the DS code. For
example, the overall slope and intercept is defined to be the sum of a global (one value for the entire
spatial domain) and local (values specific to each spatial point) component. This gives more flexibility in
fitting a model to the data to optimize the fit (i.e. minimize s(s)).
Further information about the development and inner workings of the current version of DS can be found
in Berrocal, Gelfand and Holland (2012)30 and references therein. The DS outputs that accompany this
report are described below, along with some additional analyses that include assessing the accuracy of the
DS predictions. Results are then summarized, and caveats are provided for interpreting them in the
context of air quality management activities.
5.3 Downscaler Concentration Predictions
In this application, DS was used to predict daily concentration and associated uncertainty values at the
2010 US census tract centroids across the continental U.S. using 2017 measurement and CMAQ data as
inputs. For ozone, the concentration unit is the daily maximum 8-hour average in ppb and for PM2.5 the
concentration unit is the 24-hour average in |j,g/m3.
5.3.1 Summary of 8-hour Ozone Results
Figure 5-1 summarizes the AQS, CMAQ and DS ozone data over the year 2017. It shows the 4th max
daily maximum 8-hour average ozone for AQS observations, CMAQ model predictions and DS model
results. The DS model estimated that for 2017, about 51% of the US Census tracts (36890 out of 72283)
experienced at least one day with an ozone value above the NAAQS of 70 ppb.
30 Berrocal, V., Gelfand, A., and D. Holland. Space-Time Data Fusion Under Error in Computer Model Output: An Application
to Modeling Air Quality. Biometrics. 2012. September; 68(3): 837-848. doi:10.1111/j.l541-0420.2011.01725.
Ill
-------
AQS
CMAQ
2017
4'th Max, Daily max
8-hour avg
ozone (ppb)
<-lnf,55]
(55,60]
¦ (60,65]
(65,70]
(70,75]
(75,80]
¦ (80,85]
¦ (85,90]
¦ (90, Inf]
Figure 5-1. Annual 4th max (daily max 8-hour ozone concentrations) derived from AQS, CMAQ
and DS data.
112
-------
5.3.2 Summary of PM2.5 Results
Figures 5-2 and 5-3 summarize the AQS, CMAQ and DS PM2.5 data over the year 2017. Figure 5-2
shows annual means and Figure 5-3 shows 98th percentiles of 24-hour PM2.5 concentrations for AQS
observations, CMAQ model predictions and DS model results. The DS model estimated that for 2017
about 21% of the US Census tracts (15486 out of 72283) experienced at least one day with a PM2.5 value
above the 24-hour NAAQS of 35 |j,g/m3. This 21% exceedance rate is substantially higher than the 16%
reported in the 2016 report
(https://ofmpub.epa.gov/rsig/rsigserver7data/FAQSD/docs/2016_DS_Annual_Report.pdf) due to the
increase in wildfire frequency in the northwestern US and associated high concentrations predicted by the
DS model in that region.
113
-------
AQS
CMAQ
2017
Annual mean,
24-hour avg
PM2.5 (ug/m3)
(0,3]
(3,5]
(5,8]
(8,10]
¦ (10,12]
(12,15]
(15,18]
¦ (18,lnf]
Figure 5-2. Annual mean PM2.5 concentrations derived from AQS, CMAQ and DS data.
114
-------
AQS
2017
98'th percentile,
24-hour avg
PM2.5 (ug/m3)
(0,10]
(10,15]
(15,20]
(20,25]
¦ (25,30]
(30,35]
(35,40]
¦ (40,45]
¦ (45,50]
¦ (50,lnf]
CMAQ
Figure 5-3. 98th percentile 24-hour average PM2.5 concentrations derived from AQS, CMAQ and
DS data.
115
-------
5.4 Downscaler Uncertainties
5.4.1 Standard Errors
As mentioned above, the DS model works by drawing random samples from built-in distributions during
its parameter estimation. The standard errors associated with each of these populations provide a measure
of uncertainty associated with each concentration prediction. Figure 5-4 shows the percent errors
resulting from dividing the DS standard errors by the associated DS prediction. The black dots on the
maps show the location of EPA sampling network monitors whose data was input to DS via the AQS
datasets (Chapter 2). The maps show that, in general, errors are relatively smaller in regions with more
densely situation monitors (ie the eastern US), and larger in regions with more sparse monitoring
networks (ie western states). These standard errors could potentially be used to estimate the probability
of an exceedance for a given point estimate of a pollutant concentration.
116
-------
% DS Error:
ozone
¦ (0.5]
¦ (5,10]
¦ (10,15]
¦ (15,20]
¦ (20,30]
(30,40]
(40,50]
¦ (50,75]
¦ (75,100]
% DS Error:
pm25
¦ (0.5]
¦ (5,10]
¦ (10,15]
¦ (15,20]
¦ (20,30]
(30,40]
(40,50]
¦ (50,75]
¦ (75,100]
Figure 5-4. Annual mean relative errors (standard errors divided by predictions) from the DS 2017
runs. The black dots show the locations of monitors that generated the AQS data used as input to
the DS model.
117
-------
5.4.2 Cross Validation
To check the quality of its spatial predictions, DS can be set to perform "cross-validation" (CV), which
involves leaving a subset of AQS data out of the model run and predicting the concentrations of those left
out points. The predicted values are then compared to the actual left-out values to generate statistics that
provide an indicator of the predictive ability. In the DS runs associated with this report, 10% of the data
was chosen randomly by the DS model to be used for the CV process. The resulting CV statistics are
shown below in Table 5-1.
Pollutant
# Monitors
Mean Bias
RMSE
Mean Coverage
PM2.5
958
0.21
3.35
0.95
03
1275
-0.03
4.3
0.96
Table 5-1. Cross-validation statistics associated with the 2017 DS runs.
The statistics indicated by the columns of Table 5-1 are as follows:
- Mean Bias: The bias of each prediction is the DS prediction minus the AQS value. This column
is the mean of all biases across the CV cases.
- Root Mean Squared Error (RMSE): The bias is squared for each CV prediction, then the square
root of the mean of all squared biases across all CV predictions is obtained.
- Mean Coverage: A value of 1 is assigned if the measured AQS value lies in the 95% confidence
interval of the DS prediction (the DS prediction +/- the DS standard error), and 0 otherwise. This
column is the mean of all those 0's and l's.
5.5 Summary and Conclusions
The results presented in this report are from an application of the DS fusion model for characterizing
national air quality for Ozone and PM2.5. DS provided spatial predictions of daily ozone and PM2.5 at
2010 U.S. census tract centroids by utilizing monitoring data and CMAQ output for 2017. Large-scale
spatial and temporal patterns of concentration predictions are generally consistent with those seen in
ambient monitoring data. Both Ozone and PM2.5 were predicted with lower error in the eastern versus
the western U.S., presumably due to the greater monitoring density in the east.
An additional caution that warrants mentioning is related to the capability of DS to provide predictions at
multiple spatial points within a single CMAQ grid cell. Care needs to be taken not to over-interpret any
within-grid cell gradients that might be produced by a user. Fine-scale emission sources in CMAQ are
diluted into the grid cell averages, but a given source within a grid cell might or might not affect every
spatial point contained therein equally. Therefore DS-generated fine-scale gradients are not expected to
represent actual fine-scale atmospheric concentration gradients, unless possibly where multiple monitors
are present in the grid cell.
118
-------
Apper
Acronyms
ARW
Advanced Research WRF core model
BEIS
Biogenic Emissions Inventory System
BlueSky
Emissions modeling framework
CAIR
Clean Air Interstate Rule
CAMD
EPA's Clean Air Markets Division
CAP
Criteria Air Pollutant
CAR
Conditional Auto Regressive spatial covariance structure (model)
CARB
California Air Resources Board
CEM
Continuous Emissions Monitoring
CHIEF
Clearinghouse for Inventories and Emissions Factors
CMAQ
Community Multiscale Air Quality model
CMV
Commercial marine vessel
CO
Carbon monoxide
CSN
Chemical Speciation Network
DQO
Data Quality Objectives
EGU
Electric Generating Units
Emission Inventory
Listing of elements contributing to atmospheric release of pollutant
substances
EPA
Environmental Protection Agency
EMFAC
Emission Factor (California's onroad mobile model)
FAA
Federal Aviation Administration
FDDA
Four Dimensional Data Assimilation
FIPS
Federal Information Processing Standards
HAP
Hazardous Air Pollutant
HMS
Hazard Mapping System
ICS-209
Incident Status Summary form
IPM
Integrated Planning Model
ITN
Itinerant
LSM
Land Surface Model
MOBILE
OTAQ's model for estimation of onroad mobile emissions factors
MODIS
Moderate Resolution Imaging Spectroradiometer
MOVES
Motor Vehicle Emission Simulator
NEEDS
National Electric Energy Database System
NEI
National Emission Inventory
NERL
National Exposure Research Laboratory
NESHAP
National Emission Standards for Hazardous Air Pollutants
NH
Ammonia
NMIM
National Mobile Inventory Model
NONROAD
OTAQ's model for estimation of nonroad mobile emissions
NO
Nitrogen oxides
OAQPS
EPA's Office of Air Quality Planning and Standards
OAR
EPA's Office of Air and Radiation
ORD
EPA's Office of Research and Development
119
-------
ORIS Office of Regulatory Information Systems (code) - is a 4 or 5 digit
number assigned by the Department of Energy's (DOE) Energy
Information Agency (EIA) to facilities that generate electricity
ORL One Record per Line
OTAQ EPA's Office of Transportation and Air Quality
PAH Polycyclic Aromatic Hydrocarbon
PFC Portable Fuel Container
PM2.5 Particulate matter less than or equal to 2.5 microns
PM10 Particulate matter less than or equal to 10 microns
PMc Particulate matter greater than 2.5 microns and less than 10 microns
Prescribed Fire Intentionally set fire to clear vegetation
RIA Regulatory Impact Analysis
RPO Regional Planning Organization
RRTM Rapid Radiative Transfer Model
SCC Source Classification Code
SMARTFIRE Satellite Mapping Automatic Reanalysis Tool for Fire Incident
Reconciliation
SMOKE Sparse Matrix Operator Kernel Emissions
TSD Technical support document
VOC Volatile organic compounds
VMT Vehicle miles traveled
Wildfire Uncontrolled forest fire
WRAP Western Regional Air Partnership
WRF Weather Research and Forecasting Model
120
-------
United States Office of Air Quality Planning and Standards Publication No. EPA-454/R-20-005
Environmental Protection Air Quality Assessment Division August 2020
Agency Research Triangle Park, NC
------- |