Clean Air Act Section 211(v)(l)
Anti-backs tiding Study
United States
Environmental Protection
^1	Agency

-------
Clean Air Act Section 211(v)(l)
Anti-backs tiding Study
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
£%	United States
Environmental Protection
^1	Agency
EPA-420-R-20-008
May 2020

-------
Table of Contents
Table of Contents	1
1	Glossary of Acronyms	3
2	Executive Summary	5
3	Introduction	7
3.1	Statutory Requirement for Anti-backsliding Study	7
3.2	Background on Renewable Fuel Standard Program	7
3.3	Overview of Vehicle and Engine Emissions	7
4	Overview of Study Approach	8
5	Renewable Fuel Scenarios	10
5.1	2016 "With-RFS" Scenario	10
5.2	2016 "Pre-RFS" Scenario	12
6	Emissions Inventories and Air Quality Modeling	14
6.1	Overview of Analysis Methods	14
6.2	Onroad Mobile Emissions Inventory	14
6.2.1 Onroad Mobile Emissions Inventory Methodology	16
6.3	Nonroad Mobile Emissions Inventory	17
6.3.1 Nonroad Mobile Emissions Inventory Methodology	18
6.4	Onroad and Nonroad Emissions Inventory Summaries	19
6.5	Other Inventory Sectors	28
6.6	Emissions Modeling	28
6.7	Air Quality Modeling Methodology	30
6.7.1	Air Quality Model	30
6.7.2	Model Domain and Configuration	31
6.7.3	Model Inputs	33
6.7.4	Ozone and PM2.5 Fused Fields	34
6.7.5	CMAQ Evaluation	35
7	Air Quality Modeling Results	35
7.1	Ozone	35
7.2	Particulate Matter	37
7.3	Nitrogen Dioxide	41
7.4	Carbon Monoxide	42
1

-------
7.5 Air Toxics (acetaldehyde, acrolein, benzene, 1,3-butadiene, formaldehyde,
naphthalene)	44
7.5.1	Acetaldehyde	45
7.5.2	Acrolein	46
7.5.3	Benzene	48
7.5.4	1,3-Butadiene	49
7.5.5	Formaldehyde	50
7.5.6	Naphthalene	52
8	Study Limitations and Uncertainties	53
8.1	Study Scope and Design	53
8.2	Data and Model Limitations and Uncertainties	54
9	Appendix A: Emissions Modeling and Air Quality Modeling	56
9.1	Chemical Mechanisms in Air Quality Modeling	56
9.1.1	Acetaldehyde	57
9.1.2	Organic Aerosols	59
9.1.3	Ozone	65
9.1.4	Uncertainties Associated with Chemical Mechanisms	66
9.2	Creating Ozone and PM2.5 Fused Fields Based on Observations and Model Surfaces 67
9.3	Air Quality Model Performance Evaluation	69
9.4	Monthly Air Quality Difference Maps	71
9.5	National Emission Inventories for Criteria and Toxic Pollutants	88
2

-------
1 Glossary of Acronyms
ASTM	American Society of Testing and Materials
CAA	Clean Air Act
CAP	Criteria Air Pollutants
CARB	California Air Resources Board
CG	Conventional Gasoline
CMAQ	Community Multiscale Air Quality Model
CMAS	Community Modeling and Analysis System
CNG	Compressed Natural Gas
CONUS	Continental United States
EIA	Energy Information Administration
EISA	Energy Independence and Security Act
EPAct	Energy Policy Act of 2005
eVNA	enhanced Veronoi Neighbor Average
HAP	Hazardous Air Pollutants
HOTELING Hours of extended idle
LPG	Liquid Petroleum Gas
MCIP	Meteorology-Chemistry Interface Processor
MOVES	MOtor Vehicle Emission Simulator
MSAT	Mobile Source Air Toxics
NATA	National Air Toxics Assessment
NOAA	National Oceanic and Atmospheric Administration
OC	Organic Carbon
OM	Organic Matter
PAN	Peroxyacetyl Nitrate
PM	Particulate Matter
PMC	Coarse Particulate Matter
POA	Primary Organic Aerosols
RFG	Reformulated Gasoline
RFS	Renewable Fuel Standard
RPD	Rate-per-di stance
RPH	Rate-per-hour
RPP	Rate-per-profile
RPV	Rate-per-vehicle
RVP	Reid Vapor Pressure
SCC	Source Classification Code
SMAT-CE Software for the Modeled 8-30 Attainment Test - Community Edition
SMOKE	Sparse Matrix Operator Kernel Emissions
SMPS	Scanning Mobility Particle Sizer
SO A	Secondary Organic Aerosols
TSD	Technical Support Document
VMT	Vehicle Miles Traveled
VOC	Volatile Organic Compound
3

-------
VPOP	Vehicle Population
WRF	Weather Research Forecasting Model
4

-------
2 Executive Summary
As required by Clean Air Act section 21 l(v)(l), this study examined the impacts on air quality as
a result of changes in vehicle and engine emissions resulting from required renewable fuel
volumes under the Renewable Fuel Standard (RFS). Specifically, this study compared two
scenarios for calendar year 2016, one with actual ethanol and biodiesel volumes ("with-RFS"
scenario) and another with ethanol and biodiesel use approximating 2005 levels (the "pre-RFS"
scenario).
The "with-RFS" scenario assumed 10 percent ethanol (E10) was used nationwide in all onroad
and nonroad gasoline-fueled vehicles and engines, and biodiesel was used at a five percent blend
(B5) in all onroad diesel vehicles nationwide. This was compared to the "pre-RFS" scenario,
which assumed E10 usage only in the 2016 reformulated gasoline (RFG) areas and no biodiesel
usage (except in California). In California, we assumed that the "pre-RFS" scenario was the
same as the "with-RFS" scenario and therefore did not model any emissions changes there.
Everything was held constant between the two scenarios except the fuel supplies for onroad and
nonroad engines; "upstream" emissions from producing, storing, and transporting fuels and
feedstocks were also held constant in both scenarios at 2016 levels.
As described in the following paragraphs, this study has a number of limitations. It is narrowly
focused on the impacts of statutorily-required renewable fuel volumes on concentrations of
criteria and toxic pollutants due to changes in vehicle and engine emissions; this study is not an
examination of the lifecycle impacts of renewable fuels on air quality, greenhouse gases, or other
environmental impacts. Clean Air Act section 21 l(v)(l)(A) specifically refers to air quality
impacts "as a result of changes in vehicle and engine emissions."
This anti-backsliding study examines the impacts of required renewable volumes as compared to
a hypothetical case where renewable fuel usage in 2016 was approximately the same as it had
been in 2005, before EPAct was enacted. Because of the very limited data on 2005 fuel
properties and their spatial distribution, the "pre-RFS" scenario is only a general approximation
of 2005 fuels. In addition, in the absence of consistent and reliable data on biodiesel use across
the country, this study assumes in the "with-RFS" scenario that biodiesel was at a B5 blend level
in all onroad diesel fuel nationwide; it does not capture the impacts of higher or lower biodiesel
blends that may have been used in specific areas. As a result, the air quality modeling results are
illustrative at a broad geographic scale (rather than being locally specific). For this reason, this
study does not attempt to estimate changes in the status of individual areas' attainment of the
National Ambient Air Quality Standards. With respect to estimating the effects of renewable
fuels on emissions, data for nonroad gasoline engines and diesel vehicles and engines is much
more limited than the data available for onroad gasoline vehicles.
This study examines impacts for a single retrospective year (2016). By analyzing calendar year
2016, EPA was able to use an existing modeling platform that includes known renewable fuel
volumes for 2016 and fuel properties based on actual data (aggregation of refinery batch reports
and fuel surveys). The study does not project future renewable fuel volumes and their impacts,
and thus it does not account for the impacts of the Tier 3 Motor Vehicle Emissions and Fuel
Standards, which took effect in 2017. These Tier 3 standards have lowered the sulfur content of
5

-------
gasoline and tightened the emissions standards for onroad motor vehicles, resulting in lower
emissions of criteria and toxic pollutants and precursors in 2017 and into the future as more Tier
3-compliant vehicles enter the fleet. The Tier 3 standards are projected to reduce concentrations
of ozone, PM2.5, NO2, toxics (such as acetaldehyde, formaldehyde, acrolein, benzene, 1,3-
butadiene, and naphthalene), and other pollutants into the future. In addition, the analysis year of
2016 used in this study does not reflect the full turnover of the diesel fleet to the most recent
highway standards. Because emissions data does not indicate that biodiesel affects emissions
from engines subject to the most recent standards, this study reflects emissions changes from
older engines that would be projected to decline in the future as the fleet turns over.
Despite these limitations, this study satisfies the requirements of Clean Air Act section 21 l(v)(l)
to determine the air quality impacts of changes in vehicle and engine emissions associated with
required renewable fuel volumes.
Compared to the "pre-RFS" scenario, the 2016 "with-RFS" scenario increased ozone
concentrations (eight-hour maximum average) across the eastern U.S. and in some areas in the
western U.S., with some decreases in localized areas. In the 2016 "with-RFS" scenario,
concentrations of fine particulate matter (PM2.5) were relatively unchanged in most areas, with
increases in some areas and decreases in some localized areas. The 2016 "with-RFS" scenario
increased concentrations of nitrogen dioxide (NO2) across the eastern U.S. and in some areas in
the western U.S., with larger increases in some urban areas. The 2016 "with-RFS" scenario
decreased concentrations of carbon monoxide (CO) across the eastern U.S. and in some areas in
the western U.S., with larger decreases in some areas.
Compared to the "pre-RFS" scenario, the 2016 "with-RFS" scenario increased concentrations of
acetaldehyde across much of the eastern U.S. and some areas in the western U.S., and resulted in
widespread increases in formaldehyde concentrations. The 2016 "with-RFS" scenario decreased
benzene concentrations across most of the U.S., as compared to the "pre-RFS" scenario. The
2016 "with-RFS" scenario also resulted in decreased concentrations of 1,3-butadiene in many
urban areas. The 2016 "with-RFS" scenario resulted in geographically limited increases and
decreases in concentrations of acrolein and naphthalene.
6

-------
3 Introduction
3.1	Statutory Requirement for Anti-backsliding Study
In 2007, the Energy Independence and Security Act (EISA) amended the Clean Air Act (CAA)
to include a requirement (CAA section 21 l(v)(l)) that EPA complete a study to determine
whether the renewable fuel volumes required by CAA section 21 l(o) would adversely impact air
quality as a result of changes in vehicle and engine emissions. This required study is commonly
known as the "anti-backsliding study." Section 21 l(v)(l)(B) requires the study to include
consideration of different blend levels, types of renewable fuels, and available vehicle
technologies, as well as appropriate national, regional, and local air quality control measures.
After considering the results of the study, EPA must either (1) promulgate fuel regulations to
implement appropriate measures to mitigate, to the greatest extent achievable, any adverse
impacts on air quality; or (2) determine that no such measures are necessary (CAA section
21 l(v)(2)).
3.2	Background on Renewable Fuel Standard Program
The Renewable Fuel Standard (RFS) program (CAA section 21 l(o)) was created by the Energy
Policy Act of 2005 (EPAct) and expanded by EISA in 2007. CAA section 21 l(o) establishes
targets for renewable fuel volumes and requires EPA to set volume requirements annually. In
2010 EPA finalized a rulemaking putting in place the regulations to implement 21 l(o) for
EISA.1 In doing so EPA projected the potential renewable fuel types and volumes that at the
time we thought might be used to meet the statutory requirement. However, since that time it
has become clear that those volumes were not attainable, and EPA has been using other
authorities provided in 21 l(o) to reduce the volumes to achievable levels in subsequent
rulemakings on an annual basis.
3.3	Overview of Vehicle and Engine Emissions
Vehicles and engines contribute to air pollution through both exhaust and evaporative emissions.
Exhaust emissions include hydrocarbons (HC), nitrogen oxides (NOx), particulate matter (PM),
carbon monoxide (CO), sulfur dioxide (SO2), and various toxic pollutants, such as benzene,
acetaldehyde, and formaldehyde. Evaporative emissions are hydrocarbon compounds, some of
which are toxic pollutants (e.g., benzene).
Vehicle and engine emissions depend on the design and functionality of the engine and the
associated exhaust and evaporative emission controls, in concert with the properties of the fuel
on which it is operating. Relevant gasoline properties include fuel-content parameters as well as
bulk properties. Fuel-content parameters include levels of oxygenate, ethanol, olefins, aromatics,
benzene, and sulfur. Bulk properties include vapor pressure and distillation properties, expressed
1 Regulation of Fuels and Fuel Additives: Changes to Renewable Fuel Standard Program, 75 FR 14799-14808,
March 26, 2010.
7

-------
as temperatures (T50, T90) or as volumes evaporated at specific distillation temperatures (E200,
E300). For diesel fuel, properties relevant to emissions include sulfur content and biodiesel
content. The impact of renewable fuels on vehicle and engine emissions results from differences
in these fuel properties.
EPA's emissions model, MOtor Vehicle Emission Simulator (MOVES), estimates the emissions
from onroad vehicles and nonroad engines, considering fuel properties and many other factors.
MOVES models the impact of ethanol and a wide range of other gasoline properties on exhaust
and evaporative emissions in onroad gasoline vehicles, and models the impact of biodiesel
blends on the emissions of diesel onroad vehicles.2'3'4 Note that MOVES does not model any
emissions effects of biodiesel in onroad heavy-duty engines that are model year 2007 and later
because no significant and consistent effects have been observed.2 For nonroad equipment,
MOVES models the HC, CO, and NOx emissions impacts of oxygenate (such as ethanol) for
nonroad gasoline engines, as well as the impacts on toxics emissions.5'6 MOVES does not
model a fuel effect on direct PM emissions from nonroad equipment due to insufficient data.
Also, MOVES does not model emissions impacts of biodiesel on nonroad diesel equipment.
Little data on emissions in nonroad engines using biodiesel exists, and an EPA analysis of this
limited data could not determine emission effects or conclude with confidence that nonroad
engines respond to biodiesel similarly to highway engines.7
4 Overview of Study Approach
This study assessed the air quality impact of vehicle and engine emissions in 2016 under two fuel
scenarios:
•	"Pre-RFS": Renewable fuel use at approximately 2005 (pre-EPAct) levels
•	"With-RFS": Renewable fuel use at 2016 levels
2	USEPA (2016). Fuel Effects on Exhaust Emissions from On-road Vehicles in MOVES2014. EPA-420-R-16-001.
Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection
Agency, Ann Arbor, MI. February 2016. https://nepis.epa. gov/Exe/ZyPDF.cgi?Dockey=P10005W2.pdf
3	USEPA (2016). Air Toxic Emissions from On-road Vehicles in MOVES2014. EPA-420-R-16-016. Assessment and
Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection Agency, Ann Arbor,
MI. November 2016. httPs://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100PUNQ.pdf
4	USEPA (2014). Evaporative Emissions from On-road Vehicles in MOVES2014. EPA-420-R-14-014. Assessment
and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection Agency, Ann
Arbor, MI. September 2014. httPs://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100KB5V.pdf
5	USEPA (2005). Exhaust Emission Effects of Fuel Sulfur and Oxygen on Gasoline Nonroad Engines. EPA-420-R-
05-016. Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental
Protection Agency, Ann Arbor, MI. December 2005. https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P1004L80.pdf
6	USEPA (2018). Speciation Profiles and Toxic Emission Factors for Nonroad Engines in MOVES2014b. EPA-420-
R-18-011. Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental
Protection Agency, Ann Arbor, MI. July 2018. https://nepis.epa. gov/Exe/ZyPDF.cgi?Dockey=P100UXK7.pdf
7	USEPA (2002).A Comprehensive Analysis of Biodiesel Impacts on Exhaust Emissions (Draft Report). EPA-420-
P-02-001. Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental
Protection Agency, Ann Arbor, MI. October 2002. https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P.1.001.ZAO.pdf
8

-------
Everything in these two scenarios was held constant except the fuels; for example, the vehicle
and engine population, activity, base emission rates, and meteorological data were the same in
both scenarios (reflecting calendar year 2016 conditions).
As described in more detail in Section 5.2, the "pre-RFS" fuel scenario assumed for 2016 that
vehicles used E10 in Reformulated Gasoline (RFG) areas and E0 everywhere else, which
approximates the ethanol usage patterns before EPAct was enacted in 2005. Except in
California, the "pre-RFS" scenario did not assume any biodiesel use. The 2016 "with-RFS"
scenario (described in Section 5.1) used EPA's 2016 air quality modeling platform (2016v7.2
beta)8, including its fuel supply and emission inventory.
By analyzing calendar year 2016, EPA was able to use this existing modeling platform that
includes known renewable fuel volumes for 2016 and fuel properties based on actual data
(aggregation of refinery batch reports and fuel surveys). However, this retrospective analysis for
2016 does not account for the effects of the Tier 3 Motor Vehicle Emissions and Fuel
Standards,9 which took effect in 2017.
Except for California, the 2016 mobile source inventory was generated using MOVES2014b, the
latest public version available when the study was initiated in 2019.10 EPA's MOtor Vehicle
Emission Simulator (MOVES) is a state-of-the-science emission modeling system that estimates
emissions for mobile sources at the national, county, and project level for criteria air pollutants,
greenhouse gases, and air toxics. MOVES computes exhaust emissions from onroad vehicles
and nonroad equipment11'12 and the evaporative emissions from onroad vehicles and nonroad
equipment.13'14'15 MOVES is the standard tool used to generate U.S. onroad and nonroad
emission inventories. As detailed in Sections 5 and 6, California fuels were the same in both
scenarios and the mobile source inventories for California were developed separately.
8	USEPA (2019). Technical Support Document: Preparation of Emissions Inventories for the Version 7.2 2016
North American Emissions Modeling Platform. Office of Air Quality Planning and Standards, U.S. Environmental
Protection Agency, Research Triangle Park, NC. September 2019. https://www.epa.gov/sites/prodiietion/files/20.1.9-
09/documents/20.1.6v7.2_regionalhaze_eniismod_tsd_508.pdf
9	79 FR 23414, April 28, 2014.
10	https://www.epa.gov/moves/latest-version-motor-vehicle-eniission-simnlator-moves
11	USEPA (2016). Fuel Effects on Exhaust Emissions from On-road Vehicles in MOVES2014. EPA-420-R-16-001.
Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection
Agency, Ann Arbor, MI. February 2016. https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100Q5W2.pdf
12	USEPA (2005). Exhaust Emission Effects of Fuel Sulfur and Oxygen on Gasoline Nonroad Engines. EPA-420-R-
05-016. Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental
Protection Agency, Ann Arbor, MI. December 2005. https://nepis.epa. gov/Exe/ZyPDF.cgi?Dockey=P1004L80.pdf
13	USEPA (2014). Evaporative Emissions from On-road Vehicles in MOVES2014. EPA-420-R-14-014. Assessment
and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection Agency, Ann
Arbor, MI. September 2014. httPs://nepis.epa.gov/Exe/ZyPDF.cgi?Dockev=P100KB5V.pdf
14	USEPA (2010). Nonroad Evaporative Emission Rates. EPA-420-R-10-021. Assessment and Standards Division,
Office of Transportation and Air Quality, U.S. Environmental Protection Agency, Ann Arbor, MI. July 2010.
https://nepis.epa.gov/Exe/Zv PDF.cgi?Dockev=P.1.008201.pdf
15	USEPA (2004). Refueling Emissions for Nonroad Engine Modeling. EPA-420-P-04-013. Assessment and
Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection Agency, Ann Arbor,
MI. April 2004. https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P.1.0002.lU.pdf
9

-------
Consistent with the requirements of CAA section 21 l(v)(l)(A), which refers specifically to
adverse air quality impacts of "changes in vehicle and engine emissions," this study did not
consider any changes to "upstream" emissions, i.e., emissions associated with fuel production
and distribution. As a result, this study does not represent a comprehensive consideration of the
air quality impacts of required renewable fuel volumes.
This study assessed the changes in emissions from motor vehicles and nonroad engines and
equipment (as estimated by MOVES) and used the Community Multiscale Air Quality model
(CMAQ) to estimate the resulting impacts on concentrations of ozone, PM, NO2, CO, and some
air toxics (including acetaldehyde, acrolein, benzene, 1,3-butadiene, formaldehyde, and
naphthalene). CMAQ, a photochemical model, allows us to understand how changes in
emissions impact air quality, especially for those pollutants which are formed secondarily and/or
are transported long distances.
This study satisfies the CAA section 21 l(v)(l)(A) requirement to determine the impacts of
required renewable fuel volumes, by comparing two scenarios with known renewable fuel
volumes (i.e., before and after the renewable fuel requirements). Other potential study
approaches would have involved highly uncertain estimates of fuel volumes and would have
been less informative. As required by CAA section 21 l(v)(l)(B), this study considered different
types and blends of renewable fuels (e.g., ethanol and biodiesel) and available vehicle
technologies (i.e., all vehicle and nonroad engine technologies in the 2016 fleet). The 2016 air
quality modeling platform includes national, regional, and local air quality control measures that
were relevant for that year.
5 Renewable Fuel Scenarios
5.1 2016 "With-RF S" Scenario
The "with-RF S" scenario used the beta version of the 2016 air quality modeling platform
(2016v7.2 beta), including the MOVES2014b fuel supply. The model has 11 fuel regions based
on the general structure of the gasoline distribution system and overlays additional detail for state
and local fuel controls; this results in a total of 22 sets of fuel properties used to compute fuel-
based emission adjustments. Table 5.1 gives summary information on these fuel regions. The
properties in these regions were determined from detailed fuel property data that EPA collects
from refiners as part of gasoline production compliance batch reporting, and they were validated
against survey data in local gasoline markets where available. Development of the MOVES fuel
supply is described in detail in the MOVES Fuel Supply Defaults technical report.16
16 USEPA (2018). Fuel Supply Defaults: Regional Fuels and the Fuel Wizard in MOVES2014b. EPA-420-R-18-008.
Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection
Agency, Ann Arbor, MI, July 2018. https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100UXDZ.pdf
10

-------
Table 5.1 Summary Description of MOVES2Q14 Fuel Regions

Base
Region
ID#

Maximum
E10 RVP
Waiver
Status
(01=No 1-psi
Waiver)
Minimum
Region II)
Base Region Name
Summer RVP
(psi) or 00 for
ASTM
Ethanol
Volume,
%
100000000


0.0
0
0
100010000


0.0
1
0
170000000
1
East Coast
7.0
0
0
178000000


7.8
0
0
178010000


7.8
1
0
200000000


0.0
0
0
270000000
2
Midwest
7.0
0
0
278000000
7.8
0
0
278010000


7.8
1
0
300000000


0.0
0
0
370000000
3
South
7.0
0
0
370010000


7.0
1
0
400000000
4
North
0.0
0
0
500000000
5
Rocky Mtns
0.0
0
0
578000000
7.8
0
0
600000000
6
CA/NV/AR/A11 Others
0.0
0
0
678000000
7.8
0
0
1170011000
11
East Coast RFG
7.0
1
10
1270011000
12
MD/VA RFG
7.0
1
10
1370011000
13
Texas RFG
7.0
1
10
1470011000
14
Midwest RFG
7.0
1
10
1570011000
15
California
7.0
1
10
Renewable fuel volumes for the 2016 "with-RFS" scenario were based on the simplifying
assumption that all non-E85 gasoline was E10 (both conventional gasoline (CG) and
reformulated gasoline (RFG)) nationwide. This was reasonable because volumes of other blends
(E0 and El5) were small and we do not have consistent and reliable data on their volume or
location.
Similarly, biodiesel was assumed to be at a B5 blend level (5 percent biodiesel) in all diesel
nationwide. We lack consistent and reliable data on biodiesel across the country, and this is a
reasonable simplifying assumption that is generally consistent with aggregate usage figures. In
addition, under ASTM D975, blends up to 5% can be labeled as "diesel fuel," confounding the
ability to determine where biodiesel is used.
A copy of the fuel formulation table used with MOVES for the 2016v7.2 beta platform is
available as part of the online supplemental materials.17
17 https://www.epa.gOY/renewable-fuel-standard-program/anti-backsliding-determination-aiid-stiidY
11

-------
5.2 2016 "Pre-RFS" Scenario
Because EPAct, which created the RFS program, was signed into law in late 2005, ethanol and
biodiesel consumption from that year was chosen for the "pre-RFS" scenario. EIA's Monthly
Energy Review presents U.S. ethanol consumption volumes on a monthly and annual basis.18
Computing the annualized fourth quarter rate for 2005 gives a volume of 4.5 billion gallons,
which closely matches the amount of ethanol required to blend all 2016 RFG, including
California gasoline, to E10 (about 4.7 billion gallons).19 Preferential use of ethanol in RFG is
reasonable because these areas invested heavily in ethanol blending infrastructure and logistics
leading up to and immediately following EPAct, quickly converting over entirely from MTBE
use to E10 use.20 21 While there likely was some continued use of E10 in some non-RFG areas in
this timeframe, the market was changing quickly and we do not have good data on precisely
which areas did and did not have E10. Allocating the E10 to RFG areas represents a reasonable
approximation of where ethanol was primarily used in this timeframe. Figure 5.1 shows the E10
RFG areas for this scenario.
w ¦iirA m
f HH [it f
IL	wv
H	« MO* H V ^
¦ *rl -I «r
Hl «.	TX	^
- 4 r ™
Reformulated area
Conventional area
Figure 5.1 RFG areas where E10 was allocated in the "pre-RFS" scenario
18	US Energy Information Administration. Monthly Energy' Review, Table 10.3. Retrieved September 2018.
19	US Energy Information Administration. U.S. Prime Supplier Sales Volumes of Petroleum Products, Table 2.
Retrieved April 2018.
2^US EPA Office of Transportation and Air Quality. Fuels Trends Report: Gasoline 1995-2005, Figure 6. EPA-420-
R-08-002. Januaty 2008.
21 US EPA Office of Transportation and Air Quality. Fuels Trends Report: Gasoline 2006-2016, Figure 8. EPA-420-
R-17-005. October 2017.
12

-------
Therefore, in order to create the "pre-RFS" scenario for 2016, the 2016 "with-RFS" fuel supply
was modified to make all CG E0 while all RFG remained E10 (including all California counties).
The market share of E85 blends was left unmodified (i.e., the same as 2016 "with-RFS" levels)
as a simplifying assumption, because they represent a small volume in the context of the overall
fuel supply. Other blend levels, such as E15, were assumed to be zero in the 2016 "pre-RFS"
scenario because El 5 was not in use in 2005.
According to the EIA Monthly Energy Review, biodiesel use in 2005 was 91 million gallons.22
Biodiesel use is modeled in MOVES as a single blend level nationwide, so dividing this volume
by the total 2016 onroad diesel use of approximately 45 billion gallons results in a blend level of
0.2%.23 For the purpose of the "pre-RFS" scenario in this analysis, this blend level was treated as
B0 (zero) in all states except California. We maintained California at the B5 blend level in the
"pre-RFS" scenario under the assumption that the state Low Carbon Fuel Standard24 would have
maintained biodiesel blending there even in the absence of EPAct. The fuel volume scenarios are
summarized in Table 5.2.
Removing ethanol from all 2016 CG required adjustment of several other fuel properties to
account for the fact that ethanol blending increases RVP, decreases distillation mid-point
temperature (T50), and dilutes aromatics and other constituents. These adjustments were done
using the E10 factors shown in Table 4 of the MOVES Fuel Supply Defaults technical report,
reproduced here as Table 5.3. These are applied as additive adjustments, and the sign has been
reversed here relative to the technical report because the scenario involved removing E10 from
CG. For RVP, an adjustment of 1.0 psi downward was applied in both summer and winter when
going from E10 to E0, except for summertime areas with "no waiver" flags in the MOVES fuel
supply (see Table 5.1). In these cases, we assumed that, in the absence of ethanol, additional
butanes/pentanes would be used in these fuels up to the allowed RVP limit. No changes were
made to RFG fuels (including California). A copy of the fuel formulation table modified for use
in the "pre-RFS" scenario is available as part of the online supplemental materials.25
Table 5.2 Summary of Fuel Cases (2016 Calendar Year)
Scenario
E0
E10
Biodiesel
2016 "With-RFS"
None
All gasoline
B5 in all onroad diesel
2016 "Pre-RFS"
All conventional gasoline
All reformulated
gasoline
B5 in California; B0
elsewhere
22	US Energy Information Administration. Monthly Energy Review, Table 10.4. Retrieved September 2018.
23	US Energy Information Administration. Monthly Energy Review, Tables 3.5 and 3.7c. Retrieved February 2020.
24	Low Carbon Fuel Standard Regulation, title 17, California Code of Regulations, sections 95480-95503.
25	https://www.epa.gov/renewable-fuel-staiidard-program/anti-backsliding-determination-aiid-stiidY
13

-------
Table 5.3 MOVES2Q14 Fuel Property Adjustment Factors Moving from ElO to EO
Fuel
RVP
(psi)
Sulfur
(ppm)
Aromatics
(vol%)
Olefins
(vol%)
Benzene
(vol%)
E200
(vol%)
E300
(vol%)
T50
(de2.F)
T90
(deg.F)
ElO
summer
-1.00
-
2.02
0.46
-
-3.11
-0.39
6.34
1.77
ElO
winter
-1.00
-
3.65
2.07
-
-4.88
-0.54
9.96
2.45
" Sulfur and benzene content are controlled by downstream regulations, and therefore aren't expected to change with
addition of ethanol.
6 Emissions Inventories and Air Quality Modeling
6.1	Overview of Analysis Methods
This analysis utilizes EPA's 2016v7.2 beta emissions modeling platform,26 which includes a
suite of 2016 inventories, ancillary emissions data, and scripts and software for preparing
emissions for air quality modeling. The onroad and nonroad inventories in the 2016v7.2 beta
platform represents the "with-RFS" scenario. These inventories were then recalculated using
modified fuels information, described in Section 5.2, to represent the "pre-RFS" scenario.
Section 6.2 describes the methodology for developing onroad mobile emissions inventories for
both scenarios; Section 6.3 describes this methodology for the nonroad mobile sector. In addition
to emissions from onroad and nonroad mobile sources, air quality modeling requires emissions
from all inventory sectors (e.g., biogenic, point, nonpoint sources); these sectors are discussed in
Section 6.5. Finally, Sections 6.6 and 6.7 provide an overview of the air quality modeling
methodology, including how emissions inventories are processed for air quality modeling.
6.2	Onroad Mobile Emissions Inventory
This section focuses on the approach and data sources used to develop gridded, hourly emissions
for the onroad mobile sector that are suitable for input to an air quality model in terms of the
format, grid resolution, and chemical species. While the fuel supplies used to develop emissions
for the "with-RFS" and "pre-RFS" scenarios differed, the approach and all other (non-fuel
supply) data sources used to calculate emissions for both scenarios were identical.
Onroad mobile source emission factors for all states except California were generated with
MOVES2014b, the latest public version available when the study was initiated in 2019. For this
analysis, MOVES2014b estimated onroad exhaust and evaporative emissions at the county level.
The MOVES2014b onroad emission estimates are based on a detailed analysis of in-use
26USEPA (2019). Technical Support Document: Preparation of Emissions Inventories for the Version 7.2 2016
North American Emissions Modeling Platform. Office of Air Quality Planning and Standards, U.S. Environmental
Protection Agency, Research Triangle Park, NC. September 2019. https://www.epa.gov/sites/prodiietion/files/20.1.9-
09/documents/2016v7.2_regionalhaze_emismod_tsd_508.pdf
14

-------
emissions from tens of thousands of light duty vehicles27 and hundreds of heavy-duty trucks.28
MOVES2014b also incorporates data from a wide range of test programs and other sources,
including data on the emissions effects of fuel properties such as gasoline sulfur and ethanol,29'30
and data on evaporative emissions from fuel leaks and from vehicles parked for multiple days.31
The impact of biodiesel blends on HC, CO, NOx and direct PM emissions from pre-2007 onroad
diesel engines is based on a detailed analysis of hundreds of emissions tests.32 Note that no
biodiesel effects are modelled for 2007 and later vehicles because no significant and consistent
effects have been observed.33
The MOVES-generated onroad emission factors were then combined with activity data to
produce emissions within the Sparse Matrix Operator Kernel Emissions (SMOKE) modeling
system. The collection of programs that compute the onroad mobile source emissions are known
as SMOKE-MOVES. SMOKE-MOVES uses a combination of vehicle activity data, emission
factors from MOVES, meteorology data, and temporal allocation information needed to estimate
hourly onroad emissions. Additional types of ancillary data are used for the emissions
processing, such as spatial surrogates which ensure emissions are developed on the grid used in
air quality modeling.
California is the only state for which EPA does not generate onroad mobile emissions based on
MOVES output. Instead, the California Air Resources Board uses the EPA-approved California-
specific model, EMFAC, to generate onroad mobile emissions in California. Because California
fuels were identical in the "pre-RFS" and "with-RFS" scenarios, a separate "pre-RFS" scenario
was not implemented for California. California, therefore, is not included in the inventory
comparisons.
National onroad emission summaries for key pollutants are provided in Section 6.4.
27	USEPA (2015). Exhaust Emission Rates for Light-Duty On-Road Vehicles in MOVES2014. EPA-420-R-15-005.
Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection
Agency, Ann Arbor, MI, October 2015. https://nepis.epa. eov/Exe/ZvPDF.egi?Dockev=P100NNVN.pdf
28	USEPA (2015). Exhaust Emission Rates for Heavy-Duty On-Road Vehicles in MOVES2014. EPA-420-R-15-015a.
Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection
Agency, Ann Arbor, MI, November 2015. httPs://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100NQ46.pdf
29	USEPA (2016). Fuel Effects on Exhaust Emissions for On-Road Vehicles in MOVES2014. EPA-420-R-16-001.
Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection
Agency, Ann Arbor, MI, February 2016. https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100Q5W2.pdf
30	USEPA (2016b). Air Toxic Emissions from On-road Vehicles in MOVES2014. EPA-420-R-16-016. Assessment
and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection Agency, Ann
Arbor, MI, November 2016. https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100PUNO.pdf
31	USEPA (2014). Evaporative Emissions from On-Road Vehicles in MOVES2014. EPA-420-R-14-014. Assessment
and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection Agency, Ann
Arbor, MI, September 2014. hltPs://nepis.epa.gov/Exe/ZyPDF.cgi?Dockev=P100KB5V.pdf
32	USEPA (2010). Regulatory Impact Analysis: Renewable Fuel Standard Program (RFS2). EPA-420-R-10-006.
Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection
Agency, Ann Arbor, MI. February 2010. (Appendix A).
33	USEPA (2016). Fuel Effects on Exhaust Emissions for On-Road Vehicles in MOVES2014. EPA-420-R-16-001.
Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection
Agency, Ann Arbor, MI, February 2016. https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100Q5W2.pdf
15

-------
6.2.1 Onroad Mobile Emissions Inventory Methodology
Onroad mobile source emissions result from motorized vehicles that normally operate on public
roadways, including passenger cars, motorcycles, minivans, sport-utility vehicles, light-duty
trucks, heavy-duty trucks, and buses. These sources are further divided between diesel, gasoline,
E-85, and compressed natural gas (CNG) vehicles. The onroad mobile emissions sector
characterizes emissions from parked vehicle processes (e.g., starts, hot soak, and hoteling) as
well as from "on-network" processes (i.e., from vehicles as they move along the roads). Except
for California, all onroad emissions are generated using the SMOKE-MOVES emissions
modeling framework that leverages MOVES-generated emission factors, county and Source
Classification Code (SCC)-specific activity data, and hourly meteorological data.
SMOKE-MOVES uses emission rate (i.e., "lookup") tables generated by MOVES as input.
These tables 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., an automated process is used to run MOVES to produce
emission factors for a series of temperatures and speeds for a set of "representative counties," to
which every other county in the country is mapped. Representative counties are used because it
is impractical to generate a full suite of emission factors for every county in the U.S. 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 age distributions and vehicle fuel
types, rather than choosing values specific to each representative county, a weighted average is
computed for all counties represented by each representative county, and the mean of those
averages was used. For the 2016v7.2 beta modeling platform used in this analysis, there are 303
representative counties, which is same as in the 2014v7.1 emissions modeling platform. A
detailed discussion of the selection of representative counties used in this analysis is available in
the 2014NEIv2 Technical Support Document (TSD), Section 6.8.2.34
Once representative counties are identified, emission factors are generated by running MOVES
for each representative county for two "fuel months" - January to represent winter months and
July to represent summer months - because different types of fuels are used in each season.
MOVES is run for the range of temperatures that occur in each representative county for each
season. SMOKE selects the appropriate MOVES emissions rates for each county, hourly
temperature, SCC, and speed bin and multiplies the emission rate by appropriate activity data:
VMT (vehicle miles traveled), VPOP (vehicle population), or HOTELING (hours of extended
idle) to produce emissions. These calculations are done for every county and grid cell in the
continental U.S. for every hour of the year. SMOKE-MOVES accounts for the temperature
sensitivity of the onroad emissions in each county by using the gridded hourly temperature
information available from the meteorological model outputs used for air quality modeling.
34 USEPA (2018). Technical Support Document: 2014 National Emissions Inventory, version 2. Office of Air
Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC. July 2018.
https://www.epa.gov/sites/production/files/2018-07/documents/nei2014v2 tsd 05iul2018.pdf
16

-------
In summary, the SMOKE-MOVES process for creating the model-ready onroad emissions for all
states except California consists of the following steps:
1)	Determine which counties will be used to represent other counties in the MOVES runs.
2)	Determine which months will be used to represent other months' fuel characteristics.
3)	Create inputs required by MOVES: county-specific information on vehicle populations,
age distributions, speed distribution, temporal profiles, and inspection-maintenance
programs for each of the representative counties.
4)	Create inputs needed both by MOVES and SMOKE, including temperatures and activity
data.
5)	Run MOVES to create emission factor tables for the temperatures and speeds that exist in
each county during the modeled period.
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.
The onroad emissions are processed as four components that are merged together into the final
onroad sector emissions:
•	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.
6.3 Nonroad Mobile Emissions Inventory
This section details the approach and data sources used to develop the 2016 emissions inventory
for the mobile nonroad equipment sector. This sector includes all mobile source emissions that
do not operate on roads, excluding commercial marine vehicles, railways, and aircraft. Types of
nonroad equipment include recreational vehicles, pleasure craft, and construction, mining, and
lawn and garden equipment.
Nonroad mobile emissions were generated using MOVES2014b without any state-provided
inputs, except for in California, where inventories are provided by the California Air Resources
Board (CARB), using an EPA-approved model. MOVES2014b incorporates EPA's previous
17

-------
NONROAD2008 model,35 including the fuel effects,36 but improves nonroad engine population
growth rates,37 nonroad Tier 4 engine emission rates,38 and sulfur levels of nonroad diesel
fuels.39 MOVES2014b models the HC, CO and NOx emissions impacts of oxygenate (such as
ethanol) for nonroad gasoline engines, as well as the impacts on toxics emissions.40' 41 MOVES
does not model a fuel effect on direct PM emissions from nonroad equipment due to insufficient
data. MOVES2014b does not model emissions impacts of biodiesel on nonroad diesel
equipment. Little data on emissions in nonroad engines using biodiesel exists, and an EPA
analysis of this data could not determine emission effects or conclude with confidence that
nonroad engines respond to biodiesel similarly to highway engines.42
For areas other than California, monthly MOVES2014b inventory outputs were used after a
limited amount of post-processing. Nonroad inventories were processed with the Sparse Matrix
Operating Kernel Emissions (SMOKE) modeling system version 4.6. SMOKE creates emissions
in a format that can be input into air quality models.
National nonroad emission summaries for key pollutants are provided in Section 6.4.
6.3.1 Nonroad Mobile Emissions Inventory Methodology
For all states except California, MOVES2014b was run to create a monthly emissions inventory
for criteria air pollutants (CAPs), a full set of hazardous air pollutants (HAPs), and additional
pollutants such as ethanol and total organic compounds less the sum of the HAPs
(NONHAPTOG), which are used for speciation. MOVES2014b provides estimates of
NONHAPTOG along with the speciation profile code for the NONHAPTOG emission source.
35	USEPA (2009). Frequently Asked Questions about NONROAD2008. EPA-420-F-09-021. Assessment and
Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection Agency, Ann Arbor,
MI, April 2009. https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P1003E8H.pdf
36	USEPA (2005). Exhaust Emission Effects of Fuel Sulfur and Oxygen on Gasoline Nonroad Engines. EPA-420-R-
05-016. Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental
Protection Agency, Ann Arbor, MI, December 2005. https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P1004L80.pdf
37	USEPA (2018). Nonroad Engine Population Growth Estimates in MOVES2014b. EPA-420-R-18-010.
Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection
Agency, Ann Arbor, MI, December July 2018. httPs://nepis.epa.gov/Exe/ZvPDF.cgi?Pockev=P100UXJK.pdf
38	USEPA (2018). Exhaust and Crankcase Emission Factors for Nonroad Compression-Ignition Engines in
MOVES2014b. EPA-420-R-18-009. Assessment and Standards Division, Office of Transportation and Air Quality,
U.S. Environmental Protection Agency, Ann Arbor, MI, December July 2018.
https://nepis.epa.gov/Exe/Zv PDF.cgi?Dockev=P.l..()0UXEN.pd:f
39	USEPA (2018). Fuel Supply Defaults: Regional Fuels and the Fuel Wizard in MOVES2014b. EPA-420-R-18-008.
Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection
Agency, Ann Arbor, MI, July 2018. https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100UXDZ.pdf
40	USEPA (2005). Exhaust Emission Effects of Fuel Sulfur and Oxygen on Gasoline Nonroad Engines. EPA-420-R-
05-016. Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental
Protection Agency, Ann Arbor, MI. December 2005. https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P1004L80.pdf
41	USEPA (2018). Speciation Profiles and Toxic Emission Factors for Nonroad Engines in MOVES2014b. EPA-
420-R-18-011. Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental
Protection Agency, Ann Aibor, MI. July 2018. httPs://nepis.epa.gov/Exe/ZvPDF.cgi?Dockev=P100UXK7.pdf
42	USEPA (2002). A Comprehensive Analysis of Biodiesel Impacts on Exhaust Emissions (Draft Report). EPA-420-
P-02-001. Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental
Protection Agency, Ann Arbor, MI. October 2002. https://nepis.epa.gov/Exe/ZyPDF.egi?Dockey=P.1.00lZA0.pdf
18

-------
MOVES2014b also provides estimates of PM2.5 by speciation profile code for the PM2.5 emission
source. To facilitate calculation of coarse particulate matter (PMC) within SMOKE, and to help
create emissions summaries, an additional pollutant representing total PM2.5 (called
PM25TOTAL) was added to the inventory.
MOVES2014b outputs emissions data in county-specific databases, and a post-processing script
converts the data into Flat File 10 (FF10) format. Additional post-processing steps were
performed as follows:
1)	County-specific FFlOs were combined into a single FF10 file.
2)	To reduce the file size of the inventory, HAPs that are not needed for air quality
modeling, such as dioxins and furans, were removed from the inventory.
3)	To reduce the file 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. For technical reasons, MOVES2014b attributes a very tiny amount of
emissions to sources that generate zero emissions, for example, snowmobile emissions in
Florida. Removing these sources from the inventory reduces the total file size of the
inventory by 7%.
4)	Gas and particulate components of HAPs that MOVES outputs separately, such as
naphthalene, were combined.
5)	VOC was renamed VOC INV so that SMOKE does not speciate both VOC (volatile
organic compounds) and NONHAPTOG, which would result in double counting.
6)	PM25TOTAL, referenced above, was also created at this stage of the process.
7)	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 ptnonipm and np oilgas sectors, respectively.
California is the only state for which EPA does not generate nonroad mobile emissions based on
MOVES output. Instead, the California Air Resources Board uses the EPA-approved California-
specific suite of offroad modeling tools to generate nonroad mobile emissions in California.
Because California fuels were identical in the "pre-RFS" and "with-RFS" scenarios, a separate
"pre-RFS" scenario was not implemented for California. California, therefore, is not included in
the inventory comparisons.
6.4 Onroad and Nonroad Emissions Inventory Summaries
National onroad and nonroad emissions totals for 2016 for the continental United States
(CONUS; excluding California) by pollutant and by sector for the "with-RFS" and "pre-RFS"
cases are provided in Table 6.1 and Table 6.2.
Comparison of state (CONUS; excluding California) total annual (2016) emissions (tons per
year) by pollutant for the "with-RFS" and "pre-RFS" scenarios are provided in Table 6.3 through
19

-------
Table 6.8. Detailed state-level emission summaries by pollutant and by sector are included with
the online supplemental materials.43
Table 6.1 Comparison of national (CONUS; excluding California) total annual (2016) emissions
	(tons per year) for the "with-RFS" and "pre-RFS" scenarios for criteria pollutants	
I'olliiliinl
Sector
"with-RIS"
"piv-UIS"
ililT
% ililT
NOx
Total (Gasoline + Diesel)
4,684,218
4,542,487
141,731
3.1 %

Gasoline - onroad
2,013,967
1,900,020
113,947
6%

Gasoline - nonroad
90,858
69,821
21,037
30.1 %

Diesel - onroad
1,819,928
1,813,180
6,748
0.4 %

Diesel - nonroad
759,466
759,466
0
0%






VOC
Total (Gasoline + Diesel)
2,600,189
2,495,537
104,652
4.2 %

Gasoline - onroad
1,708,443
1,601,992
106,451
6.6 %

Gasoline - nonroad
649,764
646,758
3,006
0.5 %

Diesel - onroad
171,044
175,850
-4,806
-2.7 %

Diesel - nonroad
70,938
70,938
0
0%






PMio
Total (Gasoline + Diesel)
345,532
347,558
-2,026
-0.6 %

Gasoline - onroad
144,320
143,605
715
0.5 %

Gasoline - nonroad
33,181
33,181
0
0%

Diesel - onroad
106,511
109,253
-2,742
-2.5 %

Diesel - nonroad
61,519
61,519
0
0%






PM2.5
Total (Gasoline + Diesel)
210,458
212,347
-1,889
-0.9 %

Gasoline - onroad
48,693
48,059
634
1.3 %

Gasoline - nonroad
30,526
30,526
0
0%

Diesel - onroad
71,565
74,088
-2,523
-3.4%

Diesel - nonroad
59,674
59,674
0
0%






so2
Total (Gasoline + Diesel)
27,577
26,914
663
2.5 %

Gasoline - onroad
22,148
21,485
663
3.1 %

Gasoline - nonroad
728
728
0
0%

Diesel - onroad
3,680
3,680
0
0%

Diesel - nonroad
1,021
1,021
0
0%






CO
Total (Gasoline + Diesel)
28,193,369
31,084,459
-2,891,090
-9.3 %

Gasoline - onroad
18,653,076
19,761,434
-1,108,358
-5.6%

Gasoline - nonroad
8,310,708
10,069,942
-1,759,234
-17.5 %

Diesel - onroad
858,309
881,807
-23,498
-2.7 %

Diesel - nonroad
371,276
371,276
0
0%
43 https://www.epa.gOY/renewable-fuel-standard-program/anti-backsliding-determination-aiid-stiidY
20

-------
Table 6.2 Comparison of national (CONUS; excluding California) total annual (2016) emissions
	(tons per year) for the "with-RFS" and "pre-RFS" scenarios for air toxics 	
Polllllil lit
Sector
"wiiii-ms"
"pro-UTS"
ililT
% ililT
Acetaldehyde
Total (Gasoline + Diesel)
30,882
21,875
9,007
41.2%

Gasoline - onroad
15,357
7,312
8,045
110%

Gasoline - nonroad
2,571
1,438
1,133
78.8 %

Diesel - onroad
6,659
6,830
-171
-2.5 %

Diesel - nonroad
6,295
6,295
0
0%






Acrolein
Total (Gasoline + Diesel)
3,779
3,700
79
2.1 %

Gasoline - onroad
890
820
70
8.5 %

Gasoline - nonroad
216
176
40
22.7 %

Diesel - onroad
1,157
1,189
-32
-2.7 %

Diesel - nonroad
1,516
1,516
0
0%






Benzene
Total (Gasoline + Diesel)
64,021
74,184
-10,163
-13.7 %

Gasoline - onroad
43,354
49,463
-6,109
-12.4 %

Gasoline - nonroad
16,680
20,696
-4,016
-19.4 %

Diesel - onroad
1,444
1,482
-38
-2.6 %

Diesel - nonroad
2,543
2,543
0
0%






1,3-Butadiene
Total (Gasoline + Diesel)
10,052
11,064
-1,012
-9.1 %

Gasoline - onroad
6,579
7,492
-913
-12.2 %

Gasoline - nonroad
2,935
3,019
-84
-2.8 %

Diesel - onroad
407
421
-14
-3.3 %

Diesel - nonroad
132
132
0
0%






Formaldehyde
Total (Gasoline + Diesel)
51,743
51,427
316
0.6 %

Gasoline - onroad
13,037
12,137
900
7.4 %

Gasoline - nonroad
4,526
4,733
-207
-4.4 %

Diesel - onroad
16,506
16,882
-376
-2.2 %

Diesel - nonroad
17,675
17,675
0
0%






Naphthalene
Total (Gasoline + Diesel)
5,192
5,258
-66
-1.3 %

Gasoline - onroad
2,258
2,263
-5
-0.2 %

Gasoline - nonroad
1,012
1,029
-17
-1.7%

Diesel - onroad
1,657
1,700
-43
-2.5 %

Diesel - nonroad
266
266
0
0%
21

-------
Table 6.3 Comparison of state (CONUS; excluding California) total annual (2016) NOx and VOC
emissions (tons per year) for the "with-RFS" and "pre-RFS" scenarios1^

\()\
VOC
Sl;ik-
"Hiih-RI-V
"piv-RI-'S"
(Mir
"..(Mir
"wilh-urs"
"pn-urs"
(Mir
"... dilT
Alabama
129,385
123,457
5,928
4.8 %
81,209
74,778
6,431
8.6 %
Arizona
117,980
115,305
2,675
2.3 %
71,372
67,945
3,427
5%
Arkansas
87,840
84,912
2,928
3.4 %
42,234
39,619
2,615
6.6 %
Colorado
87,698
83,460
4,238
5.1 %
60,001
58,313
1,688
2.9 %
Connecticut
32,042
32,017
25
0.1 %
26,834
26,855
-21
-0.1 %
Delaware
13,801
13,787
14
0.1 %
13,239
13,247
-8
-0.1 %
D.C.
4,264
4,259
5
0.1 %
3,208
3,213
-5
-0.2 %
Florida
296,038
281,020
15,018
5.3 %
229,949
211,582
18,367
8.7 %
Georgia
200,730
192,480
8,250
4.3 %
113,134
105,857
7,277
6.9 %
Idaho
50,297
48,158
2,139
4.4 %
30,875
29,671
1,204
4.1 %
Illinois
181,457
178,222
3,235
1.8%
108,380
106,140
2,240
2.1 %
Indiana
161,569
156,409
5,160
3.3 %
79,929
76,474
3,455
4.5 %
Iowa
92,042
89,141
2,901
3.3 %
46,821
44,819
2,002
4.5 %
Kansas
87,336
84,685
2,651
3.1 %
38,691
36,592
2,099
5.7%
Kentucky
97,725
94,565
3,160
3.3 %
53,741
51,139
2,602
5.1 %
Louisiana
90,054
86,343
3,711
4.3 %
54,174
50,152
4,022
8%
Maine
26,653
25,265
1,388
5.5 %
26,837
26,122
715
2.7 %
Maryland
71,762
70,843
919
1.3%
45,943
45,368
575
1.3%
Massachusetts
53,671
53,621
50
0.1 %
43,186
43,224
-38
-0.1 %
Michigan
132,215
124,963
7,252
5.8 %
120,141
115,718
4,423
3.8 %
Minnesota
118,743
113,021
5,722
5.1 %
97,407
94,554
2,853
3%
Mississippi
75,429
72,404
3,025
4.2 %
41,360
38,283
3,077
8%
Missouri
168,040
163,448
4,592
2.8 %
80,935
77,339
3,596
4.6 %
Montana
44,541
42,964
1,577
3.7%
22,706
21,892
814
3.7%
Nebraska
66,524
64,558
1,966
3%
29,318
27,887
1,431
5.1 %
Nevada
53,123
51,227
1,896
3.7%
28,569
26,880
1,689
6.3 %
New Hampshire
17,797
17,350
447
2.6 %
16,434
16,143
291
1.8%
New Jersey
78,396
78,295
101
0.1 %
50,705
50,779
-74
-0.1 %
New Mexico
71,235
69,178
2,057
3%
28,019
26,366
1,653
6.3 %
New York
149,458
145,826
3,632
2.5 %
112,298
110,316
1,982
1.8%
North Carolina
161,062
152,638
8,424
5.5 %
106,244
99,305
6,939
7%
North Dakota
57,558
56,601
957
1.7%
15,786
15,518
268
1.7%
Ohio
166,708
159,309
7,399
4.6 %
109,372
105,021
4,351
4.1 %
Oklahoma
90,011
86,413
3,598
4.2 %
52,111
48,747
3,364
6.9 %
Oregon
76,487
72,675
3,812
5.2 %
51,201
49,315
1,886
3.8 %
Pennsylvania
171,017
165,922
5,095
3.1 %
100,329
97,230
3,099
3.2 %
Rhode Island
12,631
12,611
20
0.2 %
7,329
7,342
-13
-0.2 %
South Carolina
107,013
101,930
5,083
5%
66,149
61,335
4,814
7.8 %
South Dakota
39,152
38,155
997
2.6 %
16,014
15,463
551
3.6 %
Tennessee
146,310
140,179
6,131
4.4 %
83,413
78,090
5,323
6.8 %
Texas
423,981
414,507
9,474
2.3 %
204,951
196,808
8,143
4.1 %
Utah
74,015
71,518
2,497
3.5 %
37,523
36,113
1,410
3.9 %
Vermont
11,237
10,812
425
3.9 %
8,523
8,321
202
2.4 %
Virginia
129,334
125,669
3,665
2.9 %
81,244
78,622
2,622
3.3 %
Washington
134,715
128,199
6,516
5.1 %
85,766
82,662
3,104
3.8 %
West Virginia
36,462
35,084
1,378
3.9 %
19,917
18,947
970
5.1 %
Wisconsin
112,680
108,443
4,237
3.9 %
82,775
80,588
2,187
2.7 %
Wyoming
28,563
27,696
867
3.1 %
12,467
12,087
380
3.1 %
t State annual totals only include emissions from gasoline and diesel vehicles and engines and exclude emissions from alternative
fuels (i.e., CNG, LPG).
22

-------
Table 6.4 Comparison of state (CONUS; excluding California) total annual (2016) PMio and PM2.5
emissions (tons per year) for the "with-RFS" and "pre-RFS" scenarios1^

PMn,
I'M: 5
Sl;ik-
"Hiih-RI-V
"piv-RI-'S"
din'
(liir
"wilh-KrS"
"piv-urs"
iiiir
"... dilT
Alabama
8,131
8,182
-51
-0.6 %
4,742
4,790
-48
-1 %
Arizona
7,610
7,663
-53
-0.7 %
4,950
4,998
-48
-1 %
Arkansas
5,232
5,275
-43
-0.8 %
3,576
3,616
-40
-1.1 %
Colorado
6,401
6,422
-21
-0.3 %
3,970
3,990
-20
-0.5 %
Connecticut
2,818
2,828
-10
-0.4 %
1,595
1,604
-9
-0.6 %
Delaware
947
952
-5
-0.5 %
589
594
-5
-0.8 %
D.C.
615
617
-2
-0.3 %
249
251
-2
-0.8 %
Florida
26,276
26,372
-96
-0.4 %
14,247
14,338
-91
-0.6 %
Georgia
13,306
13,380
-74
-0.6 %
7,706
7,775
-69
-0.9 %
Idaho
2,911
2,937
-26
-0.9 %
2,026
2,049
-23
-1.1 %
Illinois
15,637
15,721
-84
-0.5 %
9,417
9,495
-78
-0.8 %
Indiana
11,724
11,790
-66
-0.6 %
7,292
7,353
-61
-0.8 %
Iowa
6,070
6,100
-30
-0.5 %
4,556
4,583
-27
-0.6 %
Kansas
5,342
5,375
-33
-0.6 %
3,923
3,954
-31
-0.8 %
Kentucky
6,025
6,078
-53
-0.9 %
3,794
3,843
-49
-1.3%
Louisiana
6,260
6,315
-55
-0.9 %
3,607
3,657
-50
-1.4%
Maine
2,066
2,080
-14
-0.7 %
1,371
1,384
-13
-0.9 %
Maryland
6,334
6,376
-42
-0.7 %
3,439
3,478
-39
-1.1 %
Massachusetts
6,105
6,127
-22
-0.4 %
3,146
3,166
-20
-0.6 %
Michigan
10,403
10,421
-18
-0.2 %
6,306
6,323
-17
-0.3 %
Minnesota
9,118
9,136
-18
-0.2 %
6,797
6,814
-17
-0.2 %
Mississippi
4,356
4,386
-30
-0.7 %
2,668
2,696
-28
-1 %
Missouri
10,345
10,435
-90
-0.9 %
7,070
7,153
-83
-1.2%
Montana
2,550
2,568
-18
-0.7 %
1,945
1,962
-17
-0.9 %
Nebraska
4,033
4,054
-21
-0.5 %
3,095
3,114
-19
-0.6 %
Nevada
4,192
4,207
-15
-0.4 %
2,747
2,760
-13
-0.5 %
New Hampshire
1,516
1,523
-7
-0.5 %
971
978
-7
-0.7 %
New Jersey
7,291
7,333
-42
-0.6 %
4,276
4,315
-39
-0.9 %
New Mexico
3,736
3,784
-48
-1.3%
2,441
2,485
-44
-1.8%
New York
16,839
16,926
-87
-0.5 %
8,971
9,051
-80
-0.9 %
North Carolina
11,212
11,243
-31
-0.3 %
6,448
6,478
-30
-0.5 %
North Dakota
3,967
3,992
-25
-0.6 %
3,316
3,339
-23
-0.7 %
Ohio
13,505
13,547
-42
-0.3 %
8,028
8,067
-39
-0.5 %
Oklahoma
5,787
5,830
-43
-0.7 %
3,581
3,621
-40
-1.1 %
Oregon
5,006
5,032
-26
-0.5 %
3,224
3,248
-24
-0.7 %
Pennsylvania
13,403
13,483
-80
-0.6 %
8,104
8,178
-74
-0.9 %
Rhode Island
1,084
1,093
-9
-0.8 %
618
627
-9
-1.4%
South Carolina
6,678
6,729
-51
-0.8 %
4,120
4,167
-47
-1.1 %
South Dakota
2,489
2,504
-15
-0.6 %
2,014
2,028
-14
-0.7 %
Tennessee
9,039
9,095
-56
-0.6 %
5,360
5,413
-53
-1 %
Texas
30,456
30,701
-245
-0.8 %
18,093
18,320
-227
-1.2%
Utah
5,098
5,151
-53
-1 %
3,085
3,135
-50
-1.6%
Vermont
1,145
1,146
-1
-0.1 %
752
753
-1
-0.1 %
Virginia
9,089
9,136
-47
-0.5 %
5,425
5,469
-44
-0.8 %
Washington
8,764
8,809
-45
-0.5 %
5,498
5,540
-42
-0.8 %
West Virginia
2,216
2,237
-21
-0.9 %
1,436
1,456
-20
-1.4%
Wisconsin
7,761
7,809
-48
-0.6 %
5,286
5,330
-44
-0.8 %
Wyoming
1,416
1,434
-18
-1.3%
983
999
-16
-1.6%
t State annual totals only include emissions from gasoline and diesel vehicles and engines and exclude emissions from alternative
fuels (i.e., CNG, LPG).
23

-------
Table 6.5 Comparison of state (CONUS; excluding California) total annual (2016) CO and SO2
emissions (tons per year) for the "with-RFS" and "pre-RFS" scenarios1^

CO
SO;
Sl;ik-
"Hiih-RI-V
"piv-RI-'S"
diir
"..(Mir
"wilh-urs"
"piv-urs"
iiiir
"oil ill'
Alabama
771,116
861,426
-90,310
-10.5 %
805
779
26
3.3 %
Arizona
715,015
758,134
-43,119
-5.7%
402
397
5
1.3%
Arkansas
398,809
450,934
-52,125
-11.6%
365
355
10
2.8 %
Colorado
632,377
733,034
-100,657
-13.7%
419
406
13
3.2 %
Connecticut
285,881
285,989
-108
0%
312
312
0
0%
Delaware
117,531
117,579
-48
0%
110
110
0
0%
D.C.
34,996
35,016
-20
-0.1 %
56
56
0
0%
Florida
2,421,201
2,772,409
-351,208
-12.7%
2,592
2,504
88
3.5 %
Georgia
1,206,612
1,365,641
-159,029
-11.6%
1,366
1,321
45
3.4 %
Idaho
243,103
279,835
-36,732
-13.1 %
155
151
4
2.6 %
Illinois
1,183,863
1,243,093
-59,230
-4.8 %
794
784
10
1.3%
Indiana
851,946
947,199
-95,253
-10.1 %
687
669
18
2.7 %
Iowa
414,609
472,698
-58,089
-12.3 %
354
344
10
2.9 %
Kansas
383,024
432,490
-49,466
-11.4%
327
318
9
2.8 %
Kentucky
529,833
575,891
-46,058
-8%
378
370
8
2.2 %
Louisiana
500,605
566,472
-65,867
-11.6%
591
572
19
3.3 %
Maine
183,160
213,297
-30,137
-14.1 %
176
174
2
1.1 %
Maryland
535,309
550,106
-14,797
-2.7 %
639
636
3
0.5 %
Massachusetts
502,940
503,133
-193
0%
661
661
0
0%
Michigan
1,056,388
1,207,516
-151,128
-12.5 %
752
727
25
3.4 %
Minnesota
791,305
913,050
-121,745
-13.3%
476
462
14
3%
Mississippi
404,359
447,946
-43,587
-9.7 %
465
450
15
3.3 %
Missouri
791,294
863,227
-71,933
-8.3 %
646
632
14
2.2 %
Montana
185,935
210,590
-24,655
-11.7%
116
113
3
2.7 %
Nebraska
268,706
304,913
-36,207
-11.9%
220
214
6
2.8 %
Nevada
300,472
345,712
-45,240
-13.1 %
193
188
5
2.7 %
New Hampshire
148,806
158,297
-9,491
-6%
152
151
1
0.7 %
New Jersey
646,004
646,311
-307
0%
809
809
0
0%
New Mexico
256,351
287,121
-30,770
-10.7%
253
246
7
2.8 %
New York
1,217,731
1,305,719
-87,988
-6.7 %
1,504
1,481
23
1.6%
North Carolina
1,137,222
1,289,318
-152,096
-11.8%
1,299
1,254
45
3.6 %
North Dakota
146,692
169,134
-22,442
-13.3%
125
123
2
1.6%
Ohio
1,171,741
1,337,454
-165,713
-12.4 %
924
895
29
3.2 %
Oklahoma
518,057
590,345
-72,288
-12.2 %
464
450
14
3.1 %
Oregon
489,879
568,738
-78,859
-13.9%
297
288
9
3.1 %
Pennsylvania
1,055,311
1,170,165
-114,854
-9.8 %
1,102
1,077
25
2.3 %
Rhode Island
79,013
79,070
-57
-0.1 %
91
91
0
0%
South Carolina
651,879
738,968
-87,089
-11.8%
633
613
20
3.3 %
South Dakota
138,140
156,214
-18,074
-11.6%
97
95
2
2.1 %
Tennessee
833,412
935,039
-101,627
-10.9 %
754
730
24
3.3 %
Texas
2,415,363
2,603,866
-188,503
-1.2 %
2,779
2,725
54
2%
Utah
332,276
380,154
-47,878
-12.6 %
275
268
7
2.6 %
Vermont
71,758
83,978
-12,220
-14.6 %
92
89
3
3.4 %
Virginia
902,061
959,852
-57,791
-6%
950
935
15
1.6%
Washington
807,113
924,660
-117,547
-12.7%
513
497
16
3.2 %
West Virginia
188,275
212,394
-24,119
-U.4%
171
166
5
3%
Wisconsin
696,371
778,282
-81,911
-10.5 %
464
454
10
2.2 %
Wyoming
99,493
112,775
-13,282
-11.8%
83
81
2
2.5 %
t State annual totals only include emissions from gasoline and diesel vehicles and engines and exclude emissions from alternative
fuels (i.e., CNG, LPG).
24

-------
Table 6.6 Comparison of state (CONUS; excluding California) total annual (2016) acetaldehyde
and acrolein emissions (tons per year) for the "with-RFS" and "pre-RFS" scenarios1^

.Uvhildi-lmli-
A (.nik-iii
Sl;ik-
"Hilh-RI-V
"piv-RI-'S"
(Mir
(liir
"wilh-KrS"
"piv-urs"
iiiir
"oil ill'
Alabama
774
462
312
67.5 %
83
81
2
2.5 %
Arizona
758
616
142
23.1 %
97
96
1
1 %
Arkansas
514
355
159
44.8 %
68
66
2
3%
Colorado
648
375
273
72.8 %
73
70
3
4.3 %
Connecticut
270
271
-1
-0.4 %
29
29
0
0%
Delaware
109
110
-1
-0.9 %
11
11
0
0%
D.C.
32
33
-1
-3%
4
4
0
0%
Florida
1,988
1,235
753
61 %
244
234
10
4.3 %
Georgia
1,212
771
441
57.2 %
143
138
5
3.6 %
Idaho
330
212
118
55.7%
40
39
1
2.6 %
Illinois
1,379
1,179
200
17%
177
175
2
1.1 %
Indiana
1,060
726
334
46%
135
132
3
2.3 %
Iowa
656
447
209
46.8 %
96
94
2
2.1 %
Kansas
562
396
166
41.9%
83
81
2
2.5 %
Kentucky
614
433
181
41.8%
72
70
2
2.9 %
Louisiana
540
354
186
52.5 %
67
64
3
4.7 %
Maine
217
153
64
41.8%
25
23
2
8.7 %
Maryland
472
438
34
7.8 %
53
53
0
0%
Massachusetts
464
465
-1
-0.2 %
52
52
0
0%
Michigan
1,125
582
543
93.3 %
121
112
9
8%
Minnesota
994
608
386
63.5 %
130
123
7
5.7%
Mississippi
435
271
164
60.5 %
50
49
1
2%
Missouri
1,048
796
252
31.7%
137
135
2
1.5%
Montana
313
219
94
42.9 %
44
43
1
2.3 %
Nebraska
456
329
127
38.6 %
69
68
1
1.5%
Nevada
359
254
105
41.3%
48
47
1
2.1 %
New Hampshire
152
125
27
21.6%
17
16
1
6.3 %
New Jersey
570
572
-2
-0.3 %
70
71
-1
-1.4%
New Mexico
363
253
110
43.5 %
44
44
0
0%
New York
1,172
942
230
24.4 %
145
141
4
2.8 %
North Carolina
992
568
424
74.6 %
109
105
4
3.8 %
North Dakota
368
315
53
16.8 %
67
67
0
0%
Ohio
1,249
728
521
71.6%
146
139
7
5%
Oklahoma
558
356
202
56.7 %
69
67
2
3%
Oregon
544
336
208
61.9%
62
60
2
3.3 %
Pennsylvania
1,154
813
341
41.9%
136
132
4
3%
Rhode Island
88
89
-1
-1.1 %
10
10
0
0%
South Carolina
630
387
243
62.8 %
72
69
3
4.3 %
South Dakota
265
203
62
30.5 %
44
43
1
2.3 %
Tennessee
895
551
344
62.4 %
103
99
4
4%
Texas
2,361
1,900
461
24.3 %
308
304
4
1.3%
Utah
436
293
143
48.8 %
54
53
1
1.9%
Vermont
98
63
35
55.6 %
13
12
1
8.3 %
Virginia
837
659
178
27%
94
92
2
2.2 %
Washington
948
581
367
63.2 %
108
105
3
2.9 %
West Virginia
223
135
88
65.2 %
25
24
1
4.2 %
Wisconsin
820
553
267
48.3 %
99
94
5
5.3 %
Wyoming
160
114
46
40.4 %
21
20
1
5%
t State annual totals only include emissions from gasoline and diesel vehicles and engines and exclude emissions from alternative
fuels (i.e., CNG, LPG).
25

-------
Table 6.7 Comparison of state (CONUS; excluding California) total annual (2016) benzene and
1,3-butadiene emissions (tons per year) for the "with-RFS" and "pre-RFS" scenarios1^

lil'll/.l'lll'
l.3-l>iil;i(lk'iU'
Sl;ik-
"Hiih-RI-V
"piv-RI-'S"
din'
(liir
"wilh-KrS"
"pri-RI-'S"
(Mir
"oil ill'
Alabama
1,887
2,189
-302
-13.8%
272
301
-29
-9.6 %
Arizona
1,678
1,810
-132
-7.3 %
249
261
-12
-4.6 %
Arkansas
1,003
1,176
-173
-14.7%
156
173
-17
-9.8 %
Colorado
1,625
1,997
-372
-18.6%
254
291
-37
-12.7%
Connecticut
644
644
0
0%
122
122
0
0%
Delaware
311
311
0
0%
57
57
0
0%
D.C.
68
68
0
0%
12
12
0
0%
Florida
5,388
6,310
-922
-14.6 %
764
836
-72
-8.6 %
Georgia
2,763
3,247
-484
-14.9 %
407
453
-46
-10.2 %
Idaho
731
876
-145
-16.6 %
115
126
-11
-8.7 %
Illinois
2,637
2,846
-209
-7.3 %
422
439
-17
-3.9 %
Indiana
1,945
2,294
-349
-15.2%
295
330
-35
-10.6 %
Iowa
1,168
1,390
-222
-16 %
193
216
-23
-10.6 %
Kansas
965
1,140
-175
-15.4%
153
173
-20
-11.6%
Kentucky
1,246
1,420
-174
-12.3 %
192
208
-16
-7.7 %
Louisiana
1,257
1,456
-199
-13.7%
184
200
-16
-8%
Maine
576
662
-86
-13%
106
104
2
1.9%
Maryland
1,112
1,158
-46
-4%
200
203
-3
-1.5%
Massachusetts
1,089
1,089
0
0%
208
209
-1
-0.5 %
Michigan
2,825
3,386
-561
-16.6 %
463
503
-40
-8%
Minnesota
2,338
2,817
-479
-17%
344
367
-23
-6.3 %
Mississippi
962
1,115
-153
-13.7%
139
154
-15
-9.7 %
Missouri
1,930
2,185
-255
-11.7%
308
335
-27
-8.1 %
Montana
574
684
-110
-16.1 %
87
98
-11
-11.2%
Nebraska
740
876
-136
-15.5%
119
134
-15
-11.2%
Nevada
710
840
-130
-15.5%
107
118
-11
-9.3 %
New Hampshire
382
412
-30
-7.3 %
70
70
0
0%
New Jersey
1,277
1,278
-1
-0.1 %
237
237
0
0%
New Mexico
636
747
-111
-14.9 %
103
116
-13
-11.2%
New York
2,740
3,026
-286
-9.5 %
479
499
-20
-4%
North Carolina
2,584
3,069
-485
-15.8%
385
428
-43
-10 %
North Dakota
422
505
-83
-16.4 %
56
62
-6
-9.7 %
Ohio
2,804
3,406
-602
-17.7%
444
502
-58
-11.6%
Oklahoma
1,230
1,456
-226
-15.5%
196
219
-23
-10.5 %
Oregon
1,313
1,600
-287
-17.9 %
204
230
-26
-11.3%
Pennsylvania
2,505
2,914
-409
-14 %
416
454
-38
-8.4 %
Rhode Island
182
182
0
0%
35
35
0
0%
South Carolina
1,572
1,838
-266
-14.5 %
231
253
-22
-8.7 %
South Dakota
409
488
-79
-16.2 %
54
61
-7
-11.5%
Tennessee
1,988
2,337
-349
-14.9 %
297
330
-33
-10 %
Texas
4,784
5,305
-521
-9.8 %
762
812
-50
-6.2 %
Utah
902
1,085
-183
-16.9 %
140
157
-17
-10.8 %
Vermont
206
246
-40
-16.3 %
34
36
-2
-5.6 %
Virginia
1,975
2,172
-197
-9.1 %
329
347
-18
-5.2 %
Washington
2,215
2,681
-466
-17.4 %
340
384
-44
-11.5%
West Virginia
482
577
-95
-16.5 %
75
84
-9
-10.7%
Wisconsin
1,923
2,247
-324
-14.4 %
295
309
-14
-4.5 %
Wyoming
285
340
-55
-16.2 %
46
50
-4
-8%
t State annual totals only include emissions from gasoline and diesel vehicles and engines and exclude emissions from alternative
fuels (i.e., CNG, LPG).
26

-------
Table 6.8 Comparison of state (CONUS; excluding California) total annual (2016) formaldehyde
and naphthalene emissions (tons per year) for the "with-RFS" and "pre-RFS" scenarios1^


Vmhlh.ik-iu-
Sl;ik-
"Hilh-RI-V
"piv-urs"
(liir
dill
"wilh-KrS"
"piv-RI-'S"
diir
(liir
Alabama
1,234
1,215
19
1.6%
139
139
0
0%
Arizona
1,376
1,371
5
0.4 %
128
129
-1
-0.8 %
Arkansas
969
964
5
0.5 %
85
86
-1
-1.2%
Colorado
1,007
1,002
5
0.5 %
117
120
-3
-2.5 %
Connecticut
413
415
-2
-0.5 %
46
46
0
0%
Delaware
156
157
-1
-0.6 %
22
22
0
0%
D.C.
49
50
-1
-2%
5
5
0
0%
Florida
3,415
3,389
26
0.8 %
368
372
-4
-1.1 %
Georgia
2,077
2,061
16
0.8 %
222
224
-2
-0.9 %
Idaho
566
561
5
0.9 %
67
68
-1
-1.5%
Illinois
2,412
2,404
8
0.3 %
219
220
-1
-0.5 %
Indiana
1,892
1,879
13
0.7 %
168
170
-2
-1.2%
Iowa
1,227
1,215
12
1 %
98
99
-1
-1 %
Kansas
1,071
1,066
5
0.5 %
80
81
-1
-1.2%
Kentucky
1,006
998
8
0.8 %
109
110
-1
-0.9 %
Louisiana
946
939
7
0.7 %
107
108
-1
-0.9 %
Maine
347
336
11
3.3 %
55
54
1
1.9%
Maryland
749
751
-2
-0.3 %
87
88
-1
-1.1 %
Massachusetts
736
739
-3
-0.4 %
78
78
0
0%
Michigan
1,632
1,572
60
3.8 %
224
226
-2
-0.9 %
Minnesota
1,724
1,681
43
2.6 %
187
189
-2
-1.1 %
Mississippi
719
709
10
1.4%
77
77
0
0%
Missouri
1,935
1,927
8
0.4 %
174
176
-2
-1.1 %
Montana
582
580
2
0.3 %
52
53
-1
-1.9%
Nebraska
899
894
5
0.6 %
63
63
0
0%
Nevada
699
697
2
0.3 %
56
57
-1
-1.8%
New Hampshire
237
233
4
1.7%
32
32
0
0%
New Jersey
994
999
-5
-0.5 %
101
102
-1
-1 %
New Mexico
655
653
2
0.3 %
69
70
-1
-1.4%
New York
2,015
2,001
14
0.7 %
220
222
-2
-0.9 %
North Carolina
1,567
1,546
21
1.4%
177
178
-1
-0.6 %
North Dakota
869
871
-2
-0.2 %
43
44
-1
-2.3 %
Ohio
1,991
1,965
26
1.3%
204
207
-3
-1.4%
Oklahoma
950
944
6
0.6 %
102
103
-1
-1 %
Oregon
912
911
1
0.1 %
100
102
-2
-2%
Pennsylvania
1,940
1,928
12
0.6 %
205
208
-3
-1.4%
Rhode Island
141
142
-1
-0.7 %
15
15
0
0%
South Carolina
1,051
1,040
11
1.1 %
120
121
-1
-0.8 %
South Dakota
552
550
2
0.4 %
37
38
-1
-2.6 %
Tennessee
1,485
1,467
18
1.2%
161
162
-1
-0.6 %
Texas
4,357
4,358
-1
0%
413
418
-5
-1.2%
Utah
764
763
1
0.1 %
89
91
-2
-2.2 %
Vermont
181
176
5
2.8 %
17
17
0
0%
Virginia
1,322
1,315
7
0.5 %
143
143
0
0%
Washington
1,562
1,553
9
0.6 %
169
172
-3
-1.7%
West Virginia
356
352
4
1.1 %
41
41
0
0%
Wisconsin
1,357
1,329
28
2.1 %
172
173
-1
-0.6 %
Wyoming
300
298
2
0.7 %
35
35
0
0%
t State annual totals only include emissions from gasoline and diesel vehicles and engines and exclude emissions from alternative
fuels (i.e., CNG, LPG).
27

-------
6.5 Other Inventory Sectors
Emissions for all other inventory sectors (Table 6.9) were developed for the 2016v7.2 beta
modeling platform. Along with the above-described onroad and nonroad mobile emissions,
emissions from these sectors are used as inputs for the air quality modeling discussed in Section
6.7 and Section 7. Emissions from these sectors were unchanged between the "with-RFS" and
"pre-RFS" scenarios. Documentation detailing the development of these emissions inventories is
available from the 2016 National Emissions Collaborative wiki page.44
Table 6.9 Inventory sectors included in the 2016v7.2 beta emissions modeling platform
Inventory Sector
Sector Description
Biogenic
VOC emissions from trees, shrubs, grasses, and soils
Mobile - Nonroad
See Section 6.3
Mobile - Onroad
See Section 6.2
Mobile - Commercial Marine Vessels
Commercial marine vessels with Category 1, 2, and 3 engines
Mobile - Rail
Class I, II, and II railroad emissions (including yards and
switchers), commuter rail, Amtrak
Nonpoint - Agriculture
NH3 and VOC emissions from livestock and fertilizer sources
Nonpoint - Area Fugitive Dust
PM emissions from paved roads, unpaved roads and airstrips,
construction, agriculture production, and mining and quarrying
Nonroad - Residential Wood Combustion
Residential wood burning devices such as fireplaces,
woodstoves, pellet stoves, indoor furnaces, outdoor burning in
fire pits and chimneys
Nonpoint - Other
All nonpoint sources not included in other sectors, including
solvents, industrial processes, waste disposal, storage and
transport of chemicals and petroleum, waste disposal,
commercial cooking, miscellaneous area sources
Oil & Gas - Nonpoint and Point
Oil and gas exploration and production, both onshore and
offshore
Point - Electrical Generating Units
Fossil fuel fired electrical generating units
Point - Fires - Agricultural
Agricultural burning
Point - Fires - Wild and Prescribed
Wildfires and prescribed burns
Canada - Mobile - Onroad
Onroad emissions in Canada
Mexico - Mobile - Onroad
Onroad emissions in Mexico
Canada/Mexico - Point
Canadian and Mexican point source emissions
Canada/Mexico - Point - Fires
Canadian and Mexican fire emissions
Canada/Mexico - Nonpoint
Canadian and Mexican nonpoint emissions
Canada - Nonpoint - Area Fugitive Dust
Area fugitive dust emissions in Canada
Canada - Point - Dust
Dust emissions in Canada
6.6 Emissions Modeling
The CMAQ air quality 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 Sections 6.2 - 6.5. The process of emissions modeling transforms the emissions
44 http://views.cira.colostate.edu/wiki/wiki/10197
28

-------
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 chemical speciation, temporal allocation, and spatial allocation of emissions.
SMOKE version 4.6 was used to process the raw emissions inventories into emissions inputs for
each modeling sector in a format compatible with CMAQ. When preparing emissions for
CMAQ, emissions for each sector are processed separately through SMOKE, and then merged to
combine the model-ready, sector-specific 2-D gridded emissions across sectors.
The emissions modeling step for chemical speciation creates the "model species" needed by
CMAQ 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 2016v7.2 beta platform is the CB6 mechanism45. This platform
generates the PM2.5 model species associated with the CMAQ Aerosol Module version 6 (AE6).
See Section 3.2 of the Preparation of Emissions Inventories for the Version 7.2 2016 North
American Emissions Modeling Platform Technical Support Document46 for more information
about chemical speciation in the 2016v7.2 beta platform.
Temporal allocation is the process of distributing aggregated emissions to a finer temporal
resolution, thereby converting annual emissions to hourly emissions as is required by CMAQ.
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. Temporal allocation
takes these aggregated emissions and distributes the 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, with monthly and day-of-week profiles applied only if the
inventory is not already at that level of detail. See Section 3.3 of the Preparation of Emissions
Inventories for the Version 7.2 2016 North American Emissions Modeling Platform Technical
Support Document46 for more information about the profiles used to temporally allocate
emissions to the 2016v7.2 beta platform.
Spatial allocation is the process of distributing aggregated emissions to a finer spatial resolution,
as is required by CMAQ. There are more than 100 spatial surrogates available for spatially
allocating U.S. county-level emissions to thel2-km grid cells used by the air quality model. See
Section 3.4 of the Preparation of Emissions Inventories for the Version 7.2 2016 North American
Emissions Modeling Platform Technical Support Document46 for a description of the spatial
surrogates used for allocating county-level emissions in the 2016v7.2 beta platform.
45	Yarwood, G., et al. (2010) Updates to the Carbon Bond Chemical Mechanism for Version 6 (CB6). Presented at
the 9th Annual CMAS Conference, Chapel Hill, NC. Available at
https://www.cmascenter.org/conference/2010/abstracts/emerv updates carbon 2010.pdf
46	USEPA (2019). Technical Support Document: Preparation of Emissions Inventories for the Version 7.2 2016
North American Emissions Modeling Platform. Office of Air Quality Planning and Standards, U.S. Environmental
Protection Agency, Research Triangle Park, NC. September 2019. https://www.epa.gov/sites/production/files/2019-
09/documents/2016v7.2_	regionatfaaze	emismod	tsd	508.pdf
29

-------
6.7 Air Quality Modeling Methodology
CMAQ was ran to generate hourly concentration predictions for ozone, PM2.5 component
species, nitrogen and sulfate deposition, nitrogen dioxide, carbon monoxide, and a subset of air
toxics (formaldehyde, acetaldehyde, acrolein, benzene, 1,3-butadiene, and naphthalene) for each
grid cell in the modeling domain.
The 12-kilometer (km) CMAQ modeling domain was modeled for the entire year of 2016. The
12 km domain simulations included a "ramp-up" period, comprised of 10 days before the
beginning of the annual simulation, to mitigate the effects of initial concentrations. The ramp-up
period is not included in the output analyses.
For the 8-hour ozone results, we are only using modeling results from the period between May 1
and September 30, 2016. This 153-day period generally conforms to the ozone season across
most parts of the U.S. and contains the majority of days with observed high ozone concentrations
in 2016. Data from the entire year were utilized when estimating nitrogen and sulfate deposition,
visibility, and PM2.5, nitrogen dioxide, carbon monoxide, and toxics impacts.
6.7.1 Air Quality Model
CMAQ is a non-proprietary computer model that simulates the formation and fate of
photochemical oxidants, primary and secondary PM concentrations, acid deposition, and air
toxics, over regional and urban spatial scales for given inputs of meteorological conditions and
emissions. 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. The CMAQ model is a well-known and well-respected tool and has been used in
numerous national and international applications.47'48'49'50'51
This 2016v7.2 beta platform used the most recent multi-pollutant CMAQ code available at the
time of air quality modeling (CMAQ version 5.2.1).52 The 2016 CMAQ runs utilized the CB6r3
chemical mechanism (Carbon Bond with linearized halogen chemistry) for gas-phase chemistry,
and AER06 (aerosol model with non-volatile primary organic aerosol) for aerosols. CMAQ
47	Hogrefe, C., et al. (2004) Simulating regional-scale ozone climatology over the eastern United States: model
evaluation results. Atmos. Environ., 38(17), 2627-2638.
48	USEPA (2016). Air Quality Modeling Technical Support Document: Heavy-Duty Vehicle Greenhouse Gas Phase
2 Final Rule. EPA-420-R-16-007.
49	Lin, M., et al. (2008) Long range transport of acidifying substances in East Asia Part I: Model evaluation and
sensitivity studies. Atmos. Environ., 42(24), 5939-5955.
50	USEPA (2009). Technical Support Document for the Proposal to Designate an Emissions Control Area for
Nitrogen Oxides, Sulfur Oxides, and Particulate Matter. EPA-420-R-007, 329 pp.
https://19ianuarv2017snapshot.epa.gov/sites/production/files/2016-09/documents/420r090Q7.pdf
51	Simon, H., Baker, K., Phillips, S., 2012: Compilation and interpretation of photochemical model performance
statistics published between 2006 and 2012. Atmos. Environ. 61, 124-139.
52	CMAQ version 5.2.1: doi: 10.5281; https://zetiodo.org/record/12.1.260.1. Model code for CMAQ \ 5.2.1 is also
available from the Community Modeling and Analysis System (CMAS) at: http://www.cmascenter.org
30

-------
versions 5.0.2 and 5.1 were most recently peer-reviewed in September of 2015 for the U.S.
EPA.53
6.7.2 Model Domain and Configuration
The CMAQ modeling analyses used a domain covering the continental United States, as shown
in Figure 6.1. This single domain covers the entire continental U.S. (CONUS) and large portions
of Canada and Mexico using 12 km x 12 km horizontal grid spacing. The 2016 simulation used
a Lambert Conformal map projection centered at (-97, 40) with true latitudes at 33 and 45
degrees north. The model extends vertically from the surface to 50 millibars (approximately
17,600 meters) using a sigma-pressure coordinate system with 35 vertical layers. Table 6.10
provides some basic geographic information regarding the CMAQ domains and Table 6.11
provides the vertical layer structure for the CMAQ domain.
Table 6.10 Geographic elements of domains used in air quality modeling

CMAQ Modeling Configuration
Grid Resolution
12 km National Grid
Map Projection
Lambert Conformal Projection
Coordinate Center
97 deg W, 40 deg N
True Latitudes
33 deg N and 45 deg N
Dimensions
396 x 246 x 35
Vertical extent
35 Layers: Surface to 50 millibar level
(see Table 6.11)
Table 6.11 Vertical layer structure for 2016 CMAQ anti-backsliding simulations
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
53Moran, M.D., etal. (2015) Final Report: Fifth Peer Review of the CMAQ Model,
llltps^^	Qoments/cmaq^^	This peer
review was focused on CMAQ v5.0.2, which was released in May 2014, as well as CMAQ v5.1, 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
31

-------
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
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
32

-------
¦241 2000iVi, V1620ftu0i
12US2 domain
x,y origin:
col: 396 row:246
Figure 6.1 Map of the CMAQ 12 km modeling domain (noted by the purple box)
6.7.3 Model Inputs
The key inputs to the CMAQ model include emissions from anthropogenic and biogenic sources,
meteorological data, and initial and boundary conditions.
The onroad and nonroad emissions inputs used for the 2016 "with-RFS" and "pre-RFS"
scenarios are summarized in Section 6.2 and Section 6.3 of this document, respectively, and
emissions inputs for other sectors are described in Section 6.5 and in the documentation for the
2016v7.2 beta modeling platform.54
54 USEPA (2019). Technical Support Document: Preparation of Emissions Inventories for the Version 7.2 2016
North American Emissions Modeling Platform. Office of Air Quality Planning and Standards, U.S. Environmental
Protection Agency. Research Triangle Park, NC. September 2019. https://www.epa.gov/sites/production/
files/2019-09/documents/2016v7.2 regionalhaze emismod tsd 508.pdf
33

-------
The CMAQ meteorological input files were derived from simulations of the Weather Research
and Forecasting Model (WRF) version 3.8 for the entire 2016 year.55'56 The WRF Model is a
state-of-the-science mesoscale numerical weather prediction system developed for both
operational forecasting and atmospheric research applications.57 The meteorological outputs
from WRF were processed to create 12 km model-ready inputs for CMAQ using the
Meteorology-Chemistry Interface Processor (MCIP) version 4.3. These inputs included hourly
varying horizontal wind components (i.e., speed and direction), temperature, moisture, vertical
diffusion rates, and rainfall rates for each grid cell in each vertical layer.58
The boundary and initial species concentrations were provided by a northern hemispheric
CMAQ modeling platform for the year 2016.59'60 The hemispheric-scale platform uses a polar
stereographic projection at 108 km resolution to completely and continuously cover the northern
hemisphere for 2016. Meteorology is provided by WRF v3.8. Details on the emissions used for
hemispheric CMAQ can be found in the 2016 hemispheric emissions modeling platform TSD.61
The atmospheric processing (transformation and fate) was simulated by CMAQ (v5.2.1) using
the CB6r3 and the aerosol model with non-volatile primary organic carbon (AE6nvPOA). 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.
6.7.4 Ozone and PM2.5 Fused Fields
Fused fields are spatial fields, or surfaces, of gridded modeled concentrations (within the 12 km
modeling grid used in this analysis) where the model output has been adjusted using monitored
data. The fused fields use ambient concentration data from monitors to adjust the modeled
concentration in each grid cell to match observed data at locations of monitoring sites. In grid
cells where monitor data does not exist, data is interpolated between monitors using the
enhanced Veronoi Neighbor Average (eVNA) method to create an uninterrupted surface of
monitored concentrations which can be used to adjust the modeled data.
This results in a gridded future-year projection which accounts for measured values and is a way
to attempt to minimize model bias problems posed by imperfect model performance on
55	Skamarock, W.C., et al. (2008) A Description of the Advanced Research WRF Version 3.
https://openskv.ucar.edn/islandora/obiect/teclinotes:500
56	U SEP A (2019). Meteorological Model Performance for Annual 2016 Simulation WRF v3.8
https://www3.epa. gov/ttn/scram/repoits/Met	Model_Performance-20.1.6	WRF. pdf. EPA-454/R-19-010.
57	https://www.mnim.ucar.edu/weather-research-and-forecasting-model
58	Byun, D.W., Ching, J. K.S. (1999). Science algorithms of EPA Models-3 Community Multiscale Air Quality
(CMAQ) modeling system, EPA/600/R-99/030, Office of Research and Development. Please also see:
https://www.cmascenter.org/
59	Henderson, B., et al. (2018) Hemispheric-CMAQ Application and Evaluation for 2016, Presented at 2019 CMAS
Conference, available https://cmascenter.Org/conference//2018/slides/0850 henderson hemispheric-
email application 2018.pptx
60	Mathur, R., et al. (2017) Extending the Community Multiscale Air Quality (CMAQ) modeling system to
hemispheric scales: overview of process considerations and initial applications, Atmos. Chem. Phys., 17, 12449-
12474, https://doi.org/.1.0.5.1.94/a	-20.1.7.
61	U SEP A (2019). Technical Support Document: Preparation of Emissions Inventories for the Version 7.1 2016
Hemispheric Emissions Modeling Platform. Office of Air Quality Planning and Standards.
34

-------
individual days. The monitoring network does not allow fused fields to be generated for all
pollutants, but ozone and PM2.5 had enough monitored ambient data coverage to make it feasible
to create fused field surfaces. Additional information on creating the fused fields can be found in
the Appendix, see Section 9.2.
6.7.5 CMAQ Evaluation
The CMAQ predictions for ozone, fine particulate matter, sulfate, nitrate, ammonium, organic
carbon, elemental carbon, nitrogen and sulfur deposition, and specific air toxics (formaldehyde,
acetaldehyde, acrolein, benzene, 1,3-butadiene, and naphthalene) from the 2016 "with-RFS"
scenario were compared to measured concentrations in order to evaluate the ability of the
modeling platform to replicate observed concentrations. This evaluation was comprised of
statistical and graphical comparisons of paired modeled and observed data. Details on the model
performance evaluation including a description of the methodology, the model performance
statistics, and results are provided in the Appendix, Section 9.3.
7 Air Quality Modeling Results
As described in Section 3, this study assessed the air quality impact of vehicle and engine
emissions in 2016 for two scenarios, a "pre-RFS" scenario where renewable fuel use was at
approximately 2005 levels and a "with-RFS" scenario where renewable fuel use was at 2016
levels. This section of the report presents modeled changes in ambient concentrations of air
pollutants when comparing the "pre-RFS" and "with-RFS" scenarios.
•	Decreases in concentration mean that the "with-RFS" scenario decreases the pollutant
concentration compared to the "pre-RFS" scenario.
•	Increases in concentration mean that the "with-RFS" scenario increases the pollutant
concentration compared to the "pre-RFS" scenario.
Everything in the two modeled scenarios was held constant except the onroad and nonroad
inventories, which reflected the differing fuel supplies used to develop the emissions. This
includes the vehicle and engine population, activity, base emission rates, and meteorological data
(reflecting calendar year 2016 conditions) and the emissions for all other sources, including
boundary conditions and initial conditions used in the air quality modeling methodology. This
study assumed that California fuels were the same in both the "pre-RFS" and "with-RFS"
scenarios, and as a result, we did not model California.
7.1 Ozone
Figure 7.1 and Figure 7.2 show the absolute change and percent change in the maximum average
8-hour ozone concentrations when comparing the "with-RFS" scenario to the "pre-RFS"
scenario. The results shown are for the time period of May 1st through September 30th, 2016.62
Compared to the "pre-RFS" scenario, the "with-RFS" scenario increases ozone concentrations in
62 The months of May to September are commonly known as the ozone season and represent the timeframe when
most high ozone concentrations occur.
35

-------
the eastern United States, particularly the southeast, while ozone concentrations are unchanged in
some areas of the western United States. Some localized decreases also occur. Tabular results
for all grid cells are included with the online supplemental materials.63
~ £
Max: 1.1973 Min: -0.5563
PPb
-0.6 to -0.3
-0.3 to -0.1
-0.1 to 0.1
0.1 to 0.3
0.3 to 0.5
0.5 to 0.7
0.7 to 0.9
0.9 to 1.1
1.1 to 1.4
Figure 7.1 Change in absolute concentrations of 8-hour maximum average ozone between "pre-
RFS" and "with-RFS" scenarios
63 https://www.epa.gov/renewable-fuel-standard-program/anti-backsliding-determination-and-study
36

-------
>f"T
Max: 2.8764 Miry -2.5115
%
-2.9 to -2.1
-2.1 to -1.5
-1.5 to -0.9
-0.9 to -0.3
-0.3 to 0.3
0.3 to 0.9
0.9 to 1.5
1.5 to 2.1
2.1 to 2.9
Figure 7.2 Percent change in concentrations of 8-hour maximum average ozone between "pre-
RFS" and "with-RFS" scenarios
As presented in Sections 6.2 and 6.3, emissions of NOx and VOC increase from onroad and
nonroad vehicles in the "with-RFS" scenario, as compared to the "pre-RFS" scenario. Relatively
small amounts of NOx enable ozone to form rapidly when VOC levels are relatively high; such
conditions are called "NOx-limited." Rural areas are usually NOx-limited, due to the relatively
large amounts of biogenic VOC emissions in such areas. The southeastern United States has
high levels of biogenic VOC emissions and is known as a NOx-limited region, so it is likely that
the modeled increases in NOx emissions lead to the ozone increases associated with the "with-
RFS" scenario.64
When NOx levels are relatively high and VOC levels relatively low, NOx forms inorganic
nitrates (i.e., particles) but relatively little ozone. Such conditions are called "VOC-limited."
Under these conditions, increases in NOx can decrease local ozone. The localized decreases
shown in Figure 7.1 and Figure 7.2 are likely occurring in VOC-limited areas due to this
tendency for NOx to form particles instead of ozone under certain circumstances.
7.2 Particulate Matter
Figure 7.3 and Figure 7.4 show the absolute change and percent change that would occur in the
modeled annual average 2016 PM2.5 concentrations when comparing the "with-RFS" scenario to
the "pre-RFS" scenario. Figure 7.5 presents the absolute change in the modeled average annual,
average January, and average July 2016 PM2.5 concentrations when the "with-RFS" scenario is
compared to the "pre-RFS" scenario. PM2.5 concentrations remain relatively unchanged in most
of the United States, but the "with-RFS" scenario results in PM2.5 increases in some areas and
64 Simon el al. (2014). Ozone Trends Across the United States Over a Period of Decreasing NOx and VOC
Emissions, dx.doi.org/10.102t/es504514z | Environ. Sci. Technol. 2015, 49,186-195
37

-------
some localized decreases in other areas, as compared to the "pre-RFS" scenario. The monthly
maps have larger maximum increases and decreases because they are averaged over a shorter
time period. The maximum increases and decreases in January are larger than they are in July,
potentially due to winter inversions. Tabular results for all grid cells are included with the online
supplemental materials.65
Max: 0.0151 Min: -0.0303
ug/m3
-0.034 to -0.026
-0.026 to -0.019
-0.019 to -0.012
-0.012 to -0.005
-0.005 to 0.005
0.005 to 0.012
0.012 to 0.019
0.019 to 0.026
0.026 to 0.034
Figure 7.3 Absolute change in average annual 2016 PM2.5 concentrations between "pre-RFS" and
"with-RFS" scenarios
m https://www.epa.gov/renewable-fuel-standard-program/anti-backsliding-deteniiination-and-studv
38

-------
ft
L
t
I I

1 \
H
j
Jj •' v
H 1
J y
*
i
.4
		 ;Cf
iff ¦¦¦PW
	 4	^~-r/
L--^?
¦ if"!'-
s0
7
u - -i
Wax: 0.6272 Min: -0.3435 - ' "V ¦"
A s
kf
>S~f~
*"W>V
v?:
S \
\ N: ^
V
: . *%_-
-0.41 to -0.29
-0.29 to -0.17
-0.17 to -0.05
-0.05 to 0.05
0.05 to 0.17
0.17 to 0.29
0.29 to 0.41
0.41 to 0.53
0.53 to 0.65
Figure 7.4 Percent change in average annual 2016 PM2.5 concentrations between "pre-RFS" and
"with-RFS" scenarios
39

-------

.1) 1 % "
[		 ^ \

i ' "¦V
I ( ["iLT
A,, V # -
J w
f' -*" >" .. .•"•X
7;r

Vh n \
i 4 \ \ -":K
I
I
>0.020
0.015
0.010
-0.005
0.000 |
-0005
•0.010
-0.015
0.020
¦4mC

¦i ;U
W

sir
'" T ,-*i" ¦ -
f

f\„
H N-
I
I
>0.020
0.015
0.010
0.005
c
0.000 i
c
-0.005
-0.010
-0.015
-0.020
>0.020
0.015
0.010
0.005
0.000
-0.005
-0.010
-0.015
<-0.020
Figure 7.5 Absolute difference in (a) annual average, (b) January average, and (c) July average
PM2.5 concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
40

-------
PM2.5 is emitted directly from vehicles and is also formed through atmospheric chemical
reactions of gaseous emissions (e.g., sulfur oxides (SOx), nitrogen oxides (NOx) and volatile
organic compounds (VOCs)); the former is often referred to as "primary" PM2.5, and the latter as
"secondary" PM2.5. Particles can remain in the atmosphere for days to weeks and travel through
the atmosphere hundreds to thousands of kilometers, and particle concentration and composition
is affected by several weather-related factors, such as temperature, clouds, humidity, and wind.
Thus, PM2.5 concentrations are made up of a complex mixture of different components including
sulfates, nitrates, and organic compounds, and these components vary over time and space. On
average, the "with-RFS" scenario has higher emissions of NOx, SO2, and VOC, and lower
emissions of primary PM2.5, compared to the "pre-RFS" scenario. (See emission summaries in
Section 6.) It is likely that the increased PM2.5 concentrations associated with the "with-RFS"
scenario are due to increases in secondary PM2.5 which outweigh any decreases in primary PM2.5
concentrations in those areas, and areas with decreased PM2.5 concentrations are due to decreases
in primary PM2.5 concentrations which outweigh any increases in secondary PM2.5
concentrations in those areas.
7.3 Nitrogen Dioxide
Figure 7.6 and Figure 7.7 show the absolute change and percent change in the annual average
NO2 concentrations when comparing the "with-RFS" scenario to the "pre-RFS" scenario.
Tabular results for all grid cells are included with the online supplemental materials.66 The
"with-RFS" scenario results in increases across the eastern U.S. and in some areas in the western
U.S; with larger increases in some urban areas compared to the "pre-RFS" scenario. These
absolute increases in annual average NO2 concentrations are also evident when comparing the
"pre-RFS" scenario to the "with-RFS" scenario as a percent difference. Additional monthly
(January and July 2016) average NO2 difference maps, both absolute difference and percent
difference, are available in the Appendix, Section 9.4.
66 https://www.epa.gOY/renewable-fuel-standard-program/anti-backsliding-determination-aiid-stiidY
41

-------
ppb
< -0.30
^ -0.30 to-0.20
^ -0.20 to-0.10
-0.10 to -0.01
-0.01 to 0.01
0.01 to 0.10
0.10 to 0.20
¦¦ 0.20 to 0.30
^ > 0.30
Figure 7.6 Absolute change in average annual 2016 NO2 concentrations between "pre-RFS" and
"with-RFS" scenarios
%
< -10.0
-10.0 to-5.0
H -5.0 to-1.0
-1.0 to -0.1
-0.1 to 0.1
0.1 to 1.0
1.0 to 5.0
¦¦ 5.0 to 10.0
¦ > 10.0
Figure 7.7 Percent change in average annual 2016 NO2 concentrations between "pre-RFS" and
"with-RFS" scenarios
7.4 Carbon Monoxide
Figure 7.8 and Figure 7.9 show the absolute change and percent change in the annual average
CO concentrations when comparing the "with-RFS" scenario to the "pre-RFS" scenario. Tabular
results for all grid cells are included with the online supplemental materials.67 Compared to the
67 https://www.epa.gov/renewable-fuel-standard-program/anti-backsliding-deteniiination-and-studv
Max: 0.3469 Min: -0.0003
Max: 15.2024 Miri: -0.0241
42

-------
"pre-RFS" scenario, the "with-RFS" scenario results in decreases across the eastern U.S. and
some areas in the western U.S. with some larger decreases in some areas. These absolute
decreases in annual average CO concentrations are also reflected in the percent difference map
for annual average CO concentrations when comparing the "with-RFS" and "pre-RFS"
scenarios. Additional monthly (January and July 2016) average CO difference maps, both
absolute difference and percent difference, are available in the Appendix, Section 9.4.
a
k

* i s
. *r
L JKTt,	\ f \ f
4.	\i3:
"¦ i - . - - [-¦'
-	y
W"

¦A
&
j
\ W:
\
\ % '
X-4:-

\
/ X \ '
Max: -0.0129 Min: -44.6621 ¦?' X.
\- * 1
V
4.^ i4
\ \
¦J*?--.
PPb
-49.0 to-43.0
¦¦ -43.0 to -36.0
^ -36.0 to -29.0
¦¦ -29.0 to-22.0
¦¦ -22.0 to-15.0
-15.0 to -8.0
-8.0 to -1.0
-1.0 to -0.1
-0.1 to 0.1
Figure 7.8 Absolute change in average annual 2016 CO concentrations between "pre-RFS" and
"with-RFS" scenarios
43

-------
%
< -12.0
-12.0 to -10.0
¦10.0 to -8.0
-8.0 to -6.0
-6.0 to -4.0
-4.0 to -2.0
-2.0 to -1.0
-1.0 to -0.5
> -0.5
Figure 7.9 Percent change in average annual 2016 CO concentrations between "pre-RFS" and
"with-RFS" scenarios
7.5 Air Toxics (acetaldehyde, acrolein, benzene, 1,3-butadiene, formaldehyde,
naphthalene)
In this section we describe results of our modeling of air toxics concentrations for the "pre-RFS"
and "with-RFS" scenarios. Although there are many compounds which are considered air toxics,
we focused on the following six pollutants: acetaldehyde, acrolein, benzene, 1,3-butadiene,
formaldehyde and naphthalene. These pollutants have been identified as national or regional-
scale cancer and noncancer risk drivers in the 2014 or past National Air Toxics Assessments
(NATAs).68'69 Ambient levels of air toxics pollutants dominated by primary emissions (or a
decay product of a directly emitted pollutant), such as benzene and 1,3-butadiene, have the
largest impacts. Air toxics that primarily result from photochemical transformation, such as
formaldehyde and acetaldehyde, are not impacted as much as those dominated by direct
emissions. Additional monthly (January and July 2016) absolute and percent difference maps
are available in the Appendix, 9.4 and tabular results for all grid cells are included with the
online supplemental materials.70
68 USEPA (2015). 2011 NATA: Assessment Results. >.ttps://www.epa.gov/national-air-toxics-assessment/2011-
nata-assessment-results
® USEPA (2018). Technical Support Document EPA 's 2014 National Air Toxics Assessment. Office of Air Quality
Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC. August 2018.
https://www.epa.gov/national-air-toxics-assessment/2014-nata-assessment-results
70 https://www.epa.gov/renewable-fuel-standard-program/anti-backsliding-determination-and-studv
44

-------
7.5.1 Acetaldehyde
Figure 7.10 and Figure 7.11 show the absolute change and percent change in the annual average
acetaldehyde concentrations when comparing the "pre-RFS" scenario to the "with-RFS"
scenario. Compared to the "pre-RFS" scenario, the "with-RFS" scenario results in increases
across much of the eastern U.S. and some areas in the western U.S., with larger increases in
some areas. The percent difference map for annual average acetaldehyde comparing the "with-
RFS" and "pre-RFS" scenarios also shows increases with the "with-RFS" scenario, mainly in the
upper midwest, Florida, and some urban areas in the western U.S.
45

-------
¥

I

/ A
- "H	--
/ - ¦;		 r
1,.'-.^-^.!	\ 1	J
A- '-'k. _ '"	V,. #r
& » ; .v u-I
U"-rv4r..
— --I ( \ - r.rfir..-	-^v. W'
V-'
*2>-
i
.} '
—
?
I

I
; 1
1 ^|f
r---	«%
'5-
ug/m3
< -0.300
-0.300 to -0.200
-0.200 to -0.100
-0.100 to -0.010
-0.010 to -0.001
-0.001 to 0.001
0.001 to 0.010
0.010 to 0.100
0.100 to 0.200
0.200 to 0.300
> 0.300
Mi \
n m \
s
v
k k
Max: 0.096 Min: -0.0006
|\ /U i
*. > r;	/ \
¦A
z
J-..a* Jv

Figure 7.10 Absolute change in average annual 2016 acetaldehyde concentrations between "pre-
RFS" and "with-RFS" scenarios
m
T l-~-
m §
f

i
fa
?1
f\ y

fcjfcj -{
I _,
> ,_J: \; v
J,- ( >»
^	
7w
— 1
		
•• '4 A
S: Vs Sft.
'Vt-4 S

v m

IK
HSfii
91
j
%
< -50.0
-50.0 to-25.0
-25.0 to -10.0
-10.0 to -5.0
-5.0 to -2.5
-2.5 to -1.0
-1.0 to 1.0
1.0 to 2.5
2.5 to 5.0
5.0 to 10.0
10.0 to 25.0
25.0 to 50.0
> 50.0
\
Max: 8.5787 Min: -0.0631
"¦-4 %
i % VV
	il	\			
\ A
" r !
r^f
•• , t- " ^ V
	
Figure 7.11 Percent change in average annual 2016 acetaldehyde concentrations between "pre-
RFS" and "with-RFS" scenarios
7.5.2 Acrolein
Figure 7.12 and Figure 7.13 show the absolute change and percent change in the annual average
acrolein concentrations when comparing the "pre-RFS" scenario to the "with-RFS" scenario.
Compared to the "pre-RFS" scenario, the "with-RFS" scenario results in some geographically
46

-------
limited increases and decreases in the southwestern U.S., upper midwest, and northeastern U.S.
on the percent difference map. These percent differences correspond to decreases in absolute
concentration of less than 0.001 ug/m ', which correspond to the color gray on the map.
Max: 0.0007 Min: -0.0007
ug/mB
< -0.005
¦¦ -0.005 to -0.004
HI -0.004 to -0.003
-0.003 to -0.002
-0.002 to -0.001
-0.001 to 0.001
0.001 to 0.002
0.002 to 0.003
0.003 to 0.004
M 0.004 to 0.005
^ > 0.005
Figure 7.12 Absolute change in average annual 2016 acrolein concentrations between "pre-RFS"
and "with-RFS" scenarios

J'T'"-'
i %'
> 	Ut
J t'i
M mrt-T *
EfjEsH
Km

JSI1 - \ /' > 1 N f
i)	i
5
I w
I
i Kf wm
	%
\.h
r
i I P
Max: 14.3353 Min: -3.2322 " ... "
. '•}

< -50.0
-50.0 to -25.0
-25.0 to -10.0
-10.0 to -5.0
-5.0 to-2.5
-2.5 to -1.0
-1.0 to 1.0
1.0 to 2.5
2.5 to 5.0
5.0 to 10.0
10.0 to 25.0
25.0 to 50.0
> 50.0
Figure 7.13 Percent change in average annual 2016 acrolein concentrations between "pre-RFS"
and "with-RFS" scenarios
47

-------
7.5.3 Benzene
Figure 7.14 and Figure 7.15 show the absolute change and percent change in the annual average
benzene concentrations when comparing the "with-RFS" scenario to the "pre-RFS" scenario.
Compared to the "pre-RFS" scenario, the "with-RFS" scenario results in decreases across much
of the country, with larger decreases in some areas. These absolute decreases in annual average
benzene concentrations are also reflected in the percent difference map for annual average
benzene comparing the "with-RFS" and "pre-RFS" scenarios.
.x-.}
0.0 Min: -0.1798
ug/m3
< -0.300
-0.300 to -0.200
-0.200 to -0.100
-0.100 to -0.010
-0.010 to -0.001
-0.001 to 0.001
0.001 to 0.010
0.010 to 0.100
0.100 to 0.200
0.200 to 0.300
> 0.300
Figure 7.14 Absolute change in average annual 2016 benzene concentrations between "pre-RFS"
and "with-RFS" scenarios
48

-------

-0.0134 Min:-13.7637
< -50.0
-50.0 to -25.0
-25.0 to -10.0
-10.0 to -5.0
-5.0 to -2.5
-2.5 to -1.0
-1.0 to 1.0
1.0 to 2.5
2.5 to 5.0
5.0 to 10.0
10.0 to 25.0
25.0 to 50.0
> 50.0
Figure 7.15 Percent change in average annual 2016 benzene concentrations between "pre-RFS"
and "with-RFS" scenarios
7.5.4 1,3-Butadiene
Figure 7.16 and Figure 7.17 show the absolute change and percent change in the annual average
1,3-butadiene concentrations when comparing the "pre-RFS" scenario to the "with-RFS"
scenario. Compared to the "pre-RFS" scenario, the "with-RFS" scenario results in decreases in
many urban areas. The 1,3-butadiene annual average percent difference map shows decreases in
some areas and increases in other areas when comparing "pre-RFS" and "with-RFS" scenarios.
49

-------

V '"JW
$3
ug/m3
< -0.005
-0.005 to -0.004
-0.004 to -0.003
-0.003 to -0.002
-0.002 to -0.001
-0.001 to 0.001
0.001 to 0.002
0.002 to 0.003
0.003 to 0.004
0.004 to 0.005
> 0.005
SC- \
Max: 0.0014 Min: -0.0147
¥
m
1
(
\ ,
*-V
Figure 7.16 Absolute change in average annual 2016 1,3-butadiene concentrations between "pre-
RFS" and "with-RFS" scenarios
Max: 23.0871 Min: -14.431
Figure 7.17 Percent change in average annual 2016 1,3-butadiene concentrations between "pre-
RFS" and "with-RFS" scenarios
%
-50.0 to-25.0
¦¦ -25.0 to-10.0
Hi -10.0 to-5.0
-5.0 to -2.5
-2.5 to-1.0
-1.0 to 1.0
1.0 to 2.5
2.5 to 5.0
5.0 to 10.0
H 10.0 to 25.0
H 25.0 to 50.0
^ > 50.0
7.5.5 Formaldehyde
Figure 7.18 and Figure 7.19 show the absolute change and percent change in the annual average
formaldehyde concentrations when comparing the "pre-RFS" scenario to the "with-RFS"
50

-------
scenario. Compared to the "pre-RFS" scenario, the "with-RFS" scenario results in widespread
increases in absolute concentration of formaldehyde across much of the U.S. These absolute
differences correspond to percent increases of less than 1% across much of the U.S.

?~4
\ >,i 1 I
% n
UMi J
<. "v.. >"y \
'%&: \ I
A
/ u %
Max: 0.0454 Min:-0.0053 v.;,	•• ¦.
: I'4
w
\
MA
If"
:	3K*
"W

, } /' "% /
¦C /
.V
%

) L- 0.300
Figure 7.18 Absolute change in average annual 2016 formaldehyde concentrations between "pre-
RFS" and "with-RFS" scenarios
%
ma < -50.0
¦¦ -50.0 to-25.0
¦¦ -25.0 to-10.0
¦¦ -10.0 to-5.0
-5.0 to -2.5
-2.5 to-1.0
-1.0 to 1.0
1.0 to 2.5
2.5 to 5.0
5.0 to 10.0
^ 10.0 to 25.0
^ 25.0 to 50.0
H > 50.0
Figure 7.19 Percent change in average annual 2016 formaldehyde concentrations between "pre-
RFS" and "with-RFS" scenarios
51

-------
7.5.6 Naphthalene
Figure 7.20 and Figure 7.21 show the absolute change and percent change in the annual average
naphthalene concentrations when comparing the "pre-RFS" scenario to the "with-RFS" scenario.
Compared to the "pre-RFS" scenario, the "with-RFS" scenario results in some geographically
limited increases and decreases in the western U.S. and upper midwest on the percent difference
map. These percent differences correspond to decreases in absolute concentration of less than
0.001 ug/m3, which correspond to the color gray on the map.
Max: 0.0002
-0.0011
ug/m3
< -0.005
-0.005 to -0.004
-0.004 to -0.003
-0.003 to -0.002
-0.002 to -0.001
-0.001 to 0.001
0.001 to 0.002
0.002 to 0.003
0.003 to 0.004
0.004 to 0.005
> 0.005
Figure 7.20 Absolute change in average annual 2016 naphthalene concentrations between "pre-
RFS" and "with-RFS" scenarios
52

-------
Max: 1.8195 Min: -1.7713
V\
U
Jr
jt
m

•*4.
W
X


-1.0 to 1.0
1.0 to 2.5
2.5 to 5.0
5.0 to 10.0
10,0 to 25,0
25.0 to 50.0
> 50.0
Figure 7.21 Percent change in average annual 2016 naphthalene concentrations between "pre-
RFS" and "with-RFS" scenarios
8 Study Limitations and Uncertainties
In order to assess the air quality impacts of changes in vehicle and engine emissions resulting
from required renewable fuel volumes, this study used the best models and methods that were
feasible and publicly available for that purpose at the time the study was initiated. However, as
with any study, there are inherent limitations and uncertainties. This section identifies some key
limitations and uncertainties associated with the study.
8.1 Study Scope and Design
This study is narrowly focused on the impacts of required renewable fuel volumes on
concentrations of criteria and toxic pollutants due to changes in vehicle and engine emissions;
this study is not an examination of the lifecycle impacts of renewable fuels on air quality,
greenhouse gases, or other environmental impacts. This study examines only the impacts of
renewable fuel volumes on vehicles and engines, and it does not include "upstream" emissions
impacts associated with production and distribution of renewable fuels and feedstocks.
Specifically, the study holds emissions from all sources constant at 2016 levels in both scenarios,
except for gasoline-fueled vehicles and engines (onroad and nonroad) and onroad diesel-fueled
vehicles outside the state of California. For these non-California onroad and nonroad sources,
only the fuel supply is changed.
This anti-backsliding study examines the impacts of required renewable volumes as compared to
a hypothetical case where renewable fuel usage in 2016 was approximately the same as it had
been in 2005, before EPAct was enacted. This study examines impacts for a single retrospective
year (2016). By analyzing calendar year 2016, EPA was able to use an existing modeling
53

-------
platform that includes known renewable fuel volumes for 2016 and fuel properties based on
actual data (aggregation of refinery batch reports and fuel surveys). The study does not project
future renewable fuel volumes and their impacts, and it does not account for the impacts of the
Tier 3 motor vehicle emissions and fuel standards, which took effect in 2017.71 These standards
lowered the sulfur content of gasoline and tightened the emissions standards for onroad motor
vehicles, resulting in lower emissions of criteria and toxic pollutants and precursors in 2017 and
into the future as more Tier 3-compliant vehicles enter the fleet. These standards are projected to
reduce concentrations of ozone, PM2.5, NO2, toxics (such as acetaldehyde, formaldehyde,
acrolein, benzene, 1,3-butadiene, and naphthalene), and other pollutants into the future. By
examining 2016, this study also does not reflect the full turnover of the diesel fleet to the most
recent highway standards. Thus, the modeling reflects changes in diesel emissions that would be
projected to decline into the future.
8.2 Data and Model Limitations and Uncertainties
In the absence of consistent and reliable data on biodiesel use across the country, this study
assumed in the "with-RFS" scenario that biodiesel usage was at a B5 blend level (5 percent
biodiesel) in all onroad diesel fuel nationwide, generally consistent with aggregate usage figures.
Under ASTM D975, blends up to 5% can be labeled as "diesel fuel." This study did not capture
the impacts of higher or lower biodiesel blends that may have been occurring in specific areas.
This study assumed no biodiesel was being used in the "pre-RFS" scenario, except in California.
Furthermore, this study did not assume any El5 in either the "pre-RFS" or "with-RFS" scenario,
because of low E15 sales, and lack of data about its use in 2016.
Because California had state fuels regulations that affected fuel properties and usage independent
of EPAct and EISA, this study assumed that California fuels were the same in both the "pre-
RFS" and "with-RFS" scenarios, and as a result, we did not model California. More broadly,
because of the very limited data on 2005 fuel properties and their spatial distribution, the "pre-
RFS" scenario is only a general approximation of 2005 fuels. As a result, the air quality
modeling results are illustrative at a broad geographic scale (rather than being locally specific).
With respect to estimating the effects of renewable fuels on emissions, there is much more data
available for onroad gasoline vehicles than for nonroad gasoline engines or for diesel vehicles
and engines. 72 The impact of ethanol on HC, CO and NOx from nonroad gasoline engines is
based on three studies from 1991-1997. 73 There is insufficient data to model fuel effects on PM
emissions from nonroad gasoline engines. Data on fuel effects for nonroad diesel are also
71	USEPA (2014). Control of Air Pollution From Motor Vehicles: Tier 3 Motor Vehicle Emission and Fuel
Standards, 79 FR 23414 (April 28, 2014).
72	USEPA (2016). Fuel Effects on Exhaust Emissions from On-road Vehicles in MOVES2014. EPA-420-R-16-001.
Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental Protection
Agency, Ann Arbor, MI. February 2016. https://nepis.epa.gov/Exe/ZvPDF.cgi?Dockey=P100QSW2.pdf
73	USEPA (2005). Exhaust Emission Effects of Fuel Sulfur and Oxygen on Gasoline Nonroad Engines. EPA-420-R-
05-016. Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental
Protection Agency, Ann Arbor, MI. December 2005. https://iiepis.epa.gov/Exe/ZyPDF.egi?Dockey=P.1..004L80.pdf
54

-------
limited; EPA's last comprehensive evaluation of biodiesel impacts on nonroad exhaust emissions
was completed in 2002.74
More generally, the MOVES2014b model incorporates the data and analysis available when the
model was developed. Updates and improvements have been suggested75'76'77 and are
underway78 that would likely affect the baseline emissions for both light-duty and heavy-duty
emissions, but are unlikely to have a large impact on the individual fuel effects in calendar year
2016.
CMAQ simulates the impact of a very large number of physical and chemical processes on
ambient air quality. Physical processes, such as horizontal and vertical mixing, transport, and
deposition are all based on data that are continually evolving. In addition, the chemical
mechanism specifies a series of chemical reaction pathways and reaction rates. Explicitly
tracking the many thousands of chemical compounds and reactions that occur in the atmosphere
would be too burdensome computationally for air quality models to be practical. Therefore, a
chemical mechanism represents this complexity with only tens or hundreds of "lumped" or
representative species and reactions. Such a simplification introduces additional uncertainty.
74	USEPA (2002). A Comprehensive Analysis of Biodiesel Impacts on Exhaust Emissions (Draft Report). EPA-420-
P-02-001. Assessment and Standards Division, Office of Transportation and Air Quality, U.S. Environmental
Protection Agency, Ann Arbor, MI. October 2002. https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P.1.00lZA0.pdf
75	Heiken, J. G., M. Hixson and J. Lyons (2016). Review of EPA's MOVES2014 Model. E-101. August 11, 2016.
fattp://cresite.wpengine,cotn/wp-content/nploads/2019/05/FI\ <\t hlo l.~Report~SR~20.1.608.1.0~w~€RC~€over~and~
Appendices.pdf
76	Barth (2018) "MOVES Review Work Group Update", presentation to Mobile Source Technical Subcommittee,
May 22, 2018. https://www.epa.gov/sites/production/files/2018-05/documents/052218-moves-wg-barth.Ddf
77	Barth (2017) "MOVES Review Work Group Update", presentation to Mobile Source Technical Subcommittee,
May 31, 2017. https://www.epa.gov/sites/production/files/2017-06/documents/053120217-barth.pdf
78	Beardsley, et al. (2019) "Updates to EPA's Motor Vehicle Emission Simulator (MOVES)", presentation to CRC
Real World Emissions Workshop, March 10, 2019. https://www.epa.gov/sites/production/files/2019-
.1.0/docume nts/09- .1.7-20.1.9-beardslev. pdf
55

-------
9 Appendix A: Emissions Modeling and Air Quality Modeling
9.1 Chemical Mechanisms in Air Quality Modeling
This analysis examines air quality impacts of criteria pollutants including NOx, VOC, CO,
PM2.5, SO2, and air toxics, specifically formaldehyde, acetaldehyde, acrolein, benzene, 1,3-
butadiene, and naphthalene. The air toxics were added as explicit model species to the CB6r3
mechanism used in CMAQv5.2.1.79'80 Emissions of all the pollutants included in the onroad
inventories were generated using MOVES VOC emissions and toxic-to-VOC ratios.81 In
addition to direct emissions, photochemical mechanisms are responsible for formation of some
of these compounds in the atmosphere from precursor emissions. For some pollutants such as
PM, formaldehyde, and acetaldehyde, many photochemical processes are involved. CMAQ
therefore also requires inventories for a large number of other air toxics and precursor pollutants.
Methods used to develop the air quality inventories can be found on the air emissions modeling
platform website.82
In the CB6r3 mechanism, the chemistry of thousands of different VOCs in the atmosphere are
represented by a much smaller number of model species which characterize the general behavior
of a subset of chemical bond types; this condensation is necessary to allow the use of complex
photochemistry in a fully three-dimensional air quality model.83
Complete combustion of ethanol in fuel produces carbon dioxide (CO2) and water (H2O).
Incomplete combustion produces other air pollutants, such as acetaldehyde and other aldehydes,
and the release of unburned ethanol. Ethanol is also present in evaporative emissions. In the
atmosphere, ethanol from unburned fuel and evaporative emissions can undergo
photodegradation to form aldehydes (acetaldehyde and formaldehyde) and peroxyacetyl nitrate
(PAN), and it also plays a role in ground-level ozone formation. Mechanisms for these reactions
are included in CMAQ. Additionally, alkenes and other hydrocarbons are considered because
any increase in acetyl peroxy radicals due to ethanol increases might be counterbalanced by a
decrease in radicals resulting from decreases in other hydrocarbons, particularly alkenes.
79	Yarwood, G., Whitten, G. Z., Jung, J., Heo, G., Allen, D. T. (2010). Development, evaluation and testing of
version 6 of the Carbon Bond chemical mechanism (CB6), Final report to the Texas Commission on Environmental
Quality, Work Order No. 582-7-84005-FY10-26.
80	Luecken, D.J., Yarwood, G., Hutzell, W.T. (2019). Multipollutant modeling of ozone, reactive nitrogen and HAPs
across the continental US with CMAQ-CB6. A linos. Environ., 201, 15 March 2019, 62-72.
81	USEPA (2014) Memorandum to Docket EPA-HQ-OAR-2011-0135 by David Choi: Updates to MOVES for the
Tier 3 FRM Analysis. Document number EPA-HQ-OAR-2011-0135-5063.
82USEPA (2019). Technical Support Document: Preparation of Emissions Inventories for the Version 7.2 2016
North American Emissions Modeling Platform. Office of Air Quality Planning and Standards, U.S. Environmental
Protection Agency, Research Triangle Park, NC. September 2019. https://www.epa.gov/sites/prodnction/files/2019-
09/documents/20.1.6v7.2_regionalhaze_emismod_tsd_508.pdf
83 Dodge, M.C. (2000). Chemical oxidant mechanisms for air quality modeling: critical review. Atmospheric
Environment 34, 2103-2130.
56

-------
CMAQ includes 63 inorganic reactions to account for the cycling of all relevant oxidized
nitrogen species and cycling of radicals, including the termination of NO2 and formation of nitric
acid (HNO3) without PAN formation.84
NO2 + "OH + M (air) —~ HNO3 + M	k=1.19xl0"u cnr^molecule'V1
The CB6r3 mechanism also includes more than 90 organic reactions and contains numerous
updates from CB05 chemical mechanism that include new operators to better represent alkoxy
and peroxy radicals, updated reaction rates throughout, additional explicit species to represent
long-lived and abundant chemicals and species that form SOA, and additional representation of
organic nitrates.85'86
9.1.1 Acetaldehyde
Acetaldehyde is the main photodegradation product of ethanol, as well as other precursor
hydrocarbons. Acetaldehyde is also a product of fuel combustion. In the atmosphere,
acetaldehyde can react with the OH radical and O2 to form the acetyl peroxy radical
[CH3C(0)00-].87 When NOx is present in the atmosphere this radical species can then further
react with nitric oxide (NO), to produce formaldehyde (HCHO), or with nitrogen dioxide (NO2),
to produce PAN [CH3C(0)00N02]. An overview of these reactions and the corresponding
reaction rates are provided below, as published by Atkinson, et al.88
CH3CHO + -OH -> CH3CO + H20	k = 1.5 x 10"11 cn^moleculeV1
CH3CO + 02 + M —~ CH3C(0)00- + M
CH3C(0)00- + NO -> CH3C(0)0- + NOi	k = 2.0 X 10"11 cn^moleculeV1
CH3C(0)0- -> -ch3 + co2
•CH3 + 02 + M -> CH300- + M
CH300- + NO -> CH30- + N02
CH30- + 02 -> HCHO + H02
CH3C(0)00- + N02 + M -> CH3C(0)00N02 + M k = 1.0 X 10"11 cn^moleculeV1
Acetaldehyde can also photolyze (hv), which predominantly produces -Cfb (which reacts as
shown above to form CfbOO-) and HCO (which rapidly forms HO2 and CO):
84	All rate coefficients presented in this section are listed at 298K and, if applicable, 1 bar of air.
85	Emery, C., Jung, J., Koo, B., Yarwood, G. (2015). Improvements to CAMx Snow Cover Treatments and Carbon
Bond Chemical Mechanism for Winter Ozone. Final report for Utah DAQ, project UDAQ PO 480 52000000001.
86	Available at
https://github.eom/USEPA/CMAO/blob/5.2.l/CCTM/docs/Release Notes/€B6 release notes.md#brief-description
87	Acetaldehyde is not the only source of acetyl peroxy radicals in the atmosphere. For example, dicarbonyl
compounds (methylglyoxal, biacetyl, and others) also form acetyl radicals, which can further react to form
peroxy acetyl nitrate (PAN).
88	Atkinson, R., et al. (2005). Evaluated Kinetic and Photochemical Data for Atmospheric Chemistry - IUPAC
Subcommittee on Gas Kinetic Data Evaluation for Atmospheric Chemistry. July 2005 web version.
http://iuDac.Dole-ether.fr/.
57

-------
CHsCHO + hv +2 02 -> CH300- +HO2 + CO X = 240-380 nm 89
As mentioned above, CH3OO can react in the atmosphere to produce formaldehyde (HCHO).
Formaldehyde is also a product of hydrocarbon combustion. In the atmosphere, the most
important reactions of formaldehyde are photolysis and reaction with the OH, with atmospheric
lifetimes of approximately 3 hours and 13 hours, respectively.90'91 Formaldehyde can also react
with NO3 radical, ground state oxygen atom (03P) and chlorine, although these reactions are
much slower. Formaldehyde is removed mainly by photolysis whereas the higher aldehydes,
those with two or more carbons such as acetaldehyde, react predominantly with OH radicals.
The photolysis of formaldehyde is an important source of new hydroperoxy radicals (HO2),
which can lead to ozone formation and regenerate OH radicals.
HCHO + hv + 2 02 -> 2 H02 + CO	X = 240-330 nm 89
H02 + NO -> NO2+ OH
Photolysis of HCHO can also proceed by a competing pathway which makes only stable
products: H2 and CO.
CB6r3 mechanism for acetaldehyde formation warrant a detailed discussion given the increase in
vehicle and engine exhaust emissions for this pollutant and ethanol, which can form
acetaldehyde in the air. Acetaldehyde is represented explicitly in the CB6r3 chemical
mechanism79'92'93 by the ALD2 model species, which can be both formed from other VOCs and
can decay via reactions with oxidants and radicals. The reaction rates for acetaldehyde, as well
as for the inorganic reactions that produce and cycle radicals, and the representative reactions of
other VOCs have all been updated to be consistent with recommendations in the literature.89
The decay reactions of acetaldehyde are fewer in number and can be characterized well because
they are explicit representations. In CB6r3 acetaldehyde can photolyze or react with molecular
oxygen (O (3P)), hydroxyl radical (OH), or nitrate radicals. The reaction rates are based on
expert recommendations, and the photolysis rate is from IUPAC recommendations.89
In CMAQ v5.2.1, the acetaldehyde that is formed from photochemical reactions is tracked
separately from that which is due to direct emission and transport of direct emissions. In CB6r3
there are 25 different reactions that form acetaldehyde in molar yields ranging from 0.02 (ozone
reacting with lumped products from isoprene oxidation) to 2.0 (cross reaction of acylperoxy
radicals, CXO3). The specific parent VOCs that contribute the most to acetaldehyde
concentrations vary spatially and temporally depending on characteristics of the ambient air, but
89	Sander, S.P., et al. (2003). Chemical Kinetics and Photochemical Data for use in Atmospheric Studies, Evaluation
Number 14. NASA Jet Propulsion Laboratory https://ipldataevai.ipLnasa.gov/index.htniL
90	The 3 h lifetime for HCHO photolysis is determined for Jul 1 at noon and 30° N latitude. In most daylight cases, it
is substantially longer than 3 h making the OH reaction far more competitive.
91	Calvert, J. G., et al. (2011) The mechanisms of atmospheric oxidation of the oxygenates. Oxford University Press,
New York/Oxford.
92	Yarwood, G., Rao, S., Yocke, M., Whitten, G.Z., (2005). Updates to the Carbon Bond Mechanism: CB05. Final
Report to the U.S. EPA, RT-0400675. Yocke and Company, Novato, CA.
93	Luecken, D. J., et al. (2008). Effects of using the CB05 vs. SAPRC99 vs. CB4 chemical mechanism on model
predictions: Ozone and gas-phase photochemical precursor concentrations. Atmos. Environ. 42, 5805-5820.
58

-------
alkenes in particular are found to play a large role.94 The IOLE model species, which represents
internal carbon-carbon double bonds, has high emissions and relatively high yields of
acetaldehyde. The OLE model species, representing terminal carbon double bonds, also plays a
role because it has high emissions although lower acetaldehyde yields. Production from
peroxyproprional nitrate and other peroxyacylnitrates (PANX) and aldehydes with 3 or more
carbon atoms can in some instances increase acetaldehyde, but because they also are a sink of
radicals, their effect is smaller. Thus, the amount of acetaldehyde (and formaldehyde as well)
formed in the ambient air, as well as emitted in the exhaust (the latter being accounted for in
emission inventories), is affected by changes in these precursor compounds due to the addition of
ethanol to fuels (e.g., decreases in alkenes would cause some decrease of acetaldehyde, and to a
larger extent, formaldehyde).
The reaction of ethanol (CH3CH2OH) with OH is slower than some other important reactions but
can be an important source of acetaldehyde if the emissions are large. Based on kinetic data for
molecular reactions, the only important chemical loss process for ethanol (and other alcohols) is
reaction with the hydroxyl radical (-OH).95 This reaction produces acetaldehyde (CH3CHO) with
a 90 percent yield.88 The lifetime of ethanol in the atmosphere can be calculated from the rate
coefficient, k, and due to reaction with the OH radical, occurs on the order of a day in polluted
urban areas or several days in unpolluted areas. For example, an atmospheric lifetime for
acetaldehyde under nominal oxidant conditions, OH of 1.0 x 10"6 cm3molecule"1 s"\ would be 3.5
days.
In CB6r3 reaction of one molecule of ethanol yields 0.90 molecules of acetaldehyde. It assumes
the majority of the reaction occurs through H-atom abstraction of the more weakly-bonded
methylene group, which reacts with oxygen to form acetaldehyde and hydroperoxy radical
(HO2), and the remainder of the reaction occurs at the -CH3 and -OH groups, creating
formaldehyde (HCHO), oxidizing NO to NO2 (represented by model species XO2) and creating
glycoaldehyde:
CH3CH2OH + OH -> H02 + 0.90 CH3CHO + 0.05 CH2(OH)CHO + 0.10 HCHO + 0.10 XO2
9.1.2 Organic Aerosols
Organic aerosol (OA) can be classified as either primary or secondary depending on whether it is
emitted into the atmosphere as a particle (primary organic aerosol, POA) or formed in the
atmosphere (secondary organic aerosol, SOA). SOA precursors include volatile organic
compounds (VOCs) as well as low-volatility compounds that can react to form even lower
volatility compounds. Current research suggests SOA contributes significantly to ambient OA
concentrations, and in Southeast and Midwest States may make up more than 50 percent
(although the contribution varies from area to area) of the organic fraction of PM2.5 during the
94	Luecken, D.J., et al. (2012). Regional sources of atmospheric formaldehyde and acetaldehyde, and implications
for atmospheric modeling.Atmos. Environ., 47, 477-490.
95	Atkinson, R., Arey, J. (2003). Atmospheric Degradation of Volatile Organic Compounds. Chem. Rev. 103, 4605-
4638.
59

-------
summer (but less in the winter).96'97 A wide range of laboratory studies conducted over the past
twenty years show that anthropogenic aromatic hydrocarbons and long-chain alkanes, along with
biogenic isoprene, monoterpenes, and sesquiterpenes, contribute to SOA formation.98'99'100'101'102
Modeling studies, as well as carbon isotope measurements, indicate that a significant fraction of
SOA results from the oxidation of biogenic hydrocarbons.103'104 Based on parameters derived
from laboratory chamber experiments as well as predicted via structure-activity relationships or
computational chemistry, SOA chemical mechanisms have been developed and integrated into
air quality models such as the CMAQ model and have been used to predict OA
concentrations.105'106'107
Secondary organic aerosol (SOA) chemistry in CMAQ v5.2 is documented in the work of Pye et
al. (2017) and considers both volatility-driven condensation as well as heterogeneous and
aqueous chemistry.108
In the analysis presented here, the primary organic aerosol (POA) is nonvolatile and undergoes
heterogeneous oxidation.109 Specifically, primary organic aerosol is tracked separately in terms
of its carbon and non-carbon organic matter. Non-carbon organic matter (such as oxygen and
hydrogen) is added to the reduced carbon as a result of heterogeneous reaction with OH. Diesel
POA is emitted with an organic matter to organic carbon (OM/OC) ratio of 1.25. The ratio
96	Lewandowski M., et al. (2008). Primary and secondary contributions to ambient PM in the midwestern United
States, Environ. Sci. Technol. 42(9), 3303-3309.
97	Kleindienst T. E., et al. (2007). Estimates of the contributions of biogenic and anthropogenic hydrocarbons to
secondary organic aerosol at a southeastern U.S. location, Atmos. Environ. 41(37), 8288-8300.
98	Offenberg J. H., et al. (2007). Contributions of Toluene and a-pinene to SOA Formed in an Irradiated Toluene/a-
pinene,NOx/Air Mixture: Comparison of Results Using 14C Content and SOA Organic Tracer Methods, Environ.
Sci. Technol. 41, 3972-3976.
99	Pandis, S.N., Harley, R.A., Cass, G.R., Seinfeld, J.H. (1992). Secondary organic aerosol formation and transport.
Atmos. Environ. 26, 2269-2282.
100	Takekawa, H. Minoura, H. Yamazaki, S. (2003). Temperature dependence of secondary organic aerosol
formation by photo-oxidation of hydrocarbons. Atmos. Environ. 37, 3413-3424.
101	Kleeman, M.J., et al. (2007) Source apportionment of secondary organic aerosol during a severe photochemical
smog episode. Atmos. Environ. 41, 576-591.
102	Robinson, A. L., et al. (2007). Rethinking organic aerosol: Semivolatile emissions and photochemical aging.
Science 315, 1259-1262.
103	Griffin, R. J.; Cocker, D. R.; Seinfeld, J. H.; Dabdub, D. (1999). Estimate of global atmospheric organic aerosol
from oxidation of biogenic hydrocarbons. Geophys. Res. Lett. 26 (17), 2721- 2724.
104	Lewis, C. W.; Klouda, G. A.; Ellenson, W. D. (2004). Radiocarbon measurement of the biogenic contribution to
summertime PM-2.5 ambient aerosol in Nashville, TN. Atmos Environ 38 (35), 6053- 6061.
105	Byun DW, Schere, KL (2006). Review of the Governing Equations, Computational Algorithms, and Other
Components of the Models-3 Community Multiscale Air Quality (CMAQ) Modeling System, J Applied Mechanics
Reviews 59, 51-76.
106	Pye, H. O. T., et al. (2013). Epoxide Pathways to Improve Model Predictions of Isoprene Markers and Reveal
Key Role of Acidity in Aerosol Formation. Environ. Sci. Tech. 47 (19), 11056-11064.
107	Piletic, I. R., Edney, E. O., Bartolotti, L. J. (2013). A computational study of acid catalyzed aerosol reactions of
atmospherically relevant epoxides. l'hys. Chem. Chem. Phys. 15, 18065-18076
i°8 pye [_[ q j C( aj (2017). On the Implications of aerosol liquid water and phase separation for organic aerosol
mass. Atmos. Chem. Phy. 17, 343-369.
109 Simon, H and Bhave, P. (2012). Simulating the Degree of Oxidation in Atmospheric Organic Particles. Environ.
Sci. Technol., 46 (1), 331-339.
60

-------
increases due to exposure with OH. In the absence of removal, this oxidation process results in
increasing organic aerosol concentrations. These OM/OC ratios assist with post-processing of
model output for comparison with measured OC from routine networks.
Over the past 10 years, ambient OA concentrations have been routinely measured in the U.S. and
some of these data have been used to determine, by employing source/receptor methods, the
contributions of the major OA sources, including biomass burning and vehicular gasoline and
diesel exhaust. Since mobile sources are a significant source of VOC emissions, mobile sources
are also an important source of SOA, particularly in populated areas.110'111
Toluene is an important contributor to anthropogenic SOA.112,113 Mobile sources are the most
significant contributor to ambient toluene concentrations as shown by analyses done for the 2014
National Air Toxics Assessment (NATA)114 and the Mobile Source Air Toxics (MSAT) Rule.115
The 2014 NATA indicates that onroad and nonroad mobile sources accounted for around 70
percent (0.51 |ig/m3) of the total average nationwide ambient concentration of toluene (0.71
|ig/m3).
The amount of toluene in gasoline influences the amount of toluene emitted in vehicle exhaust
and evaporative emissions, although, like benzene, some toluene is formed in the combustion
process. In turn, levels of toluene and other aromatics in gasoline are potentially influenced by
the amount of ethanol blended into the fuel. Due to the high octane of ethanol, it greatly reduces
the need for other high-octane components including aromatics such as toluene (which is the
major aromatic compound in gasoline). Since toluene contributes to SOA and the toluene level
of gasoline is decreasing, it is important to assess the effect of these reductions on ambient PM.
In addition to toluene, other mobile-source hydrocarbons such as benzene, ethylbenzene, xylene,
and alkanes form SOA and are treated accordingly in CMAQ v5.2.1.108,116,117 Similar to toluene,
the SOA produced by benzene and xylene from low-NOx pathways is expected to be less volatile
and be produced in higher yields than SOA from high-NOx conditions.113 Oxidation of alkanes
110	Jathar, S. H., et al. (2017). Chemical transport model simulations of organic aerosol in southern California: model
evaluation and gasoline and diesel source contributions. Atmos. Chem. Phys. 17, 4305-4318.
111	Jathar, S. H., et al. (2014). Unspeciated organic emissions from combustion sources and their influence on the
secondary organic aerosol budget in the United States. Proc. Natl. Acad. Sci., Ill (29), 10473-10478.
112	Hildebrandt et al. (2009). High formation of secondary organic aerosol from the photo-oxidation of toluene.
Atmos. Chem. Phys. 9, 2973-2986.
113	Ng, N. L. et al. (2007). Secondary organic aerosol formation from m-xylene, toluene, and benzene. Atmos. Chem.
Phys. 7, 3909-3922.
114	USEPA (2018). Technical Support Document EPA's 2014 National Air Toxics Assessment. Office of Air
Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC. August 2018..
https://www.epa.gov/sites/production/files/2018-09/documents/2014 nata technical support document.pdf
115	USEPA (2007). Regulatory Impact Analysis for the Control of Hazardous Air Pollutants from Mobile Sources
Rule, Chapter 3, Air Quality and Resulting Health and Welfare Effects of Air Pollution from Mobile Sources. 72 FR
8428, February 26, 2007. https://nepis.epa.gov/Exe/ZvPdf.cgi?Dockev=P1004LNN.PDF
116	Pye, H. O. T.; Pouliot, G. A. (2012). Modeling the role of alkanes, polycyclic aromatic hydrocarbons, and their
oligomers in secondary organic aerosol formation. Environ. Sci. Technol. 46 (11), 6041-6047.
117	Carlton, A. G., et al. (2010). Model representation of secondary organic aerosol in CMAQv4.7. Environ. Sci.
Technol., 44(22): 8553-8560.
61

-------
with longer chains as well as cyclic alkanes form SOA with relatively higher yields than small
straight-chain alkanes.118
It is unlikely that ethanol would form SOA directly or affect SOA formation indirectly through
changes in the radical populations due to increasing ethanol exhaust. Nevertheless, scientists at
the U.S. EPA's Office of Research and Development directed experiments to investigate
ethanol's SOA-forming potential.119 The experiments were conducted under conditions where
peroxy radical reactions would dominate over reaction with NO (i.e., irradiations performed in
the absence of NOx and OH produced from the photolysis of hydrogen peroxide). This was the
most likely scenario under which SOA formation could occur, since a highly oxygenated C4
organic could form. As expected, no SOA was produced. From these experiments, the upper
limit for the aerosol yield is less than 0.01 percent based on scanning mobility particle sizer
(SMPS) data. Given the lack of aerosol formation found in these initial smog chamber
experiments, these data were not published.
In general, measurements of OA represent the sum of POA and SOA and the fraction of aerosol
that is secondary in nature can only be estimated. CMAQ has been evaluated against POA and
SOA surrogates estimated from aerosol mass spectrometer (AMS) measurements along with
positive matrix factorization (PMF).108,110>120=121>122>123 AMS methods rely on detection of
heavily fragmented structures which limits the ability to determine specific precursors to SOA as
well as the role of biogenic vs anthropogenic VOC sources.
Upon release into the atmosphere, numerous VOC compounds can react with free radicals in the
atmosphere to form SOA. While this has been investigated in the laboratory, there is relatively
little information available on the specific chemical composition of SOA compounds themselves
from specific VOC precursors. This absence of complete compositional data from the precursors
has made the identification of aromatically-derived SOA in ambient samples challenging, which
in turn has prevented observation-based measurements of individual SOA source contributions to
ambient PM levels.
As a first step in estimating ambient SOA concentrations, EPA has developed a tracer-based
method.97'98 The method is based on using mass fractions of SOA tracer compounds, measured
in smog chamber-generated SOA samples, to convert ambient concentrations of SOA tracer
compounds to ambient SOA concentrations. This method consists of irradiating the SOA
118	Lim, Y.B., Ziemann, P.J. (2009). Effects of Molecular Structure on Aerosol Yields from OH Radical-Initiated
Reactions of Linear, Branched, and Cyclic Alkanes in the Presence of NOx. Environ. Sci. Technol. 43 (7), 2328-
2334.
119	Kleindienst, T.E. (2008). Hypothetical SOA Production from Ethanol Photooxidation. Memo to the Docket EPA-
HQ-0 AR-2005-0161.
120	Pye, H. O. T., et al. (2015). Modeling the current and future roles of particulate organic nitrates in the
southeastern United States. Environ. Sci. Technol. 49 (24), 14195-14203.
121	Murphy, B. N., et al. (2017). Semivolatile POA and parameterized total combustion SOA in CMAQv5.2: impacts
on source strength and partitioning. Atmos. Chem. Phys. 17 (18), 11107-11133.
122	Woody, M. C., et al. (2016). Understanding sources of organic aerosol during CalNex-2010 using the CMAQ-
VBS. Atmos. Chem. Phys. 16 (6), 4081-4100.
123	Baker, K. R. et al. (2015). Gas and aerosol carbon in California: comparison of measurements and model
predictions in Pasadena and Bakcrsfic/c/. Atmos. Chem. Phys., 15, 5243-5258.
62

-------
precursor of interest in a smog chamber in the presence of NOx, collecting the SO A produced on
filters, and then analyzing the samples for highly polar compounds using advanced analytical
chemistry methods. Employing this method, candidate tracers have been identified for several
VOC compounds which are emitted in significant quantities and known to produce SOA in the
atmosphere. Some of these SOA-forming compounds include toluene, a variety of
monoterpenes, isoprene, and P-caryophyllene, the latter three of which are emitted by vegetation
and are more significant sources of SOA than toluene. Smog chamber work can also be used to
investigate SOA chemical formation mechanisms.124'125'126'127
Although these concentrations are only estimates, due to the assumption that the mass fractions
of the smog chamber SOA samples using these tracers are equal to those in the ambient
atmosphere, there are presently limited other means available for estimating the SOA
concentrations originating from individual SOA precursors. Among the tracer compounds
observed in ambient PM2.5 samples are two tracer compounds that have been identified in smog
chamber aromatic SOA samples.127 To date, these aromatic tracer compounds have been
identified in the laboratory for toluene and w-xylene SOA. Additional work is underway by the
EPA to determine whether these tracers are also formed by benzene and other alkylbenzenes
(including o-xylene, ^-xylene, 1,2,4-trimethylbenzene, and ethylbenzene).
One caveat regarding this work is that a large number of VOCs emitted into the atmosphere,
which have the potential to form SOA, have not yet been studied in environmental smog
chambers. These unstudied compounds could produce SOA species that are being used as
tracers for other VOCs thus overestimating the amount of SOA formed in the atmosphere by the
VOCs studied to date. This approach may also estimate entire hydrocarbon classes (e.g., all
methylsubstituted-monoaromatics or all monoterpenes) and not individual precursor
hydrocarbons. Thus, the tracers could be broadly representative and not indicative of individual
precursors. This is still unknown. Also, anthropogenic precursors play a role in formation of
atmospheric radicals and aerosol acidity, and these factors influence SOA formation from
biogenic hydrocarbons.128'129 This anthropogenic and biogenic interaction, important to EPA
and others, needs further study. The issue of SOA formation from aromatic precursors is an
important one to which EPA and others are paying significant attention.
124	Claeys, M. R., et al (2007). Hydroxydicarboxylic acids: Markers for secondary organic aerosol from the
photooxidation of a-pinene. Environ. Sci. Technol. 41(5), 1628-1634.
125	Edney E. O., et al. (2005). Formation of 2-methyl tetrols and 2-methylglyceric acid in secondary organic aerosol
from laboratory irradiated isoprenc/NOx/SO:/air mixtures and their detection in ambient PM2 5 samples collected in
the Eastern United States. Atmos. Environ. 39, 5281-5289.
126	Jaoui, M., et al. (2005). Identification and quantification of aerosol polar oxygenated compounds bearing
carboxylic or hydroxyl groups. 2. Organic tracer compounds from monoterpenes. Environ. Sci. Technol. 39, 5661-
5673.
127	Kleindienst T. E., et al. (2004). Determination of secondary organic aerosol products from the photooxidation of
toluene and their implications in ambient PM2 5. J- Atmos. Chem. 47, 70-100.
128	Pye, H.O.T., et al. (2013). Epoxide pathways improve model predictions of isoprene markers and reveal key role
of acidity in aerosol formation. Environ. Sci. Technol. 47(19), 11056-11064.
129	Carlton, A.G., et al. (2010). To what extents can biogenic SOA be controlled? Environ. Sci. Technol. 44(9),
3376-3380.
63

-------
The aromatic tracer compounds and their mass fractions have been used to estimate monthly
ambient aromatic SOA concentrations from March 2004 to February 2005 in five U.S.
Midwestern cities.130 The annual tracer-based SOA concentration estimates were 0.15, 0.18,
0.13, 0.15, and 0.19 jag carbon/m3 for Bondville, IL, East St. Louis, IL, Northbrook, IL,
Cincinnati, OH and Detroit, MI, respectively, with the highest concentrations occurring in the
summer. On average, the aromatic SOA concentrations made up 17 percent of the total SOA
concentration. Thus, this work suggests that we are finding ambient PM levels on an annual
basis of about 0.15 [j,g/m3 associated with present toluene levels in the ambient air in these
Midwest cities. Based on preliminary analysis of recent laboratory experiments, it appears the
toluene tracer could also be formed during photooxidation of some of the xylenes.97
Over the past decade a variety of modeling studies have been conducted to predict ambient SOA
levels. While early studies focused on the contribution of biogenic monoterpenes, additional
precursors, such as sesquiterpenes, isoprene, benzene, toluene, and xylene, have been
implemented in atmospheric models such as GEOS-Chem, PMCAMx, and
CMAQ.112'113'116'131'132' 133>134 The most generally available routine measurements available from
monitoring networks for model evaluations are ambient OC or estimated OA concentrations.
Without a method to attribute measured OC to different sources or precursors, identifying causes
of the underestimates in modeled OC via model/measurement comparisons can be challenging.
However, analysis of SOA concentrations in Pasadena and Bakersfield, California during 2010
indicate CMAQ-predicted SOA from toluene and xylene is underestimated despite overestimates
of the VOC precursors.123 In addition, CMAQ-predicted aromatic SOA was underestimated in
the Midwest U.S. despite reasonable predictions of primary organic aerosol tracers, implying
underestimated SOA yields.135
Anthropogenic emissions of NOx and SOx are known to modulate the abundance of SOA from
oxidation of biogenic VOCs, thus allowing for additional influences of vehicle emissions on
130	Lewandowski, M., et al. (2008). Primary and secondary contributions to ambient PM in the midwestern United
States. Environ. Sci. Technol. 42(9), 3303-3309.
131	Henze, D. K., Seinfeld, J. H. (2006). Global secondary organic aerosol from isoprene oxidation. Geophys. Res.
Lett. 33: L09812. doi:10.1029/2006GL025976.
132	Henze, D. K., et al. (2008). Global modeling of secondary organic aerosol formation from aromatic
hydrocarbons: high-vs. low-yield pathways. Atmos. Chem. Phys., 8, 2405-2420, doi:10.5194/acp-8-2405-2008.
133	Lane, T. E., Donahue, N.M. and Pandis, S.N. (2008). Simulating secondary organic aerosol formation using the
volatility basis-set approach in a chemical transport model, Atmos. Environ., 42, 7439-7451.
134	Parikh, H.M., et al. (2011). Modeling secondary organic aerosol using a dynamic partitioning approach
incorporating particle aqueous-phase chemistry, Atmos. Environ., 45, 1126-1137.
135	Napelenok, S. L., et al. (2014). Diagnostic air quality model evaluation of source specific primary and secondary
fine particulate carbon, Environ. Sci. Technol., doi: 10.1021/es4033024w.
64

-------
ambient PM2.5.136'137'138'139'140'141 The SOA that results from oxidation of isoprene is one of the
most evaluated and constrained SOA systems in CMAQ. In CMAQ, most isoprene SOA results
from acid-catalyzed reactions of later-generation isoprene oxidation products (specifically
isoprene-epoxydiols or IEPOX). The original implementation was documented in Pye et al. 2013
(introduced v5.1) with parameter updates v5.2 that improved the magnitude and speciation of the
resulting SOA (Pye et al. 2017). The isoprene SOA has been evaluated against speciated
isoprene SOA measurements across the U.S. (Pye et al. 2013, 2-methyltetrol and organosulfates),
speciated isoprene SOA from filter measurements at Look Rock, TN (Budisulistiorini et al.
2017), and isoprene SOA surrogates (specifically AMS PMF factors) at Centerville AL and Look
Rock TN (Pye et al. 2017).142 Constraints for the monoterpene system are only starting to be
leveraged, but indicate NOx plays a major role via its regulation of oxidant abundance and by
promoting oxidation via the nitrate radical resulting in high SOA yields.136'138
9.1.3 Ozone
As mentioned above, the addition of ethanol to fuels has been shown to contribute to PAN
formation and this is one way for it to contribute therefore to ground-level ozone formation
downwind of NOx sources. PAN is a reservoir and carrier of NOx and is the product of acetyl
radicals reacting with NO2 in the atmosphere. One source of PAN is the photooxidation of
acetaldehyde, but many VOCs have the potential for forming acetyl radicals and therefore PAN
or a PAN-type compound.143 PAN can undergo thermal decomposition with a lifetime of
approximately 1 hour at 298K or 148 days at 250K.
CH3C(0)00N02 + M -> CH3C(0)00- + NOi + M	k = 3.3 x 10"4 s"1 88
The reaction above shows how NO2 is released in the thermal decomposition of PAN, along with
a peroxy radical which can oxidize NO to NO2 and form other species that convert NO to N02
136	Pye, H.O.T., et al. (2019). Anthropogenic enhancements to production of highly oxygenated molecules from
autoxidation. P. Natl. Acad. Sci., 116 (14), 6641-6646.
137	Zhang, H. F., et al. (2018). Monoterpenes are the largest source of summertime organic aerosol in the
southeastern United States. P. Natl. Acad. Sci., 115 (9), 2038-2043.
138	Ng, N. L., et al. (2017). Nitrate radicals and biogenic volatile organic compounds: oxidation, mechanisms, and
organic aerosol. Atmos. Chem. Phys., 17 (3), 2103-2162.
139	Pye, H. O. T., et al. (2013). Epoxide pathways improve model predictions of isoprene markers and reveal key
role of acidity in aerosol formation. Environ. Sci. Technol., 47 (19), 11056-11064.
140	Weber, R. J., et al. (2007). A study of secondary organic aerosol formation in the anthropogenic-influenced
southeastern United States. J. Geophys. Res. Atmos., 112, D13302.
141	Xu, L., et al. (2015). Effects of anthropogenic emissions on aerosol formation from isoprene and monoterpenes in
the southeastern United States. P. Natl. Acad. Sci., 112 (1), 37-42.
142	Budisulistiorini, S. H. (2017). Simulating aqueous-phase isoprene-epoxydiol (IEPOX) secondary organic aerosol
production during the 2013 Southern Oxidant and Aerosol Study (SOAS). Environ. Sci. Technol. 51 (9), 5026-5034.
143	Many aromatic hydrocarbons, particularly those present in high percentages in gasoline (toluene, m-, 0-, p-
xylene, and 1,3,5-, 1,2,4-trimethylbenzene), form methylglyoxal and biacetyl, which are also strong generators of
acetyl radicals (Smith, D.F., T.E. Kleindienst, C.D. Mclver (1999). Primary product distribution from the reaction of
OH with m-, p-xylene and 1,2,4- and 1,3,5-Trimethylbenzene. J. Atmos. Chem., 34: 339- 364).
65

-------
through photochemical reactions, as previously shown in Section 9.1.1. NO2 further photolyzes
to produce ozone (O3).
NOi + hv -> NO + 0(3P)	X = 300-400 nm 89
0(3P) + 02 + M -> Os + M
The temperature sensitivity of PAN allows it to be stable enough at low temperatures to be
transported long distances before decomposing to release NO2. NO2 can then participate in
ozone formation in regions remote from the original NOx source.144 A discussion of CB6
mechanisms for ozone formation can be found in Yarwood et al. (2010).79
Another important way that ethanol fuels contribute to ozone formation is by increasing the
formation of new radicals through increases in formaldehyde and acetaldehyde. The photolysis
of both aldehydes results in up to two molecules of either hydroperoxy radical or methylperoxy
radical, both of which oxidize NO to NO2 leading to ozone formation.
9.1.4 Uncertainties Associated with Chemical Mechanisms
A key source of uncertainty with respect to the air quality modeling results is the photochemical
mechanisms in CMAQ. Pollutants such as ozone, PM, acetaldehyde, formaldehyde, and acrolein
can be formed secondarily through atmospheric chemical processes. Since secondarily formed
pollutants can result from many different reaction pathways, there are uncertainties associated
with each pathway. Simplifications of chemistry must be made in order to handle reactions of
thousands of chemicals in the atmosphere. Mechanisms for formation of ozone, PM,
acetaldehyde, and peroxyacetyl nitrate (PAN) are discussed in the Section 9.1.1.
144 Finlayson-Pitts BJ, Pitts JN Jr. (1986). Atmospheric Chemistry: Fundamentals and Experimental Techniques,
Wiley, New York.
66

-------
9.2 Creating Ozone and PM2.5 Fused Fields Based on Observations and Model
Surfaces
The following data were used to create the spatial fields of ozone and PM2.5 concentrations for
the 2016 "with-RFS" and 2016 "pre-RFS" scenarios:
(1)	Daily 2016 "with-RFS" and 2016 "pre-RFS" modeling-based concentrations of 24-
hour average PM2.5 component species and maximum daily 8-hour average MDA8 ozone;
(2)	Baseline, 2016 "with-RFS" "fused surfaces" of measured and modeled air quality145
representing quarterly average PM2.5 component species concentrations and ozone
concentrations for the seasonal average ozone metrics. These "fused surfaces" use the ambient
data to adjust modeled fields to match observed data at locations of monitoring sites. Details on
the methods for creating fused surfaces are provided below.
For PM2.5, daily gridded PM2.5 species were processed into annual average surfaces which
combine observed values with model predictions using the enhanced Veronoi Neighbor Average
(eVNA) method.146'147'148 These steps were performed using EPA's software package, Software
for the Modeled 8-30 Attainment Test-Community Edition (SMAT-CE)149 and have been
previously documented both in the user's guide for the predecessor software150 and in EPA's
modeling guidance document.151 As explained above, we first create a 2016 "with-RFS" eVNA
surface for each PM component species. To create the 2016 "with-RFS" eVNA surface, SMAT-
CE first calculates quarterly average values (January-March; April-June; July-September;
October-December) for each PM2.5 component species at each monitoring site with available
measured data. For this calculation we used three years of monitoring data (2015-2017).152
SMAT-CE then creates an interpolated field of the quarterly-average observed data for each
PM2.5 component species using inverse distance squared weighting resulting in a separate 3-year
average interpolated observed field for each PM2.5 species and each quarter. The interpolated
145	In this analysis, a "fused surface" represents a spatial field of concentrations of a particular pollutant that was
derived by applying the Enhanced Voronoi Neighbor Averaging with adjustment using modeled and measured air
quality data (i.e., eVNA) technique (Ding et al. 2016).
146	Gold C, Remmele P.R., Roos T., (1997). In: Algorithmic Foundation of Geographic Information Systems. In:
Lecture Notes in Computer Science, Vol. 1340 (van Kereveld M, NievergeltJ, Roos T, Widmayer P, eds) Berlin,
Germany: Springer-Verlag. Voronoi methods in GIS. pp. 21-35.
147	US EPA (2007). Technical Report on Ozone Exposure, Risk, and Impact Assessments for Vegetation. EPA
452/R-07-002. Prepared by Abt Associates Inc. for U.S. Environmental Protection Agency, Office of Air Quality
Planning and Standards, Health and Environmental Impacts Division. Research Triangle Park, NC.
(fat!ps://www3.epa. gov/ttm/maaqs/stamdards/ozone/data/2007 01 environinentaI tsd.pdf).
148	Ding, D., Zhu, Y., Jang, C., Lin, C., Wang, S., Fu, J., Gao, J., Deng, S., Xie, J., Qui, X. (2015). Evaluation of
heath benefit using BenMAP-CE with an integrated scheme of model and monitor data during Guangzhou Asian
Games. Journal of Environmental Science. 29, 178-188.
149	Software download and documentation available at https://www.epa.gov/scram/photocheniical-modeling-tools
150	Abt Associates, 2014. User's Guide: Modeled Attainment Test Software.
https://www3.epa.gov/ttn/scram/giiidance/giiide/MATS 2-6-1 manual, pdf
151	USEPA, (2018). Modeling Guidance for Demonstrating Attainment of Air Quality Goals for Ozone, PM2.5, and
Regional Haze, Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research
Triangle Park, NC. https://www3.epa.gov/ttii/scram/giiidance/giiide/03-PM-RH-Modeiing Guidance-2018.pdf
152	Three years of ambient data is used to provide a more representative picture of air pollution concentrations.
67

-------
observed fields are then adjusted to match the spatial gradients from the modeled data. These two
steps can be calculated using Equation 9.1:
eVNA a016 «wiihrRFs«= I Weight Monitorx sqjois-ion^^'2016"™^" Equation 9.1
S ^	^	ModelXtStClt2016"with-RFS"
Where:
•	eVNAg,s,q,2oi6 "with-RFS" is the gradient adjusted quarterly-average eVNA value at grid-
cell, g, for PM component species, s, during quarter, q for the year 2016;
•	Weightx is the inverse distance weight for monitor x at the location of grid-cell, g;
•	Monitorx,s,q,20i5-2017 is the 3-year (2015-2017) average of the quarterly monitored
concentration for species, s, at monitor, x, during quarter, q;
•	Models,s,q,2016 "with-RFS" is the 2016 "with-RFS" modeled quarterly-average
concentrations of species, s, at grid cell, g, during quarter, q;
•	Modelx,s,q,20i6 "with-RFS" isthe 2016 "with-RFS" modeled quarterly-average concentration
of species, s, at the location of monitor, x, during quarter q.
The 2016 "with-RFS" eVNA field serves as the starting point for fused model surfaces. As
described in above, to create a gridded 2016 "pre-RFS" eVNA surfaces for the 2016 "with-RFS"
and 2016 "pre-RFS" scenarios, we take the ratio of the modeled 2016 "pre-RFS" quarterly
average concentration to the modeled 2016 "with-RFS" concentration in each grid cell and
multiply that by the corresponding 2016 "with-RFS" eVNA quarterly PM2.5 component species
value in that grid cell, as shown in Equation 9.2.
e VNAg S q 2016 "pre-RFS" = (e VNAg s q 2oi6 "with-RFS'') x M ,, &s,q'2016 preRFS Equation 9.2
6 * r	y Model&s q 2016 "with-RFS"
This results in a gridded future-year projection which accounts for adjustments to match
observations in the 2016 modeled data.
Particulate ammonium concentrations are impacted both by emissions of precursor ammonia gas
as well as ambient concentrations of particulate sulfate and nitrate. Because of uncertainties in
ammonium speciation measurements combined with sparse ammonium measurements in rural
areas, the SMAT-CE default is to calculate ammonium values using the degree of sulfate
neutralization (i.e., the relative molar mass of ammonium to sulfate with the assumption that all
nitrate is fully neutralized). Degree of neutralization values are mainly available in urban areas
while sulfate measurements are available in both urban and rural areas. Ammonium is thus
calculated by multiplying the interpolated degree of neutralization value by the interpolated
sulfate value at each grid-cell location which allows the ammonium fields to be informed by
rural sulfate measurements in locations where no rural ammonium measurements are available.
68

-------
The degree of neutralization is not permitted to exceed the maximum theoretical molar ratio of
2:1 for ammonium:sulfate. When creating the future year surface for particulate ammonium, we
use the default SMAT-CE assumption that the degree of neutralization for the aerosol remains at
2016 levels.
A similar method for creating 2016 "pre-RFS" eVNA surfaces is followed for the MDA8 ozone
metric with a few key differences. First, while PM2.5 is split into quarterly averages and then
averaged up to an annual value, we look at ozone as a summer-season average using definitions
that match metrics from epidemiology studies (May-Sep for MDA8). The other main difference
in the SMAT-CE calculation for ozone is that the spatial interpolation of observations uses an
inverse distance weighting rather than an inverse distance squared weighting. This results in
interpolated observational fields that better replicate the more gradual spatial gradients observed
in ozone compared to PM2.5.
9.3 Air Quality Model Performance Evaluation
An operational model performance evaluation for ozone, PM2.5 and its related speciated
components, specific air toxics (i.e., formaldehyde, acetaldehyde, benzene, 1,3-butadiene, and
acrolein), as well as nitrate and sulfate deposition, was conducted using 2016 state/local
monitoring site data in order to estimate the ability of the CMAQ modeling system to replicate
the base year concentrations for the 12 km Continental United States domain (Section 6.7.2,
Figure 6.1). Included in this evaluation are statistical measures for model versus observed pairs
that were paired in space and time on a daily or weekly basis, depending on the sampling
frequency of each network (measured data). We excluded the CMAQ predictions for certain
time periods with missing ozone, PM2.5, air toxic and nitrate and sulfate deposition observations
from our calculations. It should be noted when comparing model and observed data that each
CMAQ concentration represents a grid-cell volume-averaged value, while the ambient network
measurements are made at specific locations.
Model performance statistics were calculated for several spatial scales and temporal periods.
Statistics were calculated for individual monitoring sites and for each of the nine National
Oceanic and Atmospheric Administration (NOAA) climate regions of the 12-km U.S. modeling
domain (Figure 9.1).153 The regions include the Northeast, Ohio Valley, Upper Midwest,
Southeast, South, Southwest, Northern Rockies, Northwest and West154'155 as were originally
153	NOAA, National Centers for Environmental Information scientists have identified nine climatically consistent
regions within the contiguous U.S. https://www.ncdc.noaa.gov/mon.itoring-references/maps/us-climate-regions.php
154	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 Midwest 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.
155	Note most monitoring sites in the West region are located in California, therefore statistics for the West will be
mostly representative of California ozone air quality.
69

-------
Figure 9.1 NOAA Nine Climate Regions159
identified in Karl and Koss (1984).156 The statistics for each site and climate region were
calculated by season ("Winter" is defined as average of December, January, and February;
"Spring" is defined as average of March, April, and May; "Summer" is defined as average of
June, July, and August; and "Fall" is defined as average of September, October, and November).
For 8-hour daily maximum ozone, we also calculated performance statistics by region for the
May through September ozone season i >7 In addition to the performance statistics, we prepared
several graphical presentations of model performance. These graphical presentations include
regional maps which show the mean bias, mean error, normalized mean bias and normalized
mean error calculated for each season at individual monitoring sites. The full model
performance evaluation can be found in a memo with the online supplemental materials.158
U.S. Climate Regions
156	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.
157	In calculating the ozone season statistics, we limited the data to those observed and predicted pairs with
observations that exceeded 60 ppb in order to focus on concentrations at the upper portion of the distribution of
values.
158	https://www.epa.gov/renewable-fuel-standard-program/anti-backsliding-determination-and-studv
159	Source: https://www.ncdc.noaa.gOv/monitoring-references/maps/us-climate-regions.php#references
70

-------
Monthly Air Quality Difference Maps
a
b
%
< -1.00
^ -1.00 to -0.50
-0.50 to -0.05
-0.05 to 0,05
0.05 to 0.40
0.40 to 0.80
0.80 to 1.20
Hi 1.20 to 1.60
^ > 1.60
C
p'
t ^

» 1 JL '."a
•--- • j-\\.r
11
¦Mr

v • ; . -f;

... 4 " V >
|P?1| J'
» "	t-lT v.jJ
V - - •
9 a- § ft
}

Yl
4 rfl i'a
a 1 "i #- Hi *$..
Figure 9.2 Percent difference in (a) annual average, (b) January average, and (c) July average PM2.5
concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
71

-------
a
1
5

"" 1
~mm l:
1
t'sT
MHL \ %
-• . .J: s>~
		4m
||

uj(
* c "

Sm
^ %.V>,
; ^ is®
PPb
< -0.30
0.30 to 0.20
-0.20 to -0.10
-0.10 10 -0.01
-0.01 to 0.01
0.01 to 0.10
0.10 to 0.20
0.20 to 0.30
>0.W
c
Figure 9.3 Absolute difference in (a) annual average, (b) January average, and (c) July average
NO2 concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
72

-------

a
• r ~ - -
F
r - \ '¦¦
l. 	A
h r ^
>, i -!	"<6 V
%
< 1.0
1.0 to 5.0
5.0 to 10.0
10.0 10 Ib.O
15.0 to 20.0
20.0 to 25.0
25.0 to 30.0
30.0 to 35.0
> 3b.0
Figure 9.4 Percent difference in (a) annual average, (b) January average, and (c) July average NCh
concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
73

-------
ppb
< -35.0
35.0 to 30.0
-30.0 to -25.0
-?!>.0 to -?0.0
-20.0 to -15.0
-15.0 to -10.0
-10.0 to -5.0
-5.0 to-1.0
> -1.0
Figure 9.5 Absolute difference in (a) annual average, (b) January average, and (c) July average
CO concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
74

-------
a
-¦¦i. *:
%
< -12.0
12.0 to 10.0
-10.0 to -8.0
-a.a to -fc.o
-6.Q to -4.0
-4.0 to -2.0
2,0 to -1.0
-1,0 to -0.5
> -O.1!
'I
*
A,	'¦
v jff
¦n|
"" tiPy
" V "L f
' 4
Figure 9.6 Percent difference in (a) annual average, (b) January average, and (c) July average CO
concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
75

-------
a
b
Ljgfrn3
< 0.300
-0.300 10 -0.700
0.200 to 0,100
-0.100 10 -0.010
0.010 to 0.001
-0.001 lo 0.001
0.001 to 0.010
0.010 to 0.100
0.100 to 0.200
0.200 to 0.300
> 0.300
c

% ]
j#
• - - - * -i | Jfe' V

; ¦ 	' L/

. jffr?



«'

- :





\ • ™ §|
Muuoocaa «iir crm > "*¦ V
•!.. p ^ f ~.J
Figure 9.7 Absolute difference in (a) annual average, (b) January average, and (c) July average
benzene concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
76

-------

"5:«r • -
C
¦. EJ
jf
I


~Ik- ^
r : 	,<
If '	'¦ '; , f*	v'-*.-'
V"'-
•-	:¦¦¦'¦	-:A
t W
\ t
'<.<¦....	v iS
' ,*¦< r


%
< -ho.n
50.0 to 25.0
-?5.Q Id -Ul.CI
10.0 to 5.0
-5.0 to -?.!>
2.5 to 1.0
-1.0 to 1.0
1.0 to 2.5
2.5 to 5.0
5.0 to 10.0
10.0 to 25.0
25.0 to 50.0
> 50.0
Figure 9.8 Percent difference in (a) annual average, (b) January average, and (c) July average
benzene concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
77

-------
a
f k
IK
¦¦ Hr;J
\L:
Kg
H *
H
"V< ,	If	'
j\ ¦y'.j'.A	^

i \ >-
.-*$ \ \.J
J'""'... ••-:*&
>¦¦	r-r
'¦ '4'' .
c:4
> \ ¦,
•s, I ...» .
> ' V V' •

-'"A >
f\ ¦'**: "f vv;:-
.• ."	
...
"A

if
X
K\'>
ug/rn3
< 0.300
-0.300 10 -0.?00
0.200 to 0.100
-0.100 10 -0.010
0.010 to 0.001
-0.001 La 0.001
0.001 to 0.010
0.010 to 0.100
0.100 to 0.200
0.200 to 0.300
> 0.300
ff
^ ... W J. A
¦4 '• \ Wi ••• •
	m
• -Xo

v.	-	V
MM	| \ ¦
i	\
Figure 9.9 Absolute difference in (a) annual average, (b) January average, and (c) July average
acetaldehyde concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
78

-------
a
1J>,.
.jy % '¦ ~-
%
< -50.0
50.0 to 25.0
-75.0 10 -10.0
10.0 to 5.0
-5.0 to -?.5
2.5 to 1.0
-1.0 (0 1.0
1.0 to 2.5
2,5 to 5.0
5.0 to 10.0
10.0 to 25.0
25.0 to 50.0
> 50.0
Figure 9.10 Percent difference in (a) annual average, (b) January average, and (c) July average
acetaldehyde concentrations between "pre-RFS" and "with-RFS" Scenarios for 2016
79

-------
a
b
	c
1 •.
« fj"
T" : ^ 5 >
1 'V""-
. 	; '! Xt't . -*<
^ BfgT
- " v:. W
"i ¦ ' v 1 BHkNr
	/fln?
i %ry 	-4
v*« r
¦ •, J \ ">y
m
% Li^A
t. r \
% h \
! < \.S *V |f *§§ s
ug/m3
¦¦ < 0.005
M -0.005 10 -0.004
¦ 0.004 to 0.003
-0,003 lo -0.00?
0.002 to 0.001
-0.001 lo o.uoi
0.001 to 0.002
0.002 to 0.003
0.003 to 0.004
¦¦ 0.004 to 0.005
c
rj
1
1 « *
* *i % iK°' "'.sLf
iM
¦j $
- A -/
••• ' •- y


f
•4,
W' r '

i ""i: A (
\ iT' ''Vf ; \
A /' v
y
Mn*: G.0015 M It 0.001)7


'• A'.'
Figure 9.11 Absolute difference in (a) annual average, (b) January average, and (c) July average
acrolein concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
80

-------
a


1 V JC -- :y
F /¦;;	-

- - -V .. if \
@ § Bj-

— --'v - Km ;f




X"'i l
§1

1 	
¦—if""" p
I '% p


X1.. 1^.-1
HPi m 1
mHB r
"W"
I ¦ ft, „
** \v


MwU.mi mi,. -UjmA > V

if'
>:'¦
"*-V
\ y:,
£JL«HS '•
""y"'
J r
J

-""•r
a h  50.0
ft"

K3
Figure 9.12 Percent difference in (a) annual average, (b) January average, and (c) July average
acrolein concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
81

-------
a


H §M
„ •.». iMVi % m\
/ .J V

* m 1 B
... ¦¦ /
-	—i s. ¦' y *4
, *' A
-	rP"
i
. ;
¦ '••• "-

HrH J
\f SJ
K . •*" • t -v.
b
ug/rn3
< 0.005
¦¦ -0.005 10 -0-004
0.004 to 0.003
-0.003 10 -0.00?
0.002 to 0.001
-0.001 10 0.001
0.001 to 0.002
0.002 to 0.003
0.003 to 0.004
^ 0.004 to 0.005
> 0.005
c
¦j - -¦
1 '• % J ^ |
*«*« e a « & 1 al« 1
r - - % ' i %

J	w&
V:, ^
"T\/

% | f
' - -V
i ,Ji	1,

-f B : y
"V'
",r, V'\\
-4 Af \.'A,.
, —! . v "t '-v.
Figure 9.13 Absolute difference in (a) annual average, (b) January average, and (c) July average
1,3-butadiene concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
82

-------
a
%
H < -50.0
50.0 to 25.0
¦¦ -75.0 lo -10.CI
¦ 10.0 to 5.0
-5.0 10 -?.5
2.5 to -1.0
-1.0 to 1.0
1.0 to 2.5
2.5 to 5.0
5.0 to io.o
¦¦ 10.0 to 25.0
25.0 to 5D.0
H > 50.0
Figure 9.14 Percent difference in (a) annual average, (b) January average, and (c) July average
1,3-butadiene concentrations between "pre-RFS" and "with-RFS" scenarios for 2016

83

-------
TT
km
® ¦ ¦ 'j.
.. jfl
¦T:
f
- \
\\ V
iff;*
¦. H" .M

I 11 11 w
a m
\ \
\ \ •-'>i
ug,'m3
< 0.300
-0.300 10 -0.^00
0.200 to 0.100
-0.100 lo -0.010
0.010 to 0.001
-0-001 to 0.001
0.001 to 0.010
0.010 to 0.100
0.100 to 0.200
0.200 to 0.300
> 0.300
Figure 9.15 Absolute difference in (a) annual average, (b) January average, and (c) July average
formaldehyde concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
84

-------
a
b
%
¦ < -mlo
Ml 50.0 to 25.0
¦¦ -75.0 10 -10.a
10.0 to 5.0
-5 0 to -?.5
-2.5 to -1.0
-1.0 to 1.0
1.0 to 2.5
2.5 to 5.0
5.0 to 10.0
¦¦ 10.0 to 25.0
¦¦ 25.0 to 50.0
¦¦ > 50.0
C
Figure 9.16 Percent difference in (a) annual average, (b) January average, and (c) July average
formaldehyde concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
85

-------
a
b
v"f		;
¦ f ' -
^
i" ¦ -¦-¦€>. L J>
U ¦ Y
, "• ll *¦ S' ,. JJp
?»* w- - L.s s ««•'	iff
¦i,,, H F '

J,„	
iffl, J'


a r' v ; -
ijg/m3
mm < 0.005
H -0.005 10-0.004
0.004 to 0.003
-0.003 10 -0.00?
0.002 to 0.001
-0.001 10 0.001
0.001 to 0.002
0.002 to 0.003
0.003 to 0.004
M 0.004 to 0.005
> 0.005
c
Figure 9.17 Absolute difference in (a) annual average, (b) January average, and (c) July average
naphthalene concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
86

-------
a
m	 i

1 1 v"

JCM
•;
i \.
' ~ "~3 | !? « ||M
-- -.r .



^ll
ring
% P> :

f - -v %, ?) ./T*
8



V Sp • ' it
' * !'¦


%
£ISh 9
• '• s-,„ •

;i ^ "-W'
§ ? • -L

* :m 1
'H§a|
I
£M .V

b
If

- % X
\ -.0%
k ]. •,„
M \ ¦
%
< -hO.O
50.0 to 25.0
-75.0 10-10.0
10.0 to 5.0
-5.0 10 -?.5
2.5 to 1.0
-1.0 10 1.0
1.0 to 2.5
2.5 to 5.0
5.0 to 10.0
10.0 to 25.0
25.0 to 5D.0
> 50.0
Figure 9.18 Percent difference in (a) annual average, (b) January average, and (c) July average
naphthalene concentrations between "pre-RFS" and "with-RFS" scenarios for 2016
87

-------
9.5 National Emission Inventories for Criteria and Toxic Pollutants
In addition to the detailed county-level emission inventories generated for air quality modeling
as described in Sections 6.2 and 6.3, national emission inventories were developed as a quality
check, using the pre-aggregation feature of MOVES2014b at the national level.
In this approach, a single MOVES run can be performed to estimate the emission from the entire
United States (50 states, excluding Puerto Rico and Virgin Islands). Input variables such as
temperatures, fuel properties, effects of the Inspection and Maintenance (I/M) programs, and
activity data including vehicle age distributions, speed distributions, road type distributions
across the country are pre-aggregated and approximated by a single set of national average
values. This approach is simpler but coarser compared to the approach used to develop the
detailed inventories needed for air quality modeling.
The national emission inventories in calendar year 2016 for the "with-RFS" and "pre-RFS" cases
are provided for both onroad vehicles and nonroad equipments along with their totals in Table
9-1.
The national-level emission inventory changes between the "pre-RFS" and the "with-RFS" cases
as summarized in Table 9-1 show the trends generally similar to those for the national (CONUS,
excluding California) total emissions summarized in Table 6.1 based on detailed emission
inventories for air quality modeling.
Table 9-1 Comparison of national emission inventory for onroad vehicles and nonroad equipments
	(tons per year) for the "with-RFS" and "pre-RFS" cases in calendar year 2016	
Pollutant
Sector
"with-RFS"

"pre-RFS"
diff
% diff
NOx
Total
5,268,765

5,130,752
138,014
2.7%

Gasoline - onroad
1,854,796

1,770,048
84,748
4.8%

Gasoline - nonroad
196,087

151,763
44,324
29.2%

Diesel - onroad
2,315,858

2,306,916
8,941
0.4%

Diesel - nonroad
851,516

851,516
0
0.0%







VOC
Total
3,056,755

2,969,017
87,739
3.0%

Gasoline - onroad
1,645,135

1,570,525
74,609
4.8%

Gasoline - nonroad
1,092,235

1,072,828
19,407
1.8%

Diesel - onroad
230,086

236,364
-6,278
-2.7%

Diesel - nonroad
76,799

76,799
0
0.0%







PMio
Total
381,585

384,543
-2,958
-0.8%

Gasoline - onroad
133,357

132,666
692
0.5%

Gasoline - nonroad
42,753

42,753
0
0.0%

Diesel - onroad
138,201

141,850
-3,650
-2.6%

Diesel - nonroad
65,085

65,085
0
0.0%







88

-------
pm25
Total
245,220

247,965
-2,746
-1.1%

Gasoline - onroad
46,735

46,123
612
1.3%

Gasoline - nonroad
39,333

39,333
0
0.0%

Diesel - onroad
94,094

97,452
-3,358
-3.4%

Diesel - nonroad
63,133

63,133
0
0.0%







S02
Total
29,600

28,978
622
2.1%

Gasoline - onroad
22,784

22,162
622
2.8%

Gasoline - nonroad
1,007

1,007
0
0.0%

Diesel - onroad
4,546

4,546
0
0.0%

Diesel - nonroad
1,148

1,148
0
0.0%







CO
Total
30,994,245

34,219,868
-3,225,623
-9.4%

Gasoline - onroad
18,091,224

19,008,086
-916,862
-4.8%

Gasoline - nonroad
11,311,391

13,597,070
-2,285,679
-16.8%

Diesel - onroad
895,826

918,908
-23,082
-2.5%

Diesel - nonroad
395,127

395,127
0
0.0%







Acetaldehyde
Total
35,697

27,596
8,101
29.4%

Gasoline - onroad
15,359

8,605
6,754
78.5%

Gasoline - nonroad
4,111

2,541
1,570
61.8%

Diesel - onroad
9,103

9,326
-223
-2.4%

Diesel - nonroad
6,810

6,810
0
0.0%







Acrolein
Total
4,549

4,463
87
1.9%

Gasoline - onroad
925

854
71
8.3%

Gasoline - nonroad
359

302
57
19.0%

Diesel - onroad
1,570

1,611
-42
-2.6%

Diesel - nonroad
1,653

1,653
0
0.0%







Benzene
Total
74,640

84,871
-10,231
-12.1%

Gasoline - onroad
43,186

48,352
-5,166
-10.7%

Gasoline - nonroad
26,715

31,731
-5,016
-15.8%

Diesel - onroad
1,965

2,014
-49
-2.4%

Diesel - nonroad
2,748

2,748
0
0.0%







1,3-Butadiene
Total
12,048

12,973
-925
-7.1%

Gasoline - onroad
6,768

7,594
-825
-10.9%

Gasoline - nonroad
4,597

4,678
-81
-1.7%

Diesel - onroad
536

554
-18
-3.3%

Diesel - nonroad
142

142
0
0.0%







Formaldehyde
Total
63,957

63,912
45
0.1%

Gasoline - onroad
12,537

11,781
756
6.4%

Gasoline - nonroad
7,156

7,376
-220
-3.0%
89

-------

Diesel - onroad
22,831

23,322
-491
-2.1%

Diesel - nonroad
19,115

19,115
0
0.0%







Naphthalene
Total
6,439

6,529
-90
-1.4%

Gasoline - onroad
2,208

2,223
-14
-0.6%

Gasoline - nonroad
1,682

1,701
-18
-1.1%

Diesel - onroad
2,260

2,316
-57
-2.5%

Diesel - nonroad
288

288
0
0.0%
90

-------