Air Quality Analysis for the Light-
and Medium-Duty Vehicle
Multipollutant Rule
Memo to the Docket
rnA United States
Environmental Protection
Agency
-------
Air Quality Analysis for the Light-
and Medium-Duty Vehicle
Multipollutant Rule
Memo to the Docket
Assessment and Standards Division
Office of Transportation and Air Quality
U.S. Environmental Protection Agency
United States
Environmental Protection
Agency
EPA-420-R-24-008
March 2024
-------
Table of Contents
1 Introduction/Overview 1
2 Emissions Inventory Methodology 2
2.1 Emissions Inventory Sector Summary 2
2.2 The Emissions Modeling Process 4
2.3 Emissions Inventory Methodology for 2016v3-Compatible Sectors 5
2.4 2055 Emissions Inventory Methodology for the Nonroad Sector 6
2.5 2055 Emissions Inventory Methodology for Fugitive Dust 6
3 Onroad Emissions Inventory Methodology 6
3.1 Emissions Factor Table Development 7
3.2 Activity and Other Data Development 8
3.2.1 2016 Base Year Activity data 8
3.2.2 2055 Projected Activity Data 12
3.3 Onroad Emissions Modeling 23
3.3.1 Spatial Surrogates 26
3.3.2 Temporal Profiles 26
3.3.3 Chemical Speciation 27
3.3.4 Other Ancillary Files 28
4 EGU Emissions Inventory Methodology 28
4.1 Integrated Planning Model (IPM) 29
4.2 IPM 2022 Post-IRA 29
4.2.1 AQM Reference Scenario and Incremental Demand Input Files 29
4.2.2 AQM Policy Scenario and incremental demand inputs 33
4.3 Air Quality Model-Ready EGU inventory generation 34
5 Petroleum Sector Emissions Inventory Methodology 34
5.1 Refinery Emissions 35
5.1.1 Projection of Refinery Emissions to 2050/2055 35
5.1.2 Identifying Refineries to Adjust for Air Quality Analysis 35
5.1.3 Apportioning Total Refinery Emissions to Gasoline and Diesel Fuel Production... 35
5.1.4 Total Refined Fuel and Onroad Fuel Consumed Associated with AQM Cases 37
5.1.5 Projected Change in U.S. Refinery Activity Related to Decreased Domestic
Demand 38
5.1.6 Generation of Adjustment Factors 39
5.1.7 Limitations of Modeling Impacts on Refinery Emissions 39
-------
5.2 Crude production well and pipeline emissions 42
5.2.1 Reference Case Crude Production Well Site and Pipeline Inventories for 2050/2055
42
5.2.2 Policy Scenario and Associated Crude Demand 42
5.2.3 Projected Change in U.S. Crude Production Activity Related to Decreased
Domestic Demand 43
5.2.4 Generation of Crude Production Well and Pipeline Adjustment Factors 43
5.2.5 Limitations of Modeling Impacts on Crude Production Wells and Pipeline Pumps 43
5.3 Natural gas production wells and pipeline pumps emissions 43
5.3.1 Reference Case Natural Gas Production Well Site and Pipeline Inventories for
2050/2055 43
5.3.2 Policy Scenario and Associated Natural Gas Demand 44
5.3.3 Generation of Natural Gas Production Well and Pipeline Adjustment Factors 44
5.3.4 Limitations of Modeling Impacts on Natural Gas Production Wells
and Pipeline Pumps 44
6 Inventory Summary Tables 45
7 Air Quality Modeling Methodology 47
7.1 Air Quality Model - CMAQ 47
7.2 CMAQ Domain and Configuration 47
7.3 CMAQ Inputs 49
7.4 CMAQ Model Performance Evaluation 50
7.4.1 Monitoring Networks 52
7.4.2 Model Performance Statistics 53
7.4.3 Evaluation for 8-hour Daily Maximum Ozone 55
7.4.4 Seasonal Evaluation of PM2.5 Component Species 61
7.4.5 Seasonal Hazardous Air Pollutants Performance 126
7.4.6 Seasonal Nitrate and Sulfate Deposition Performance 128
7.5 Model Simulation Scenarios 131
8 Additional Results of Air Quality Analysis 132
8.1 Annual 2055 Reference, LMDV Regulatory, and Onroad-Only Scenario Maps 132
8.2 Seasonal Air Toxics Maps 151
8.3 Projected Visibility in Mandatory Class I Federal Areas 181
-------
1 Introduction/Overview
The Environmental Protection Agency (EPA) has finalized a rule to build on and improve the
previous emission control program for on-highway light- and medium-duty engines and vehicles
by further reducing air pollution from light- and medium-duty engines across the United States.
This rulemaking is formally titled "Multi-Pollutant Emissions Standards for Model Years 2027
and Later Light-Duty and Medium-Duty Vehicles," and is more generally referred to as the
"Light Medium Duty Vehicle" (LMDV) rule. The rule impacts emissions of criteria and air toxic
pollutants as well as greenhouse gases (GHGs). This document includes information related to
the air quality modeling analysis done in support of the final rule and focuses on impacts to
ambient concentrations of criteria and air toxic pollutants.
EPA conducted an air quality modeling analysis of a regulatory scenario involving light- and
medium-duty "onroad" vehicle emission reductions and corresponding changes in "upstream"
emission sources like EGU (electric generating unit) emissions and refinery emissions. For this
analysis, emission inventories were produced, and air quality modeling was performed for three
scenarios: a 2016 base case, a 2055 reference scenario, and a 2055 LMDV regulatory or policy
case.1 Decisions about the emission scenarios and other elements used in the air quality
modeling were made early in the analytical process for the final rulemaking, and the decision
was made to model the proposed standards as the policy case. Accordingly, the air quality
analysis does not fully represent the final regulatory scenario; however, we consider the
modeling results to be a fair reflection of the impact the standards will have on air quality in
2055. The policy case assumes battery electric vehicle (BEV) penetration will reach 71 percent
for passenger cars and 66 percent for light-duty trucks in model year 2050. The policy case also
assumes a phase-in of gasoline particulate filters for gasoline vehicles beginning in model year
2027.
An air quality modeling platform consists of all the emissions inventories and ancillary data
files used for emissions modeling, as well as the meteorological, initial condition, and boundary
condition files needed to run the air quality model. An emissions modeling platform consists of
the emissions modeling data and techniques including the emission inventories, the ancillary data
files, and the approaches used to transform inventories for use in air quality modeling.
This analysis utilizes the 2016v3 emissions modeling platform,2 which includes a base year
(2016) and projection year (2023 and 2026) inventories, along with ancillary emissions data, and
scripts and software for preparing the emissions for air quality modeling. The Technical Support
Document (TSD) Preparation of Emissions Inventories for the 2016v3 North American
Emissions Modeling Platform describes how the emission inventories for each year of data
available in the platform were developed.3
Section 2 of this document gives a summary of the emissions inventory inputs to the air
quality modeling. Section 3 of this document describes the methodology for developing onroad
1 The reference case represents a scenario without the light- and medium-duty standards being analyzed. Additional
information about the use of the base case is available in Section 7.5.
2 2016v3 Emissions Modeling Platform, https://www.epa.gov/air-emissions-modeiing/2016v3-platform SMOKE
inputs available from https://gaftp.epa. gov/Ai r/em is mod/2016/v3/
3 U.S. EPA (2023) Technical Support Document: Preparation of Emissions Inventories for the 2016v3 North American
Emissions Modeling Platform, https://www.epa.gov/air-em.issions-modeiing/2016-version-3-techiiical-snpport-
docuinent.
1
-------
mobile emission inventories, Section 4 focuses on the methodology for developing electrical
generating unit (EGU) emission inventories, and Section 5 focuses on the methodology for
developing petroleum sector emission inventories. Section 6 provides emissions summary tables.
Sections 7 and 8 provide an overview of the air quality modeling methodology and supplemental
air quality modeling results.
2 Emissions Inventory Methodology
This section provides an overview of the emission inventories used in the air quality analysis
for the final rule. These inventories include point sources, nonpoint sources, onroad and nonroad
mobile sources, commercial marine vessels (CMV), locomotive and aircraft emissions, biogenic
emissions, and fires for the U.S., Canada, and Mexico. The emissions used for the 2055 policy
scenario were the same as those in the 2055 reference scenario for all emissions sectors except
for onroad mobile source emissions, EGU emissions, and petroleum sector emissions
(specifically refineries, crude oil production well sites and pipelines, and natural gas production
well sites and pipelines).
For this study, the 2016 emission inventories used were based on those for the 2016v3
platform except for the U.S. onroad and nonroad4 mobile sources. For the 2055 cases, the U.S.
onroad and nonroad mobile sources were projected to year 2055 levels, while other
anthropogenic emissions sources were retained at the 2016v3 platform projected emissions levels
for the year 2026. A high-level summary of the emission inventories used is provided in this
section, while the development of the U.S. onroad mobile source emissions is described in detail
in Section 3, the development of the EGU emissions is described in Section 4, and the
development of petroleum sector emissions is described in Section 5.
2.1 Emissions Inventory Sector Summary
For the purposes of preparing the air quality model-ready emissions, emission inventories are
split into "sectors". The significance of a sector is that each sector includes a specific group of
emission sources, and those data are run through the emissions modeling system independently
from the other sectors up to the point of the final merging process. The final merging process
combines the sector-specific low-level (of the vertical levels in the air quality model) gridded,
speciated, hourly emissions together to create CMAQ-ready emission inputs. While pertinent
atmospheric emissions related to the problem being studied are included in each modeling
platform, the splitting of inventories into specific sectors for emissions modeling varies by
platform. The sectors for the 2016v3 emissions modeling platform used in this study are shown
in Table 2-1. Descriptions for each sector are provided. For more detail on the data used to
4 The 2016 U.S. nonroad mobile source emissions inventory in the 2016v3 platform includes emissions for Texas and
California which were developed using their own tools. For this study, those state-supplied emissions were replaced
with 2016 nonroad emissions computed with an updated version of the Motor Vehicle Emission Simulator,
MOVES4.RC2.
2
-------
develop the 2016v3 inventories and on the processing of those inventories into air quality model-
ready inputs, see the 2016v3 emissions modeling platform TSD.5
Table 2-1 Inventory sectors included in the emissions modeling platform
Inventory Sector
Sector Description
Mobile - Nonroad
Mobile sources that do not drive on roads, excluding
locomotives, aircraft, and commercial marine vessels (see
Section 2.3)
Mobile - Onroad
Onroad mobile source gasoline and diesel vehicles from moving
and non-moving vehicles that drive on roads (see Section 3)
Mobile - Category 3 Commercial Marine
Vessels
Commercial marine vessels with Category 3 engines within and
outside of U.S. waters
Mobile - Category 1 and 2 Commercial
Marine Vessels
Commercial marine vessels with Category 1 and 2 engines
within and outside of U.S. waters
Mobile - Rail
U.S. Class I line haul, Class II/III line haul, passenger, and
commuter locomotives (does not include railyards and
switchers)
Nonpoint - Fertilizer
NH3 emissions from U.S. fertilizer sources
Nonpoint - Livestock
Primarily NH3 and VOC emissions from U.S. livestock sources
Nonpoint - Area Fugitive Dust
PM emissions from paved roads, unpaved roads and airstrips,
construction, agriculture production, and mining and quarrying
in the U.S.
Nonpoint - Residential Wood Combustion
U.S. residential wood burning emissions from devices such as
fireplaces, woodstoves, pellet stoves, indoor furnaces, outdoor
burning in fire pits and chimneys
Nonpoint - Oil and Gas
Oil and gas exploration and production, both onshore and
offshore
Nonpoint - Solvents
Nonpoint VOC emissions from solvents such as cleaners,
personal care products, and adhesives.
Nonpoint - Other
All nonpoint emissions in the U.S. not included in other sectors,
including industrial processes, waste disposal, storage and
transport of chemicals and petroleum, waste disposal,
commercial cooking, and miscellaneous area sources
Point - Airports
Aircraft engines and ground support equipment at U.S. airports
Point - Electrical Generating Units
Electric generating units that provide power to the U.S. electric
grid
Point - Oil and Gas
Point sources related to the extraction and distribution of oil and
gas in the U.S.
Point - Other
All point sources in the U.S. not included in other sectors.
Includes rail yards and refineries.
Point - Fires - Agricultural
Fires due to agricultural burning in the U.S.
Point - Fires - Wild and Prescribed
Wildfires and prescribed burns in the U.S.
Point - Non-U.S. Fires
Fires within the domain but outside of the U.S.
Biogenic (beis)
Emissions from trees, shrubs, grasses, and soils within and
outside of the U.S.
Canada - Mobile - Onroad
Onroad mobile sources in Canada (see Section 2.5)
Mexico - Mobile - Onroad
Onroad mobile sources in Mexico (see Section 2.5)
5 U.S. EPA (2023) Technical Support Document: Preparation of Emissions Inventories for the 2016v3 North American
Emissions Modeling Platform, https://www.epa.gov/air-emissions-modeling/2016-version-3-technical-siipport-
document.
3
-------
Inventory Sector
Sector Description
Canada/Mexico - Point
Canadian and Mexican point sources
Canada/Mexico - Nonpoint and Nonroad
Canadian and Mexican nonroad sources and nonpoint sources
not included in other sectors
Canada - Agricultural Point
Canadian agricultural ammonia sources
Canada - oil and gas 2D
Canadian low-level point oil and gas sources
Canada - Nonpoint - Area Fugitive Dust
Area source fugitive dust sources in Canada
Canada - Point - Point Fugitive Dust
Point source fugitive dust sources in Canada
2.2 The Emissions Modeling Process
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 emissions inventories for the sectors described above. The process of
emissions modeling transforms the emissions inventories from their original temporal, pollutant,
and spatial resolution into the hourly, speciated, gridded resolution required by the air quality
model. Emissions modeling includes the chemical speciation, temporal allocation, and spatial
allocation of emissions along with final formatting of the data that will be input to the air quality
model.
Chemical speciation creates the "model species" needed by CMAQ, for a specific chemical
mechanism, from the "inventory pollutants" of the input emission inventories. 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 this platform is the CB6 mechanism.6
This platform generates the PM2.5 model species associated with the CMAQ Aerosol Module
version 7 (AE7). See Section 3.2 of the 2016v3 platform TSD for more information about
chemical speciation in the 2016v3 platform.
Temporal allocation is the process of distributing aggregated emissions to a finer temporal
resolution, for example converting annual emissions to hourly emissions as is required by
CMAQ. While the total annual, monthly, or daily emissions are important, the hourly 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 2016v3
platform TSD for more information about temporal allocation of emissions in the 2016v3
platform.
Spatial allocation is the process of distributing aggregated emissions to a finer spatial
resolution, as is required by CMAQ. Over 60 spatial surrogates are used to spatially allocate U.S.
county-level emissions to thel2-km grid cells used by the air quality model. See Section 3.4 of
the 2016v3 platform TSD for a description of the spatial surrogates used for allocating county-
level emissions in the 2016v3 platform.
6 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.
4
-------
The primary tool used to perform the emissions modeling to create the air quality model-
ready emissions was the SMOKE modeling system, version 4.9 (SMOKE 4.9).7 When preparing
emissions for CMAQ, emissions for each sector are processed separately through SMOKE. The
elevated point source emissions are passed to CMAQ directly so the model can perform plume
rise based on hourly meteorological conditions, while the low-level emissions are combined to
create model-ready 2-D gridded emissions. Gridded emissions files were created for a 36-km
national grid named 36US3 and for a 12-km national grid named 12US2, both of which include
the contiguous states and parts of Canada and Mexico as shown in Figure 2-1. This figure also
shows the region covered by other grids that are relevant to the development of emissions for this
2.3 Emissions Inventory Methodology for 2016v3-Compatible Sectors
Except for the onroad mobile source emissions, the emissions used for the 2016 air quality
case are consistent with those developed through the 2016v3 Platform. For the 2055 cases, the
following were made to be consistent with the 2026 emissions developed by the Inventory
Collaborative (described in the 2016v3 Emissions Modeling Platform TSD): emissions for
sectors other than onroad and nonroad mobile sources in the U.S. and emissions for the onroad
mobile source sector in Canada and Mexico. Development of the 2055 nonroad emissions is
described in Section 2.4. The development of the U.S. onroad mobile source emissions for each
7 http://www.smoke-model.org/
5
-------
case is described below in Section 3. Additionally, the 2016v3 inventories, which have improved
state and county apportionment as compared to 2016v2, were used for CMV. For the point (non-
EGU) sector, 2016v3 was used. Another update that was made for this modeling was to use the
Biogenic Emissions Inventory System (BEIS) version 4 coupled with the Biogenic Emissions
Landuse Dataset version 6 within CMAQ, which was run using inline biogenics.
2.4 2055 Emissions Inventory Methodology for the Nonroad Sector
To prepare the nonroad mobile source emissions, an updated version of the Motor Vehicle
Emission Simulator (MOVES), MOVES4.RC2, was run using inputs compatible with the
2016v3 platform for all states. The nonroad component of MOVES was configured to create a
national nonroad inventory for 2055. The 2055 MOVES nonroad inventory was used in all
states.
2.5 2055 Emissions Inventory Methodology for Fugitive Dust
The inventory for road dust is generated using vehicle miles traveled (VMT)8, and the total
projected VMT in 2055 did not change between the reference and LMDV regulatory scenario
(only the fraction of EVs changed). Road dust inventories for 2055 were projected using 2055
VMT (see Section 3.2.2) and are presented in Table 6-4.
3 Onroad Emissions Inventory Methodology
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 emission factors used to develop
emissions for the reference and policy scenarios differed, the approach and all other data sources
used to calculate emissions for both scenarios were identical.
Onroad mobile source emissions result from motorized vehicles operating on public
roadways. These include passenger cars, motorcycles, minivans, sport-utility vehicles, light-duty
trucks, heavy-duty trucks, and buses. The sources are further divided by the fuel they use,
including diesel, gasoline, E-85, electricity, and compressed natural gas (CNG). The sector
accounts for emissions from parked vehicle processes (e.g., starts, hot soak, and extended idle)
and on-network processes (i.e., from vehicles as they move along the roads). The onroad
emissions are generated using Sparse Matrix Operator Kernel Emissions (SMOKE) programs
that leverage MOVES-generated emission factors with county, fuel type, source type, and road
type-specific activity data, along with hourly meteorological data.
The MOVES-generated onroad emission factors were combined with activity data (e.g.,
VMT, vehicle populations) to produce emissions within the 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 spatially allocate emissions to the grid used for air quality modeling.
8 See Section 4.2.3.1 of the 2016v2 TSD for more detail on how fugitive dust is projected.
6
-------
More details on the generation of the emission factors, activity data, and on the modeling of
the emissions are in the following subsections. National onroad emission summaries for key
pollutants are provided in Section 4.
3.1 Emissions Factor Table Development
Onroad mobile source emission factors were generated for the modeled cases by running
versions of MOVES49 (MOVES4.RC2, MOVES4.R1 and MOVES4.R2). The MOVES4
versions used for air quality modeling incorporated updated information not available for the
MOVES4 release. MOVES4.R2 also included policy case-specific inputs, including higher EV
fractions, reduced energy consumption, and reductions in HC, NOx and PM emission rates to
reflect rule requirements. Detailed information on the model updates is available in a memo to
the docket.10
The LMDV reference and regulatory cases include assumptions about light-, medium-, and
heavy-duty EV sales. The reference case EV fractions are based on modeling of light-duty
electric vehicle costs and consumer preferences while the heavy-duty fractions account for our
understanding of state adoption of California's Advanced Clean Trucks rule. For the policy
case, the heavy-duty EV fractions remained the same, but the light- and medium-duty EV
fractions were updated for consistency with EV sales fractions generated by the OMEGA model
for the NPRM action case. For air quality modeling, the case-specific BEV fractions were
incorporated into each county's fuel mix described in Section 3.2.2.5 below.
The emission factor tables input to SMOKE-MOVES are generated by running MOVES.
These tables differentiate emissions by process (i.e., running, start, vapor venting, etc.), fuel type,
vehicle type, road type, temperature, speed bin for rate per distance processes, hour of day, and
day of week. To generate the MOVES emission factors across the U.S., MOVES was run 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. The representative counties for
which emission factors are generated are selected according to their state, elevation, fuels used in
the region, vehicle age distribution, and inspection and maintenance programs. Every county in
the country is mapped to a representative county based on its similarity to the representative
county with respect to those attributes. The representative counties selected for the 2016v3
platform were retained for this analysis. More details on the methodology behind choosing
representative counties is available in the 2016v3 TSD.
Emission factors were generated by running MOVES for each representative county for two
"fuel months" - January to represent winter months and July to represent summer months -
because in some parts of the country different types of fuels are used in each season. MOVES
was run for the range of temperatures that occur in each representative county for each season.
The calculations of the temperature ranges needed for each fuel month were based on
meteorology for every county and grid cell in the continental U.S. for each hour of the year. The
SMOKE interface accounts for the sensitivity of the on-road emissions to temperature and
9 USEPA (2023) Motor Vehicle Emission Simulator: MOVES4. Office of Transportation and Air Quality. US
Environmental Protection Agency. Ann Arbor, MI. August 2023. https://www.epa.gov/moves.
10 Mo (2024). Revisions to MOVES for Air Quality Modeling to support the FRMfor the Multi-Pollutant Emissions
Standards for Model Years 2027 and Later Light-Duty and Medium-Duty Vehicles. Memorandum to Docket EPA-
HQ-OAR-2022-0829. February, 2024
7
-------
humidity by using the gridded hourly temperature information available from the meteorological
model outputs used for air quality modeling.
Appropriate versions of MOVES were run using the above approach to create emission
factors for each of the three modeling cases: 2016 base year, 2055 reference, and a 2055
regulatory case. A new set of emission factor tables were developed for this study using the same
representative counties as were used the 2016v3 platform. The county databases (CDBs) input to
MOVES for 2016 were equivalent to those used for the 2016v3 platform. To prepare the 2055
CDBs used to generate year 2055 emissions factors, the vehicle age distributions were projected
to reflect the year 2055 as were the tables representing the inspection and maintenance programs.
The fuels used were also representative of year 2055. The CDBs for each of the 2055 modeling
cases incorporated the case-specific fuel mix as detailed in Section 3.2.2.5 below.
3.2 Activity and Other Data Development
To compute onroad mobile source emissions, SMOKE selects the appropriate MOVES
emission rates for each county, hourly temperature, speed bin, and source classification code
(SCC) (which includes the fuel type, source type and road type), then multiplies the emission
rate by the appropriate activity data such as VMT (vehicle miles travelled), VPOP (vehicle
population), SPEED/SPDIST (speed distributions and averages), HOTELING (hours of extended
idle), ONI (hours of off-network idling), or STARTS (engine starts), to produce emissions. For
each of these activity datasets, first a national dataset was developed; this national dataset is
called the "EPA default" dataset. Data submitted by state agencies were incorporated into the
activity datasets used for the study where they were available and passed quality assurance
checks.
The activity data for the 2016 base year were consistent with the activity data used in the
2016v3 platform. Additional details on the development of activity data are available in the
2016v3 platform TSD.
In addition to activity data, this section also describes inputs for fuel parameters and county-
specific vehicle inspection and maintenance programs.
3.2.1 2016 Base Year Activity data
3.2.1.1 Vehicle Miles Traveled (VMT)
EPA calculated default 2016 VMT by backcasting the 2017 NEI VMT to 2016. The 2017 NEI
Technical Support Document11 contains details on the development of the 2017 VMT. The data
backcast to 2016 were used for states that did not submit 2016 VMT data. The factors to adjust
VMT from 2017 to 2016 were based on VMT data from the Federal Highway Administration
(FHWA) county-level VM-2 reports. For most states, EPA calculated county-road type factors
based on FHWA VM-2 County data for 2017 and 2016. Separate factors were calculated by
vehicle type for each MOVES road type. Some states have a very different distribution of urban
activity versus rural activity between 2017 NEI and the FHWA data, due to inconsistencies in the
definition of urban versus rural. For those states, a single county-wide projection factor based on
total FHWA VMT across all road types was applied to all VMT, independent of road type.
11 U.S. EPA (2021) 2017 National Emissions Inventory: January 2021 Updated Release, Technical Support Document.
https://www.epa.gov/air-emissions-inventories/2017-national-emissions-inventory-nei-technical-support-document-
tsd
8
-------
County-total-based (instead of county+road type) factors were used for all counties in IN, MS,
MO, NM, TN, TX, and UT because many counties had large increases in one particular road
type and decreases in another road type.
For the 2016v3 platform, VMT data submitted by state and local agencies were incorporated
and used in place of EPA defaults. Note that VMT data need to be provided to SMOKE for each
county and SCC. The onroad SCCs characterize vehicles by MOVES fuel type, vehicle (aka
source) type, emissions process, and road type. Any VMT provided at a different resolution than
this were converted to a full county-SCC resolution to prepare the data for processing by
SMOKE.
A final step was performed on all state-submitted VMT. The distinction between a "passenger
car" (MOVES source type 21) versus a "passenger truck" (MOVES source type 31) versus a
"light commercial truck" (MOVES source type 32) is not always consistent between different
datasets. This distinction can have a noticeable effect on the resulting emissions, since MOVES
emission factors for passenger cars are quite different than those for passenger trucks and light
commercial trucks. To ensure consistency in the 21/31/32 splits across the country, all state-
submitted VMT for MOVES vehicle types 21, 31, and 32 (all of which are part of HPMS vehicle
type 25) was summed, and then re-split using the 21/31/32 splits from the EPA 2016v2 default
VMT. VMT for each source type as a percentage of total 21/31/32 VMT was calculated by
county from the EPA default VMT. Then, state-submitted VMT for 21/31/32 was summed and
re-split according to those percentages.
For 2016v3, total 2016 VMT is unchanged from 2016v2. However, road type distributions
were updated to be consistent with those in 2020 NEI12 in Florida, Illinois, Minnesota, Missouri,
South Carolina, and West Virginia to correct anomalies found in the 2016vl and 2016v2 data.
3.2.1.2 Vehicle Population (VPOP)
The EPA default VPOP dataset was based on the EPA default VMT dataset described above.
In the areas where EPA backcasted 2017 NEI VMT:
2016v3 VPOP = 2016v3 VMT * (VPOP/VMT ratio by county-SCC6).
Where the ratio by county-SCC is based on 2017 NEI with MOVES3 fuel splits and SCC6
means the first six digits of the SCC code that include fuel type and source type but exclude the
road type and process. In the areas where we used 2016vl VMT resplit to MOVES3 fuels,
2016v3 VPOP = 2016v2 VPOP = 2016vl VPOP with two resplits: first, source types 21/31/32
were resplit according to 2017 NEI EPA default 21/31/32 splits so that the whole country has
consistent 21/31/32 splits. Next, fuels were resplit to MOVES3 fuels. There are some areas
where 2016 VMT was submitted but 2016 VPOP was not; those areas are using 2016vl VPOP
(with resplits). The same method was applied to the 2016 EPA default VMT to produce an EPA
default VPOP dataset.
3.2.1.3 Speed Activity (SPEED/SPDIST)
SMOKE-MOVES uses speed distributions similarly to how they are used when running
MOVES in inventory mode. The speed distribution file, called SPDIST, specifies the amount of
12 U.S. EPA (2023) 2020 National Emissions Inventory, Technical Support Document, https://www.epa.gov/air-
emissions-inventories/2020-national-emissions-inventory-nei-technical-support-document-tsd
9
-------
time spent in each MOVES speed bin for each county, vehicle (aka source) type, road type,
weekday/weekend, and hour of day. This file contains the same information at the same
resolution as the Speed Distribution table used by MOVES but is reformatted for SMOKE.
Using the SPDIST file results in a SMOKE emissions calculation that is more consistent with
MOVES than the previous hourly speed profile (SPDPRO) approach, because emission factors
from all speed bins can be used, rather than interpolating between the two bins surrounding the
single average speed value for each hour as is done with the SPDPRO approach.
The SPEED inventory that includes a single overall average speed for each county, SCC, and
month, was also read in by SMOKE. SMOKE requires the SPEED dataset to exist even when
speed distribution data are available, even though only the speed distribution data affects the
selection of emission factors. The SPEED and SPDIST datasets are from the 2017 NEI and are
based on a combination of the Coordinating Research Council (CRC) A-100 data and 2017 NEI
MOVES CDBs.
3.2.1.4 Hoteling Hours (HOTELING)
Hoteling hours activity is used to calculate emissions from extended idling and auxiliary
power units (APUs) for heavy duty diesel vehicles. For the 2016v3 platform, hoteling hours were
computed using a factor calculated by EPA's Office of Transportation and Air Quality based on
recent studies.
The method used in 2016v3 is the following:
1 Start with 2016 VMT for combination long haul trucks (i.e., MOVES source type 62)
on restricted roads, by county. Only VMT on urban and rural restricted highways for
MOVES source type 62 is included in the hoteling calculation.
2 Multiply the VMT by 0.007248 hours/mile.13
3 Apply parking space reductions to keep hoteling within the estimated maximum
hours by county, except for states that requested EPA do not do that (CO, ME, NJ,
NY).
Hoteling hours were adjusted down in counties for which there were more hoteling hours
assigned to the county than could be supported by the known parking spaces. To compute the
adjustment, the hoteling hours for the county were computed using the above method, and
reductions were applied directly to the 2016 hoteling hours based on known parking space
availability so that there were not more hours assigned to the county than the available parking
spaces could support if they were full every hour of every day.
A dataset of truck stop parking space availability with the total number of parking spaces per
county was used in the computation of the adjustment factors.14 This same dataset is used to
develop the spatial surrogate for hoteling emissions. Since there are 8,784 hours in the year
2016; the maximum number of possible hoteling hours in a particular county is equal to 8,784 *
13 USEPA (2023). Population and Activity of Onroad Vehicles in MOVES4. EPA-420-R-23-005. Office of
Transportation and Air Quality. US Environmental Protection Agency. Ann Arbor, MI. August 2023.
https://www.epa.gov/moves/moves-technical-reports.
14 From 2016 version 1 hoteling workbook.xlsx developed based on the input dataset for the hoteling spatial surrogate
in the 2016vl platform.
10
-------
the number of parking spaces in that county. Hoteling hours for each county were capped at that
theoretical maximum value for 2016 in that county unless the number of parking spaces listed
was less than 12, in which case the hours were not reduced.
For 2016v3, hoteling was calculated as:
2016v3 HOTELING = 2017NEI HOTELING * 2016v3 VMT/2017NEI VMT
This is effectively consistent with applying the 0.007248 factor directly to the 2016v3 VMT.
Then, for counties that provided 2017 hoteling but did not have vehicle type 62 restricted VMT
in 2016 - that is, counties that should have hoteling, but do not have any VMT from which to
calculate it - EPA backcast 2017 hoteling to 2016 using the FHWA-based county total 2017 to
2016 trend. Finally, the annual parking-space-based caps for hoteling hours were applied. The
same caps were used as for 2017 NEI, except recalculated for a leap year (multiplied by
366/365).
For 2016v3, road type distributions and/or hoteling were adjusted in states where there was
hoteling in every county in the state: FL, IL, MN, MO, SC, and WV. 2016v2 VMT in those six
states was redistributed by road type based on 2020 NEI road type distributions (by
county/vehicle, with county/HPMS filling in where a county/vehicle isn't available in 2020
NEI), and then hoteling was recalculated based on the new VMT in those six states using the
standard VMT/HOTELING factor and parking space adjustments. Notably, this resulted in an
overall increase in hoteling in Missouri, although hoteling is now in fewer counties).
3.2.1.5 Off-Network Idling Hours (ONI)
After creating VMT inputs for SMOKE-MOVES, off-network idle (ONI) activity data were
also needed. ONI is defined in MOVES as time during which a vehicle engine is running idle
and the vehicle is somewhere other than on the road, such as in a parking lot, a driveway, or at
the side of the road. This engine activity contributes to total mobile source emissions but does
not take place on the road network.
Examples of ONI activity include:
light duty passenger vehicles idling while waiting to pick up children at school or to
pick up passengers at the airport or train station,
single unit and combination trucks idling while loading or unloading cargo or making
deliveries, and
vehicles idling at drive-through restaurants.
Note that ONI does not include idling that occurs on the road, such as idling at traffic signals,
stop signs, and in trafficthese emissions are included as part of the running and crankcase
running exhaust processes on the other road types. ONI also does not include long-duration
idling by long-haul combination trucks (hoteling/extended idle), as that type of long duration
idling is accounted for in other MOVES processes.
ONI activity hours were calculated based on VMT. For each representative county, the ratio
of ONI hours to onroad VMT (on all road types) was calculated using the MOVES ONI Tool by
source type, fuel type, and month. These ratios were then multiplied by each county's total VMT
(aggregated by source type, fuel type, and month) to get hours of ONI activity.
11
-------
3.2.1.6 Engine Starts (STARTS)
Onroad "start" emissions are the instantaneous exhaust emissions that occur at the engine start
(e.g., due to the fuel rich conditions in the cylinder to initiate combustion) as well as the
additional running exhaust emissions that occur because the engine and emission control systems
have not yet stabilized at the running operating temperature. Operationally, start emissions are
defined as the difference in emissions between an exhaust emissions test with an ambient
temperature start and the same test with the engine and emission control systems already at
operating temperature. As such, the units for start emission rates are instantaneous grams/start.
MOVES uses vehicle population information to sort the vehicle population into source bins
defined by vehicle source type, fuel type (gas, diesel, etc.), regulatory class, model year and age.
The model uses default data from instrumented vehicles (or user-provided values) to estimate the
number of starts for each source bin and to allocate them among eight operating mode bins
defined by the amount of time parked ("soak time") prior to the start. Thus, MOVES accounts
for different amounts of cooling of the engine and emission control systems. Each source bin and
operating mode has an associated g/start emission rate. Start emissions are also adjusted to
account for fuel characteristics, LD inspection and maintenance programs, and ambient
temperatures.
2016v3 STARTS = 2016v3 VMT * (2017 STARTS/ 2017 VMT by county&SCC6)
For 2016v3, Georgia Environmental Protection Division provided new weekday activity for
starts per day for 20 counties. These new starts were used for the weekdays for those 20 counties,
while MOVES default starts/day were used for weekend days. Since annual activity data are
required by the FF10 activity file format, the number of starts/day was multiplied by the number
of weekdays and weekends in the year to calculate the annual total starts for the 20 counties by
county and source type. The starts for light duty vehicle source types 21,31, and 32 were
summed and then re-split between the 21, 31, and 32 source types based on splits from EPA
default activity data, so that 21/31/32 splits are from a consistent data source nationwide. Since
George only provided their activity data by county and vehicle type, the 2016v2 splits were used
as the basis for distribution of the starts to fuel type and month.
3.2.1.7 Fuels
The 2016 scenario used MOVES4.RC2 default fuels. These fuels are the same as the fuels in
MOVES4.0.0.15
3.2.2 2055 Projected Activity Data
The projected 2055 activity data are primarily based on the 2016v3 platform's projected 2026
data, updated to be consistent with the default data and algorithms in MOVES4.R1, as well as to
estimate geographic differences in fuel and age distributions. To accomplish this analysis, the
following steps were taken:
15 U.S. EPA (2023) Fuel Supply Defaults: Regional Fuels and the Fuel Wizard inMOVES4. Office of Transportation
and Air Quality. US Environmental Protection Agency. Ann Arbor, MI. August 2023. EPA-420-R-23-025
12
-------
1. Calendar year 2055 CDBs were developed for each representative county, as
described in more detail later in this section. Each scenario (the reference case and the
policy case) had its own set of CDBs.
2. MOVES was run with each CDB to calculate detailed activity data for each
representative county. MOVES4.R1 was used for the 2055 reference case scenario and
MOVES4.R2 was used for the 2055 policy case scenario.
3. The MOVES activity results for each representative county were allocated to the
individual counties represented by each representative county using the 2016v3
platform allocations.
The following sections describe how the 2055 CDBs were developed to calculate the 2055
projected activity data.
3.2.2.1 Data Used As-is from the 2016v3 Platform
The starting point for developing the 2055 CDBs was the 2016v3 platform for calendar year
2026. The following data were used as-is from the 2016v3 platform data in the 2055 CDBs:
Geography tables: State, County, Zone, and ZoneMonthHour
VMT distribution tables: MonthFraction, DayFraction, and HourFraction
Speed distribution table: AvgSpeedDistribution
Road distribution tables: RoadTypeDistribution and ZoneRoadType
Retrofit table: OnroadRetrofit
3.2.2.2 Default Data Used As-is from MO VES
National default data and algorithms in MOVES4.R1 and MOVES4.R2 were used for the
following tables:
Some (but not all) fuels tables: FuelFormulation, FuelSupply, and FuelUsageFraction
Starts tables: StartsPerDayPerVehicle, StartsMonthAdjust, StartsHourFraction,
Starts Age Adjustment, and StartsOpModeDistribution
Hotelling tables: HotellingHoursPerDay, HotellingMonthAdjust,
HotellingHourFraction, HotellingAgeFraction, and HotellingActivityDistribution
Off-Network Idle tables: TotalldleFraction, IdleModelYearGrouping,
IdleMonthAdjust, and IdleDayAdjust
I/M table: IMCoverage
13
-------
Note that in MOVES4.R1 and MOVES4.R2, starts, hotelling, and off-network idle tables are
optional tables, and therefore can be empty in a CDB if the intention is to use default data.
Therefore, these tables are empty in the 2055 CDBs. However, the fuels tables and I/M table are
required inputs. Since the default database contains county (or region) specific data, the 2055
CDBs contain the relevant subset of the default database's data. See the MOVES4 technical
reports 16'17'18'19'20'21for more information about how these default data were derived.
3.2.2.3 Default Data from MOVES4.R1 Allocated Using 2016v3 Platform
National default data in MOVES4.R1 were allocated to representative counties for the
following tables:
VMT table: HPMSVTypeYear
VPOP table: SourceTypeYear
VMT fractions by HPMSVTypelD and county were calculated from the 2026 VMT
projections in the 2016v3 platform and used to allocate the national default VMT projections for
2055 to the county level. Similarly, VMT fractions by sourceTypelD and county were calculated
from the 2016v3 platform to allocate the national default VPOP projections for 2055. See the
MOVES4 technical report for more information about how the national default data were
derived.20
3.2.2.4 2055 Age Distributions
Each CDB has a sourceTypeAgeDistribution table. The 2055 age distributions by
representing county were primarily derived using July 1, 2020 vehicle registration data
purchased from IHS Markit-Polk, vehicle stock and sales projections from the Annual Energy
Outlook (AEO) 202322, and vehicle scrappage rates presented in the Transportation Energy Data
Book (TEDB).23 The age distributions were calculated using a modified version of the age
16 U.S. EPA (2023) Exhaust Emission Rates for Light Duty Onroad Vehicles in MOVES4. Office of Transportation
and Air Quality. US Environmental Protection Agency. Ann Arbor, MI. August 2023. EPA-420-R-23-028
17 U.S. EPA (2023) Exhaust Emission Rates for Heavy Duty Onroad Vehicles in MOVES4. Office of Transportation
and Air Quality. US Environmental Protection Agency. Ann Arbor, MI. August 2023. EPA-420-R-23-027
18 U. S. EPA (2023) Emission Adjustments for Onroad Vehicles in MOVES4. Office of Transportation and Air Quality.
US Environmental Protection Agency. Ann Arbor, MI. August 2023. EPA-420-R-23-021
19 U.S. EPA (2023) Evaporative Emissions from Onroad Vehicles in MOVES4. Office of Transportation and Air
Quality. US Environmental Protection Agency. Ann Arbor, MI. August 2023. EPA-420-R-23-023
20 U.S. EPA (2023) Population and Activity of Onroad Vehicles in MOVES4. Office of Transportation and Air
Quality. US Environmental Protection Agency. Ann Arbor, MI. August 2023. EPA-420-R-23-005
21 U.S. EPA (2023) Greenhouse Gas and Energy Consumption Rates for Onroad Vehicles in MOVES4. Office of
Transportation and Air Quality. US Environmental Protection Agency. Ann Arbor, MI. August 2023. EPA-420-R-23-
026
22 US Energy Information Administration (EIA), Annual Energy Outlook 2023, Supplemental Tables 38, 39, 44, 45
and 49, Washington, DC: March 2023.
23 Davis, S. and R Boundy (2022), Transportation Energy Data Book, Ed. 40, Oak Ridge National Laboratory,
ORNL/TM-2022/2376, https://tedb.onil.gov/wp-content/uploads/2022/03/TEDB_Ed_40.pdf
14
-------
distribution projection algorithm described in Appendix C of the Population and Activity of
Onroad Vehicles in MOVES4 technical report.20 The algorithm was modified to maintain
differences between counties, such that counties that had newer-than-average fleets in 2020
continue to have newer-than-average fleets in 2055 and, similarly, counties with older fleets now
have older fleets in the future. The fundamental approach to solving this problem was to define
how age distributions in a local area are different from the national average, and then apply that
difference to future years.
The following algorithm was implemented for calculating a representative county's base age
distribution:
1. Subset the 2020 registration data to get vehicle counts by source type and model year
for all counties represented by the representative county.
2. Group all model years 1990 and older together, because MOVES groups all vehicles
ages 30 and older together.
3. Calculate age fractions by source type.
4. Replace age 0 (model year 2020) fractions with the ratio of vehicle sales to stock from
AEO. This is because the July 1 registration data pull represents an incomplete year.
5. Renormalize the age distributions, retaining the age 0 fractions.
The following equations were used to project a representative county's base age distribution
one year into the future:
Population distribution for the next calendar year = Population distribution for the
current calendar year, minus vehicles scrapped in the current calendar year, plus
locally adjusted new vehicle sales in the next year
Vehicles scrapped in the current calendar year = Scrappage factor times the base
scrappage rate times the population distribution for the current calendar year
Scrappage factor = (Total number of vehicles in the current year, minus total number
of vehicles in the next year, plus locally adjusted new vehicle sales in the next year)
divided by the sum of the base scrappage rate times the current year's population
distribution. The purpose of the scrappage factor is to scale the base scrappage rate to
balance the equation accounting for the total number of vehicles in each calendar year.
For example, if the total number of vehicles remains constant from one year to the
next and vehicle sales are high, then the scrappage factor would be high as well, as
more vehicles would be scrapped to balance out the higher sales while maintaining
constant number of total vehicles.
The population distribution of the current calendar year is known; thus, the algorithm starts
with the base age distribution and then the algorithm iterates, so the output of the algorithm is the
input for the next year. The total number of vehicles in the next year and the vehicle sales in the
next year are also known, based on AEO. The base scrappage curve is also known, based on data
presented in TEDB.
15
-------
The differences between local areas were accounted for by applying a local sales scaling
factor to the number of new vehicles sold in the next year in the equations above. This scaling
factor was defined as the difference between the local and the national population fractions
summed over an age range [1, j], divided by the national population fraction over the same age
range. Essentially, this is using the fraction of newer vehicles in a local fleet compared to the
national average as a surrogate for what future sales in a local area might be.
The precise age range [1, j] used was determined for each source type, chosen so that the
difference between the local average age and the projected national average age in 2055 was as
close as possible to the difference between the local and national average ages in 2020. That is,
the chosen age ranges tried to maintain the same delta in average age between the local and the
national case in the future. The chosen age ranges by source type were:
Motorcycles: [1, 7]
Passenger cars: [1, 10]
Passenger trucks: [1, 4]
Light commercial trucks: [1,4]
Other buses: [1, 10]
Transit buses: [1, 10]
School buses: [1,8]
Refuse trucks: [1, 9]
Single unit trucks: [1,7]
Motor homes: [1, 9]
Combination short-haul trucks: [1, 9]
Note that for some counties, some source types were not present in the IHS Markit-Polk data.
In these rare cases, the national default age distributions were assumed. Additionally,
combination long-haul trucks were assumed to have the same age distribution nationally.
The algorithm and data described above were used to calculate SourceTypeAgeDistribution
tables for each representing county in 2055. The same age distributions were used for all
scenarios. The following figures show the resulting projected average age in 2055 by county for
the light-duty source types.
16
-------
Years
I 17.5
1,0
12.5
10.0
7.5
Figure 3-1 Projected average age of passenger cars in 2055
-------
Figure 3-3 Projected average age of light commercial trucks in 2055
3.2.2.5 2055 Fuel Mix
The mix of the fuel types used in vehi cles (or "fuel distributions") for 2055 rely on national
projections, which vary by scenario. The national projected fuel distributions for ICE vehicles in
the reference case rely on July % 2020, vehicle registration data purchased from IHS Markit-
Polk, vehicle sales projections from AEO 2023,22 EPA's Revised 2023 and Later Model Year
Light-Duty Vehicle Greenhouse Gas Emissions Standards,24 and CARB's Advanced Clean
Trucks regulation. More information about the national projected fuel distributions for the
reference case can be found in the MOVES4 technical report. Electric vehicle fractions in the
2055 reference case were based on EVI-X reference case estimates as explained in the RIA20
Fuel distributions for the regulatory case assume a shift to more electric vehicles. We assume
BEV penetration will reach 71 percent for passenger cars and 66 percent for light-duty trucks in
model year 2050. Additional details are available in the RIA.
To maintain consistency with the scenario being modelled, we used a different approach from
that used in the NPRM to project representative county fuel type distributions. For the FRM, the
starting data for representative county fuel type distributions were the results of a geospatial
allocation analysis from EVI-X: EV stock by calendar year, vehicle type, and county, which
were in turn based on national EV stock from OMEGA. For more information on the EVI-X
analysis, see Chapter 5.1 of the RIA.
24 U.S. EPA (2021). Revised 2023 and Later Model Year Light-Duty Vehicle Greenhouse Gas Emissions Standards
(86 FR 74434, December 30, 2021)
18
-------
Since MOVES' fuel type distributions inputs are by model year (not calendar year), we
calculated the annual fraction of EV sales to EV stock from OMEGA's national projections, and
then applied this fraction to EVI-X's EV stock by county to get annual EV sales by vehicle type
and county. This assumes the fraction of new EVs compared to existing EVs is relatively
constant throughout the country.
Then, because the air quality modeling is performed at the representative county level and not
for all individual counties, the EV sales by vehicle type and county were aggregated to the
representative county level.
To assure these values were reasonable, we set the national average to be the limit on the
fraction of EVs that we modeled per representative county in non-Metropolitan Statistical Areas
(MSAs). That is, in this analysis, counties that represent only non-MSA counties were limited to
the national EV sales fraction. All other counties had a maximum limit of 100% EVs. While not
every county approached these limits, some surpassed them. When we compared the EV sales to
all sales by vehicle type and representative county (calculated by applying the representative
county's age distribution to the vehicle population and taking age 0 vehicles as the sales
estimate), we counted the number of "excess" EVs according to these limits and recategorized
and reallocated where they appear using the following algorithm:
1. Where there were excess car EVs, we recategorized as many as possible as truck EVs
in the same representative county (or vice-versa).
2. Where there were excess pickup EVs, we recategorized as many as possible as van
EVs in the same representative county (or vice-versa).
3. In representative counties where there were still excess EVs, we proportionally
reallocated them to all representative counties in the same IPM region that did not
have excess EVs.
4. In the remaining representative counties where there were still excess EVs after the
above steps, we proportionally reallocated them to all counties in the country that did
not have excess EVs.
After the step described above, we mapped the OMEGA vehicle categories to MOVES
vehicles as follows:
"Car" corresponds one-to-one to passenger cars in MOVES (sourceTypelD 21 and
regClassID 20).
"Truck" corresponds to regClassID 30 in MOVES. These vehicles are split into
passenger trucks (sourceTypelD 31) and light commercial trucks (sourceTypelD 32)
using the FF10 31/32 splits by representative county.
"MDV" or medium-duty vans are assumed to be all 2bs, and therefore correspond to
source types 31 and 32 with regClassID 41. They are split into 31s and 32s using the
same splits as "trucks".
"MDP" or medium-duty pickups are assumed to be all 3 s, and therefore correspond to
sourceTypelD 52 with regClass 41.
The number of ICE sales per representative county were calculated by subtracting the EV
sales from the total sales by vehicle type. These were then distributed between the ICE fuel types
using the national default ICE distributions for each model year.
19
-------
Once all excess EVs were reallocated, the light-duty fuel distributions were formatted for use
in the MOVES SampleVehiclePopulation table and were stored in the CDBs.
Note that the heavy-duty fuel distributions were not assumed to vary geographically. The
national average fuel distributions for all heavy-duty source types were used uniformly across all
representative counties. The following figures compare the projected EV penetration rates by
county in 2055 between the reference case and the regulatory case for each light- and medium-
duty source type.
EV Sales
100%
75%
50%
25%
m
0%
Figure 3-4 Comparing passenger car EV penetrations in 2055
20
-------
-------
-------
EV Sales
100%
ฆ
75%
50%
25%
ฆ
0%
Figure 3-7 Comparing class 3 single unit truck EV penetrations in 2055
3.3 Onroad Emissions Modeling
The SMOKE-MOWS process for creating the air quality model-ready onroad mobile
emissions consists of the following steps:
1) Select the representative counties to use in the MOVES runs.
2) Determine which months will be used to represent other months' fuel characteristics.
3) Create inputs needed only by MOVES. MOVES requires county-specific information on
vehicle populations, age distributions, speed distribution, road type distributions,
23
-------
temporal profiles, inspection-maintenance programs, and presence of Low Emission
Vehicle (LEV) program for each of the representative counties.
4) Create inputs needed both by MOVES and by SMOKE, including temperatures and
activity data.
5) Run MOVES to create emission factor tables for the temperatures and speeds that exist in
each county during the modeled period.
6) Run SMOKE to apply the emission factors to activity data (VMT, VPOP, HOTELING,
STARTS, ONI) to calculate emissions based on the gridded hourly temperatures in the
meteorological data.
7) Aggregate the results to the county-SCC level for summaries and quality assurance.
The onroad emissions are processed as five components that are merged 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, and permeation 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;
rate-per-start (RPS) uses START activity data to compute off-network emissions from
vehicle starts;
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; and
rate-per-hour-ONI (RPHO) uses off-network idling hours activity data to compute
emissions for vehicles while idling off-network, (e.g., idling in a parking lot or unloading
freight). This is a new emission calculation which was added to the Cleaner Trucks
Initiative (CTI) version of MOVES.
As described above, versions of MOVES49 (MOVES4.RC2, MOVES4.R1 and
MOVES4.R2). were run for three scenarios: 2016, a 2055 reference case, and a 2055 regulatory
case. Scenario specific EV fractions were developed for each representative county. MOVES
was used to compute onroad emissions in California.
24
-------
SCC descriptions for onroad emissions
SCCs in the onroad sector follow the pattern 220FVV0RPP. where:
F = MOVES fuel type (1 for gasoline, 2 for diesel, 3 for CNG, 5 for E-85, and 9 for
electric)
VV = MOVES vehicle (aka source) type, see Table 3-1
R = MOVES road type (1 for off-network, 2 for rural restricted, 3 for rural unrestricted, 4
for urban restricted, 5 for urban unrestricted)
PP = SMOKE aggregate process. In the activity data, the last two digits of the SCC are
always 00, because activity data is process independent. MOVES separately tracks over a
dozen processes, but for computational reasons it is not practical to model all of these
processes separately within SMOKE-MOVES. Instead, "aggregate" processes are used in
SMOKE. To support this, the MOVES processes are mapped to SMOKE aggregate
processes according to Table 3-2. The MOVES3.R1 model includes a process, 92, that
corresponds to emissions from off-network idling (ONI).
MOVES Vehicle Type
Description
11
Motorcycle
21
Passenger Car
31
Passenger Truck
32
Light Commercial Truck
41
Intercity Bus
42
Transit Bus
43
School Bus
51
Refuse Truck
52
Single Unit Short-haul Truck
53
Single Unit Long-haul Truck
54
Motor Home
61
Combination Short-haul Truck
62
Combination Long-haul Truck
25
-------
Table 3-2 SMOKE-MOVES aggregate processes
MOVES Process ID
Process description
SMOkl. aggregale process
01
Running Exhaust
72
02
Start Exhaust
72
09
Brakewear
40
10
Tirewear
40
11
Evap Permeation
72
12
Evap Fuel Vapor Venting
72
13
Evap Fuel Leaks
72
15
Crankcase Running Exhaust
72
16
Crankcase Start Exhaust
72
17
Crankcase Extended Idle Exhaust
53
18
Refueling Displacement Vapor Loss
62
19
Refueling Spillage Loss
62
90
Extended Idle Exhaust
53
91
Auxiliary Power Exhaust
91
92
Off-network Idle Exhaust
92
3.3.1 Spatial Surrogates
Onroad county activity data were allocated to a national 12 km grid for air quality modeling
using spatial surrogates. For all processes other than the ONI process present in the MOVES3
model, the spatial surrogates used to allocate onroad activity to the national 12km grid are the
same as in the 2016v3 platform and are described in the 2016v3 platform TSD. ONI and other
off-network activity data including VPOP and STARTS were spatially allocated using the
surrogates listed in Table 3-3.
Table 3-3 Spatial surrogates for on-network idling (ONI)
Source Type
Description
Spatial Surrogate
Description
11
Motorcycle
307
NLCD All Development
21
Passenger Car
307
NLCD All Development
31
Passenger Truck
307
NLCD All Development
32
Light Commercial Truck
308
NLCD Low + Med + High
41
Other Bus (non-transit, non-
school)
258
Other Bus Terminals
42
Transit Bus
259
Transit Bus Terminals
43
School Bus
506
Education
51
Refuse Truck
306
NLCD Med + High
52
Single Unit Short-haul Truck
306
NLCD Med + High
53
Single Unit Long-haul Truck
306
NLCD Med + High
54
Motor Home
304
NLCD Open + Low
61
Combination Short-haul Truck
306
NLCD Med + High
62
Combination Long-haul Truck
306
NLCD Med + High
3.3.2 Temporal Profiles
For on-network and hoteling emissions, VMT and hoteling activity were temporally allocated
from annual or monthly values to hourly and SMOKE was run for every day of the year. The
temporal profiles for VMT and hoteling activity are the same as in the 2016v3 platform and are
described in more detail in the 2016v3 platform TSD. ONI monthly activity data were
26
-------
temporally allocated to hourly values using a subset of the temporal profiles that are used to
temporally allocate VMT. VMT data were temporally allocated using temporal profiles which
vary by region (e.g., county, MSA), source type, and road type. ONI activity was developed for
each county and source type, but not road type. This means ONI cannot be temporalized in
exactly the same way as VMT. Instead, a subset of the VMT temporal profiles was selected to be
applied to ONI. Only temporal profiles for unrestricted road types were chosen to be used for
ONI, since off-network idling activity is assumed to better match the temporal pattern of
unrestricted road type driving, rather than on freeways. There are also different VMT temporal
profiles for urban road types and rural road types. ONI activity has no urban or rural designation,
and so within each county, we can only apply either a rural temporal profile or an urban temporal
profile. Therefore, we used the MOVES county classification as either an urban county or a rural
county for the purposes of choosing appropriate temporal profiles for ONI in each county.13 In
urban counties, ONI activity was temporally allocated using VMT profiles for urban unrestricted
roads. For rural unrestricted roads, ONI activity was temporally allocate using VMT profiles.
3.3.3 Chemical Speciation
For onroad and nonroad mobile sources, historically the speciation of total organic gas and
particulate matter emissions has been done by MOVES. However, this is now largely done
outside of MOVES as a post-processing step. This has the advantages of making MOVES
simpler and faster to run and making it easier to change or update chemical mechanisms and
speciation profiles used in the emissions modeling process. Some speciation is still done inside
MOVES for "integrated species" - species of gases and particulate matter which are calculated
directly by MOVES. In many cases, these integrated species are affected by parameters like
temperature or fuel formulation, which are better accounted for within MOVES. For total
organic gases, MOVES calculates 15 integrated species, such as methane and benzene, and the
remainder is called NonHAPTOG and speciated outside MOVES. PM emissions can be
speciated outside of MOVES using similar methodology, but for this platform, PM2.5 onroad
emissions were speciated within MOVES. For nonroad, PM speciation profiles were assigned in
MOVES post-processing and then applied in SMOKE.
In MOVES, speciation profiles for both gaseous and PM emissions are assigned by emission
process, fuel subtype, regulatory class, and model year. Each of these dimensions are available in
MOVES output except for fuel subtype, which is aggregated as part of each fuel type. To apply
speciation outside of MOVES and make it compatible with the needs of SMOKE, we need to
determine the speciation profile mapping by SMOKE process (aggregation of MOVES emission
processes) and SMOKE Source Classification Code (SCC), which are defined by fuel type,
source type, and road type.
For this platform, MOVES runs were performed in inventory mode for each representative
county and season (i.e., winter and summer) to compute NonHAPTOG output by emission
process, fuel type, regulatory class, and model year. Emissions were then disaggregated by fuel
subtype using the market share of each fuel blend in each county, so that speciation profiles can
be accurately assigned. After this step, emissions were normalized and aggregated to calculate
the percentage of total NonHAPTOG and (for nonroad) PM emissions that should be speciated
by each profile for each SMOKE SCC and process. Finally, these percentages were applied in
SMOKE-MOVES to all counties based on their representative county. A MOVES post-
processing tool was then used to generate the needed data for preparing speciation cross-
references (GSREFs) for SMOKE from the outputs of the inventory mode runs. Although they
27
-------
are similar in nature and outcome, the post-processing tools used for onroad and nonroad
emissions output from MOVES are different.
To generate onroad emissions and to perform the subsequent speciation, SMOKE-MOVES
was first run to estimate emissions and both the MEPROC and INVTABLE files were used to
control which pollutants are processed and eventually integrated. From there, the NONHAPTOG
emission factor tables produced by MOVES were speciated within SMOKE using the GSREF
files output from the MOVES postprocessing and the NONHAPTOG GSPRO files generated by
the S2S-Tool. Overall, this process allows most speciation to occur outside of MOVES, which
better supports processing of onroad emissions for multiple chemical mechanisms without
having to rerun the MOVES model. Further details on speciation methods involving MOVES
can be found in the associated technical reports (EPA-420-R-22-017, EPA-420-R-23-006).25
3.3.4 Other Ancillary Files
SMOKE-MOVES requires several other types of ancillary files to prepare emissions for air
quality modeling:
Mobile county cross reference (MCXREF): Maps individual counties to representative
counties.
Mobile fuel month cross reference (MFMREF): Maps actual months to fuel months for
each representative county. May through September are mapped to the July fuel month,
and all other months to the January fuel month.
MOVES lookup table list (MRCLIST): Lists emission factor table filenames for each
representative county.
Mobile emissions processes and pollutants (MEPROC): Lists which pollutants to include
in the SMOKE run.
Meteorological data for MOVES (METMOVES): Gridded daily minimum and maximum
temperature data. This file is created by the SMOKE program Met4moves and is used for
RatePerProfile (RPP) processing.
4 EGU Emissions Inventory Methodology
This section focuses on the approach and data sources used to develop gridded, hourly
emissions for the electrical generating unit (EGU) or "power plant" sector that are suitable for
input to an air quality model in terms of the format, grid resolution, and chemical species.
25 https://www.epa.gov/moves/moves-onroad-technical-reports
28
-------
4.1 Integrated Planning Model (IPM)
IPM is a linear programming model that accounts for variables and information such as
energy demand, planned unit retirements, and planned rules to project unit-level energy
production and configurations.
4.2 IPM 2022 Post-IRA
The version of IPM used to generate EGU inventories for the AQM analysis is the 2022 Post-
IRA version with Final Good Neighbor Plan (GNP), which includes the Inflation Reduction Act
Provisions reflecting supply-side impacts.26
The IRA provisions modeled within IPM included:
Clean Electricity Production and Investment Tax Credits
Existing Nuclear Production Tax Credit
Carbon Capture and Storage 45Q Tax Credit
This modeling did not include other power sector impacts, such as demand impacts from
higher levels of vehicle electrification or IRA energy efficiency provisions.
IPM was run for a set of years, including 2050, with 2055 as the furthest out year. We used
the 2050 outputs, and assumed they are constant through 2055, to avoid end of timeframe issues.
All inputs, outputs and full documentation of EPA's IPM Post-IRA 2022 Reference Case and the
associated NEEDS version is available on the power sector modeling website. The inputs and
outputs for the AQM reference and policy scenarios described in this Section are also available
in the docket for the rule.27
4.2.1 AQM Reference Scenario and Incremental Demand Input Files
IPM requires an electricity demand, and the default electricity demand for the version of IPM
used to run the LMDV AQM reference scenario is based on AEO 2021, which does not include
the full forecasted zero emission vehicle (ZEV) adoption. Relative to AEO 2021, the LMDV
AQM reference case has increased HD ZEV adoption (to account for California's Advanced
Clean Trucks Regulation)28 and LD BEV adoption (to account for EPA's Revised 2023 and
Later Model Year Light-Duty Vehicle Greenhouse Gas Emissions Standards (LD GHG 2023-
2026)) final rule (86 FR 74434, December 30, 2021).29 Therefore, we developed IPM input files
specific to the demand of electric vehicles not captured by IPM's defaults, which we call
26 https://www.epa.gov/power-sector-modeling/final-pm-ncKMs
27 Web-ready IPM files for the LMDV FRM AQM Reference scenario and LMDV FRM AQM Policy scenario.
28 California Air Resources Board, Final Regulation Order - Advanced Clean Trucks Regulation. Filed March 15,
2021. Available at: https://ww2.arb.ca.gov/sites/default/files/barcu/regact/2019/act2019/fro2.pdf.
29 Beardsley, Megan. 2023. "Updates to MOVES for the Multi-Pollutant Emissions Standards for Model Years 2027
and Later Light-Duty and Medium-Duty Vehicles." Memorandum to the Docket EPA-HQ-OAR-2022-0829.
29
-------
incremental demand input files. The IPM incremental demand for LD and HD is the NPRM No-
Action case, detailed in Chapter 5 of the Draft RIA for the proposal.30
4.2.1.1 Light-duty incremental demand
Charging profiles for light-duty PEVs were sourced from the Electric Vehicle Infrastructure
Projection Tool (EVI-Pro) Lite developed by the National Renewable Energy Laboratory in
collaboration with others.31 EVI-Pro Lite allows users to generate charging profiles32 for
different scenarios based on the number33 and mix of vehicles, daily vehicle miles traveled,
ambient temperature, and availability and preference for certain charging types and charging
strategies. While full customization isn't possible in the tool, we generally tried to make
selections among the available options most consistent with our reference case where applicable,
using default selections for other variables.34 The resulting weekday and weekend charging
profiles35 are shown in Figure 4-1.
30 US EPA, 2023. Multi-Pollutant Emissions Standards for Model Years 2027 and Later Light-Duty and Medium-
Duty Vehicles - Draft Regulatory Impact Analysis. EPA-420-D-23-003. See Table 5-2.
https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P10175J2.pdf
31 U.S. Department of Energy, Alternative Fuels Data Center. 2023. "Electric Vehicle Infrastructure Projection Tool
(EVI-Pro) Lite." Available at: https://afdc.energy.gov/evi-pro-lite/load-profile.
32 The tool asks users to select a city or urban area, which changes default selections for average ambient temperature
and vehicle miles traveled. Since we use the resulting profiles nationwide, we made selections (e.g., 50ฐF) intended
to reflect that.
33 We selected 30,000 PEVs (the highest default option available in the tool). However, it is important to note that we
do not use the charging profiles from EVI-Pro Lite to estimate the amount of PEV demand. Rather, we use the profiles
only to distribute our estimate of PEV demand for the Reference and Regulatory cases by hour of day.
34 We made the following selections: average daily miles traveled per vehicle: 35 miles; average ambient temperature:
50ฐF; PEVs that are all-electric: 75% (highest available option); PEVs that are sedans: 50%; mix of workplace
charging: 20% Level 1 and 80% Level 2; access to home charging: 75%; mix of home charging: 50% Level 1 and
50% Level 2; preference for home charging: 100%; home charging strategy: immediate - as fast as possible; work
charging strategy: immediate - as fast as possible.
35 Profiles from the EVI-Pro Lite tool are generated in 15-minute increments. Here we have aggregated to hourly
shares for use in IPM. We also normalized profiles such that the sum of hourly demand shares totals 100%.
30
-------
Figure 4-1: Charging profiles for light-duty PEV demand in the reference Case36
4.2.1.2 Heavy-duty incremental demand
We used the output of national MOVES3.R1 runs to develop the set of IPM incremental
heavy-duty demand input files. Electricity demand was calculated using the MOVES national
modeling domain, with output by each type of day (i.e., for an average weekday and weekend).
IPM requires grid demand to be specified by day type, by each of IPM's geographic regions, and
by each hour of the day.
IPM requires grid demand to be geographically allocated by IPM region. We developed
regional allocation factors based on county-level CO2 emissions in the 2016v2 emissions
modeling platform.37-38 We used CO2 emissions as our basis for regional allocation because CO2
scales well with VMT while capturing differing fleet characteristics in different counties. IPM
includes a mapping of each county to an IPM region, which we used to aggregate county
allocation factors by IPM region.
Inputs to the IPM model include not only the anticipated electricity demand from plug-in
electric vehicles (PEVs), but also how that demand is distributed by time of day. This will
depend on when PEVs charge. We develop and apply charging profiles to reflect the share of
demand from PEV charging that we assume occurs each hour on weekdays and weekends.
36 We use light-duty charging profiles to distribute PEV demand for cars, passenger trucks, and light commercial
trucks (MOVES vehicle types 21, 31, and 32, see Table 3-1).
37 The emissions modeling platform is a product of the National Emissions Inventory Collaborative consistent of more
than 245 employees of state and regional air agencies, EPA, and Federal Land Management agencies. It includes a
full suite of base year (2016) and projection year emission inventories modeled using EPA's full suite of emissions
modeling tools, including MOVES, SMOKE, and CMAQ.
38 U.S. EPA. "2016v2 Platform". January 23, 2023. Available online: https://www.epa.gov/air-emissions-
modeling/2016v2-platform
31
-------
Heavy-duty vehicles comprise a broad spectrum of vehicle types and applications, and we
would expect charging patterns to vary accordingly. For this reason, we develop individual
charging profiles for seven vehicle categories: transit buses, school buses, other buses, refuse
trucks, single unit short-haul trucks, combination short-haul trucks, and motorhomes. We start
from data on vehicle soaks (or times when vehicles are not operating) in MOVES3.R1 for each
of the above categories. For our analysis, we considered only soak lengths that were greater than
or equal to 12 hours, using this as a proxy for when vehicles may be parked at a depot,
warehouse, or other off-shift location and may have an opportunity to charge. How long a
particular vehicle will take to charge will depend on a variety of factors including the vehicle's
daily electricity consumption and the power level of the charging equipment. The time that
charging occurs will also depend on the charging preferences of BEV owners or operators. Some
may choose to start charging as soon as the vehicle is parked, while others may delay charging to
accommodate other vehicles in a fleet, take advantage of time-of-use electricity rates, or for other
reasons. In developing national, fleetwide profiles, we made the simplifying assumption that
charging demand would be evenly distributed across the 12 hours before vehicles start daily
operation, i.e. when the soak periods end.
As a final step, we weighted the seven individual charging profiles by the relative share of
electricity demand for each vehicle category in MOVES3.R1. The resulting aggregate weekday
and weekend profiles are shown in Figure 4-2.
Figure 4-2: Charging profiles for heavy-duty PEV demand in the reference case39
Finally, upstream emissions that would be incurred for fuel cell electric vehicles (FCEVs) due
to the production of hydrogen are not captured by MOVES. We made a simplifying assumption
that all hydrogen used to fuel FCEVs is produced via the electrolysis of water, and thus in this
39 We use heavy-duty charging profiles to distribute demand for PEVs of MOVES vehicle type 41 and higher (see
Table 3-1).
32
-------
analysis, all hydrogen production is represented as additional demand to EGUs and the emissions
are modeled using IPM. Hydrogen in the U.S. today is primarily produced via steam methane
reforming (SMR) largely in support of petroleum refining and ammonia production. New
transportation demand and economic incentives may shift how hydrogen is produced, and
electrolysis is a key mature technology for hydrogen production. The relative emissions impact
of hydrogen production via SMR versus electrolysis depends on the source of electricity
generation, and this varies significantly by region across the country. Electrolysis powered by
electricity from the grid on average in the U.S. may overestimate the upstream emissions impacts
that are attributable to HD FCEVs in the near-term.
We developed yearly scalar multipliers which were applied to MOVES FCEV energy
consumption to represent total grid demand from the hydrogen production necessary to support
the projected levels of FCEVs. First, we assumed hydrogen is produced by a series of
decentralized, grid-powered polymer electrolyte membrane (PEM) electrolyzer systems, each
with a hydrogen production capacity around 1,500 kilograms per day.40'41 Next, we assumed the
gaseous hydrogen is compressed and pre-cooled for delivery to vehicles using grid-powered
electrical equipment. Finally, we assumed a linear improvement between our estimated current
and future efficiency for hydrogen production. The linear interpolation is between current values
that start in 2025 and future values represented for 2055, assuming a period of diffusion for more
efficient electrolysis technology improvements to spread. The final scaling factors range from
1.748 in 2025 to 1.616 in 2055.
4.2.2 AQM Policy Scenario and incremental demand inputs
The default electricity demand for the version of IPM42 used to run the LMDV AQM policy
scenario is based on AEO 2021, which does not include the full forecasted zero emission vehicle
(ZEV) adoption. As mentioned above, we developed light- and medium-duty incremental
demand input files for the LMDV AQM reference case and in addition to those files, described
in Section 4.2.1.1, the incremental light-duty demand input files for the policy scenario also
included light- and medium-duty EV demand to represent the NPRM Action case. The light- and
medium-duty demand associated with the NPRM action case is detailed in Chapter 5.2.3 of the
Draft RIA for the proposal.43 We used regional profiles, generated using EVI-X, for the light-
duty PEV demand in our policy case, see Chapter 5.1 and 5.3 in the DRIA for more detail on
EVI-X.
We also developed heavy-duty incremental demand input files for the LMDV AQM policy
case. We used the output of national MOVES3.R1 runs to develop the set of IPM incremental
heavy-duty demand input files. Electricity demand was calculated using the MOVES national
modeling domain, with output by each type of day (i.e., for an average weekday and weekend).
40 This is based on assumptions from the Hydrogen Analysis Production (H2A) Model from the National Renewable
Energy Laboratory (NREL).
41 National Renewable Energy Laboratory (NREL). "H2A: Hydrogen Analysis Production Model: Version 3.2018".
Available online: https://www.nrel.gov/hydrogen/h2a-production-archive.html
42 https://www.epa.gov/power-sector-modeling/post-ira-2022-reference-case
43 US EPA, 2023. Multi-Pollutant Emissions Standards for Model Years 2027 and Later Light-Duty and Medium-
Duty Vehicles - Draft Regulatory Impact Analysis. EPA-420-D-23-003.
https://nepis.epa.gov/Exe/ZyPDF.cgi?Dockey=P10175J2.pdf
33
-------
The heavy-duty EV charging profiles were the same for the reference and policy cases, see
Section 4.2.1.1.2.
4.3 Air Quality Model-Ready EGU inventory generation
The EGU emissions are calculated for the AQM inventory using the output of the IPM model
for the forecast year. Units that are identified to have a primary fuel of landfill gas, fossil waste,
non-fossil waste, residual fuel oil, or distillate fuel oil may be missing emissions values for
certain pollutants in the generated inventory flat file. Units with missing emissions values are
gapfilled using projected base year values.
The projections are calculated using the ratio of the future year seasonal generation in the IPM
parsed file and the base year seasonal generation at each unit for each fuel type in the unit as
derived from the 2018 EIA923 tables and the 2018 NEI. New controls identified at a unit in the
IPM parsed file are accounted for with appropriate emissions reductions in the gapfill projection
values. When base year unit-level generation data cannot be obtained no gapfill value is
calculated for that unit. Additionally, some units, such as landfill gas, may not be assigned a
valid SCC in the initial flat file. The SCCs for these units are updated based on the base year
SCC for the unit-fuel type. Combined cycle units produce some of their energy from process
steam that turns a steam turbine. The IPM model assigns a fraction of the total combined cycle
production to the steam turbine. When the emissions are calculated these steam units are
assigned emissions values that come from the combustion portion of the process. In the base year
NEI steam turbines are usually implicit to the total combined cycle unit. To achieve the proper
plume rise for the total combined cycle emissions, the stack parameters for the steam turbine
units are updated with the parameters from the combustion release point. Large EGUs in the
IPM-derived flat file inventory are associated with hourly CEMS data for NOx and SO2
emissions values in the base year. To maintain a temporal pattern consistent with the 2016 base
year, the NOx and SO2 values in the hourly CEMS inventories are projected to match the total
seasonal emissions values in the future years.
5 Petroleum Sector Emissions Inventory Methodology
This section focuses on the approach and data sources used to develop adjusted gridded,
hourly emissions for some of the sectors related to producing petroleum liquid fuels for mobile
sources. While the emission factors used to develop emissions for the reference and policy
scenarios differed, the approach and data sources used to calculate emissions for both scenarios
were consistent.
Emission sources related to producing petroleum liquid fuels for mobile sources include
extracting, transporting, and storing crude oil; extracting, transporting, and storing natural gas;
and refining, transporting, and storing finished fuels like gasoline and diesel. These sources are
described in the emissions modeling platform TSD in Section 2.1.2 (point oil and gas) and 2.2.4
(nonpoint oil and gas).44
More details on the modeling of the petroleum sector emissions are in the following
subsections, and national emission summaries for key pollutants are provided in Section 6. The
44 U.S. EPA (2023) Technical Support Document: Preparation of Emissions Inventories for the 2016v3 North
American Emissions Modeling Platform, https://www.epa.gov/air-emissions-modeling/2016-version-3-technical-
suppo rt~docume nt.
34
-------
docketed spreadsheet "LMDV FRM AQM petroleum adjustment factors.xlsx" presents the
calculations described in this Section.
5.1 Refinery Emissions
5.1.1 Projection of Refinery Emissions to 2050/2055
The 2016v3 emissions modeling platform, which includes projection years 2023 and 2026,
was the starting point for the air quality analysis to develop refinery inventories.45 The 2026
refinery inventory from the 2016v3 emissions modeling platform was projected to 2050 using
AEO 2023 growth factors.46'47 We assumed no change in refinery emissions between 2050 and
2055. The national total refinery inventory is presented in Table 5-1, and see docketed
spreadsheet "2050 national refinery summary for OTAQ FRM.xlsx".
Table 5-1 2016v3 Emissions Modeling Platform Refinery Inventory Projected to 2026 and 2055
Pollutant
Projected emissions in 2026 (tons/yr)
Projected emissions in 2055 (tons/yr)
NOx
76,447
81,607
pm25
18,231
19,243
so2
25,164
26,287
voc
63,033
64,091
5.1.2 Identifying Refineries to Adjust for Air Quality Analysis
To isolate the impact of this rule on refinery emissions, only refineries that produce gasoline
or diesel fuel for onroad vehicles were adjusted in the air quality modeling. For the NPRM
illustrative air quality analysis, eligible refineries were identified from the 2016v2 emissions
modeling platform refineries report, and those that did not produce gasoline or diesel fuel for
onroad vehicles were excluded (see docketed spreadsheet, "2016v2 platform refineries
report.xlsx"). In preparation for the final rule, the same approach was applied to new refineries
that had been added to the 2016v3 emissions modeling platform (see docketed spreadsheet
"refineries in 2016v3 not 2016v2.xlsx"). Ultimately, 118 refineries that produce onroad fuel
were adjusted in the air quality modeling.
5.1.3 Apportioning Total Refinery Emissions to Gasoline and Diesel Fuel Production
Scaling factors were calculated to apportion total refinery emissions to the refining of gasoline
and diesel versus other refined fuels and refinery operations. The scaling factors are based on the
45 https://www.epa.gov/air-emissions-modeling/2016v3-platform
46 Specifically, a projection packet was prepared for 2026->2050 using AEO 2023 for refineries. AEO categories were
mapped to SCCs and SCC+NAICS combinations (with SCC+NAICS taking precedence if a mapping exists for the
refinery NAICS, which are 32411/324110) using the usual industrial source AEO-SCC and AEO-SCC-NAICS xrefs
from past platforms. Only refineries NAICS and SCCs which have refinery emissions were included when making
the packet, so the 2026-2050 packet is not something that can be used to project the entire ptnonipm sector. Each
record in the packet references the refineries NAICS so that it can be applied to the entire ptnonipm sector without
changing any non-refineries.
47 https://www.eia.gov/outlooks/aeo/tables_ref.php
35
-------
relative energy demand of refining various fuels calculated by Wang et al.48 Wang et al.
expressed the energy demand of refining fuels in terms of mass and included outputs that are not
refinery products (i.e., fuel gas), so we removed non-refinery products and adjusted the energy
demand factors to be based on volume instead of mass.
Relative emissions related to the refining of various products are determined primarily by the
energy needed to refine those products, but also depend on pollutant-specific emissions from
refining those products. For example, the refining of gasoline causes higher methane emissions
than an equivalent volume of diesel. We developed pollutant-specific apportionment factors
based on relative emissions of refining gasoline, diesel, and other products using emission
factors from GREET 2021.49 We use the apportionment factors to calculate the portion of the
refinery inventory attributable to the refining of each fuel type. Final apportionment factors for
each pollutant that we modeled in our refinery analysis appear in Table 5-2.
Table 5-2 Refinery emission apportionment factors by fuel type
Pollutant
Refinery Emissions Apportionment Factor
Gasoline
Diesel
Other
Carbon Dioxide (CO2)
0.591
0.061
0.348
Methane (CH4)
0.640
0.053
0.307
Nitrous Oxide (N20)
0.583
0.063
0.354
Nitrogen Oxides (NOx)
0.610
0.056
0.334
Particulate Matter (PM2 5)
0.620
0.054
0.326
Sulfur Dioxide (SO2)
0.596
0.058
0.346
Volatile Organic Compounds (VOC)
0.570
0.058
0.372
Table 5-3 shows how we estimated 2050 refinery emissions that are attributable to the
refining of gasoline and diesel fuel. We began with the total refinery inventory, which was
reduced to only represent refineries that produce onroad fuels (see Section 5.1.2.). Then, we
further apportioned emissions to be specific to the refining of gasoline or the refining of diesel.
Table 5-3 2050 refinery emission inventory apportioned by refinery type and fuel type
Pollutant
Emission Inventory by Refinery
Group (U.S. Tons)
Inventory Apportioned by
Fuel Type (U.S. Tons)
All
Refineries
Refineries that produce
gasoline and diesel
Gasoline
Diesel
Nitrogen Oxides (NOx)
81,607
77,830
47,437
4,335
Particulate Matter (PM2 5)
19,243
18,253
11,324
605
Sulfur Dioxide (SO2)
26,287
23,501
14,017
819
Volatile Organic Compounds (VOC)
64,091
57,829
32,972
1,924
48 Wang, M., Lee, H. & Molburg, J. Allocation of energy use in petroleum refineries to petroleum products. Int J
LCA 9, 34-44 (2004). https://doi.org/10.1007/BF02978534
49 Wang, Michael, Elgowainy, Amgad, Lee, Uisung, Bafana, Adarsh, Baneijee, Sudhanya, Benavides, Pahola T.,
Bobba, Pallavi, Burnham, Andrew, Cai, Hao, Gracida, Ulises, Hawkins, Troy R., Iyer, RakeshK., Kelly, Jarod C.,
Kim, Taemin, Kingsbury, Kathryn, Kwon, Hoyoung, Li, Yuan, Liu, Xinyu, Lu, Zifeng, Ou, Longwen, Siddique,
Nazib, Sun, Pingping, Vyawahare, Pradeep, Winjobi, Olumide, Wu, May, Xu, Hui, Yoo, Eunji, Zaimes, George G.,
and Zang, Guiyan. Greenhouse gases, Regulated Emissions, and Energy use in Technologies Model ฎ (2021
Excel). Computer Software. U.S. Department of Energy (DOE) Office of Energy Efficiency and Renewable Energy
(EERE). 11 Oct. 2021. Web. doi:10.11578/GREET-Excel-2021/dc.20210902.1.
36
-------
5.1.4 Total Refined Fuel and Onroad Fuel Consumed Associated with AQM Cases
We estimated total reference case refinery activity in terms of gasoline and diesel produced.
The total refined fuel supplied in 2050 was obtained from AEO 2023 Table ll50'51 and is
presented here in Table 5-4. We assume that 2050 projected volumes stay constant through 2055.
It is important to note that an error was made in interpreting the total refined fuel supplied that is
presented in AEO 2023 Table 11. "Product suppled" was assumed to be the volume of fuel
refined in the United States; however, after the AQM was underway, we learned that this value
presented in AEO 2023 Table 11 does not include fuel that was refined and exported. The United
States is a net exporter of gasoline and diesel and therefore, the total refinery activity that was
assumed for the reference case is underestimated. We estimate that this error has had a relatively
small impact on the air quality modeling results compared to the total emission reductions from
the policy scenario; its implications are discussed further in Section 5.1.7.
Table 5-4 Total Refined Fuel Supplied in 2050 from AEO 2023 Reference Case (billion gallons/yr)
Total Refined Fuel
Supplied3
Gasoline
113.92
Diesel
49.12
a Total refined fuel supplied from Table 11 of AEO 2023, with units converted from million barrels per day
The fuel demanded in 2055 by onroad vehicles (gallons of gasoline and gallons of diesel) in
the policy scenario was generated using MOVES (see docketed spreadsheets "FRM reference
petroelumconsumption.xlsx" and "FRM policy petroelumconsumption.xlsx").
There are methodological differences in how onroad fuel demand is calculated by MOVES
and by AEO. An adjustment factor to account for the difference between MOVES4.R1 and AEO
2023 reference was applied to the MOVES onroad fuel demand numbers to make them more
consistent with AEO 2023, see Table 5-5.
Table 5-5 Factor to apply to MOVES fuel demand to make consistent with AEO fuel demand
MOVES adjustment
factor
Gasoline
1.01
Diesel
0.93
The adjusted MOVES onroad fuel demand was then used to calculate the change in fuel
demand from the reference onroad fuel demand obtained from AEO 2023 Table 36
('Transportation Energy Use: Light-Duty Vehicle: Total').52 The reduction in onroad fuel
demand due to the policy scenario was calculated separately for gasoline and for diesel by
50 https://www.eia.gov/outlooks/aeo/data/browser/#/?id=ll-AE02023&cases=ref2023&sourcekey=0
51 From the Annual Energy Outlook 2023 Table 11 'Petroleum and Other Liquids Supply and Disposition
Case: Reference case', "Product Supplied - by Fuel - Motor Gasoline" in 2050 was used as the total refined
gasoline supplied, and "Product Supplied - by Fuel - Distillate Fuel Oil - of which: Diesel" in 2050 was used as the
total refined diesel supplied.
52 https://www.eia.gov/outlooks/aeo/data/browser/#/?id=46-AE02023&cases=ref2023&sourcekey=0
37
-------
subtracting the fuel demands in Table 5-6, in which the MOVES policy case estimates have been
adjusted to account for AEO/MOVES methodological differences.
Table 5-6 2055 Onroad Fuel Demand for Air Quality Analysis Scenarios (billion gallons/yr)
Reference Onroad Fuel
Demand
LMDV Policy Onroad
Fuel Demand
Gasoline
105.55
55.60
Diesel
36.95
31.41
5.1.5 Projected Change in U.S. Refinery Activity Related to Decreased Domestic Demand
We estimate the change in refinery activity by assuming a reduction in onroad fuel demand
will lead to a reduction in the total amount of fuel refined. However, U.S. refineries can
theoretically respond to lower domestic demand by increasing volumes of exported liquid fuels,
thus allowing them to refine at the same volume and leaving refinery emissions unchanged.
In the NPRM air quality analysis, we assumed that 7% of the reduced domestic demand for
refined fuels would be made up by an increase in exports, based on a comparison of the reference
case and low economic growth case in AEO 2023. We received comments from several
organizations that refineries would increase exports more than we assumed.
There are several reasons to expect refineries to increase exports in the case that domestic
demand for refined fuels drops in the future. First, most refineries refine products in addition to
onroad fuels. In fact, the refining of gasoline and diesel fuel produces coproducts that have
economic value of their own, so refineries may continue refinery activity but focus on other
products. Second, it can be economically advantageous to refine crude oil in the United States
because feedstock prices tend to be lower, thus leading to higher profit margins.
Despite the favorable economic conditions for refiners in the United States, there have been
some refinery closures and conversions in recent years, often at least partially in response to
lower domestic fuel demand from the COVID-19 pandemic and the desire for low-carbon fuels.
The closure or conversion of some U.S. refineries in recent years suggests that the closure or
conversion of additional refineries is likely as domestic demand for gasoline and diesel fuel
declines, especially for those that have lower margins or face other issues. The extent to which
U.S. refineries keep operating, shut down, or are converted is difficult to project, since it depends
on the economics of individual refineries, the economic condition of the parent company, and the
long-term strategy pursued by each company's board for providing a return to its shareholders.
After carefully taking into consideration stakeholder comments, the more desirable economic
conditions for refiners in the U.S., and the recent closures and conversions of some U.S.
refineries, we have updated our projection of how refineries will be impacted by this rulemaking.
For the final rule, we estimated refinery emissions by assuming that U.S. refineries would
increase exports to offset half of the projected reductions in domestic demand for liquid fuels.
Thus, the total decrease in refinery activity, measured in gallons of gasoline and diesel refined, is
half of the estimated drop in domestic fuel demand (see Table 5-7). However, there remains
significant uncertainty in how U.S. refineries will respond to lower demand for liquid onroad
fuels. Therefore, we performed a sensitivity analysis in which we estimate emission impacts if
refineries had no change in activity as a result of reduced domestic demand from this rule. In the
sensitivity analysis, presented in Chapter 7 of the RIA, we present total emission impacts of the
policy scenario with no change in refinery emissions from the reference case.
38
-------
Table 5-7 Reductions in Onroad Fuel Demand and Refinery Activity for 2050 Policy Scenario (billions
gallons/year)
Reductions in Onroad
Fuel Demand
Reductions in Refined
Gasoline and Diesel
Gasoline
49.94
24.97
Diesel
5.54
2.77
5.1.6 Generation of Adjustment Factors
The reduced gallons of onroad gasoline and diesel that would be refined domestically (Table 5-7)
was subtracted from the total refined fuel supplied in Table 5-4 and used to create an adjustment
factor to be applied to the gasoline and diesel portions of the 2050 onroad refinery inventory.
Adjustment factors of 0.78 and 0.94 were applied to gasoline and diesel portions, respectively.
The resulting emissions, associated with refining gasoline and diesel fuel only, are presented in
Table 5-8.
Table 5-8 Projected 2055 Emissions from Refineries Associated with Producing Gasoline and Diesel
Only
Scenario
NOx (tons/yr)
PM2.5 (tons/yr)
SO2 (tons/yr)
VOC (tons/yr)
LMDV
Policy
gasoline
37,039
8,842
10,944
25,744
diesel
4,091
571
773
1,815
The total refinery emissions in the policy scenario is estimated as the total refinery emissions
in the reference case less the projected reductions in gas and diesel refining associated with this
rule. A final adjustment factor, equal to the ratio of the total refinery emissions in the regulatory
case to the total refinery emissions in the reference case, was then calculated for each of the
pollutants included in air quality analysis (see Table 5-9). These adjustment factors were applied
in air quality modeling to each of the refineries that produce onroad fuel (see Section 5.1.2).
Table 5-9 Adjustment Factor to Apply to 2050 Refinery Inventory
Scenario
NOx
PM25
SO2
VOC
LMDV Policy
0.86
0.86
0.87
0.87
5.1.7 Limitations of Modeling Impacts on Refinery Emissions
5.1.7.1 Uncertainty in impact on refinery activity
As noted in Section 5.1.5, we recognize that there is significant uncertainty in the impact that
reduced domestic demand for gasoline and diesel fuel will have on refinery emissions and that
the refinery industry could respond differently than how we have predicted in our air quality
analysis. For example, many US refineries may continue their production of refined products and
instead import less refined product because they experience lower crude oil and natural gas
prices than refineries elsewhere. Some refineries may also increase exports of US refined
products. If refineries employ these strategies and their production is unaffected by lower
39
-------
domestic demand, we would project no emission reductions from refineries rather than those
associated with the adjustment factors presented in Table 5-9.
5.1.7.2 Overestimation of reduction in refinery inventory
As mentioned in Section 5.1.4, we underestimated the total refined fuel in the reference case
by not including fuel that was refined in the US and then exported. Therefore, the adjustment
factors we applied in the air quality analysis overestimate the relative reduction in the refinery
inventory between the reference and policy cases. This error was discovered after air quality
modeling was already underway, so we were unable to correct it due to time constraints.
However, we have estimated its impact on the total refinery emissions and the total emissions
across all sectors in the policy case, and we conclude that although the overestimate of emissions
reductions is non-negligible, it is relatively small.
To quantify the magnitude of the overestimate, we first estimated exports of gasoline and
diesel in the reference case. Growth factors for exports were estimated for 2050 versus 2022
using the projected net exports provided in Table 11 of AEO 2023.53 These growth factors were
then applied to apportioned exports of refined fuels measured by EIA in 2022 to estimate net
exports of gasoline and diesel in 2050. The projected net exports were added to the total refined
fuel that had been used for the reference case, and the method for estimating adjustment factors
for the regulatory scenario refinery emissions was repeated using the updated estimate of total
refined fuel that included net exports of gasoline and diesel in 2050.54 Compared to the
adjustment factors that were applied to relevant refineries in the AQM (see Table 5-9), the
corrected adjustment factors after the addition of refined fuels exports were ~2 percentage points
higher (see Table 5-10). We then applied the corrected adjustment factor to the emissions
inventories for all refineries that produce onroad fuel and compared the estimated refinery
emissions impacts against those calculated from the uncorrected adjustment factors to quantify
the magnitude of the error on projected refinery emissions impacts (see Table 5-11). Using this
method, we conclude that the exclusion of net exports from the initial adjustment factor
calculations has likely resulted in an overestimate of refinery emissions reductions of 16%.
Finally, to understand the impact that this overestimate had on the total projected emissions
impacts from all sources, we reduced the refinery-related emissions impacts by 16% and
summed new, corrected total emissions reductions (see Table 5-12). Note that the uncorrected
emissions reductions presented in Table 5-12 represent projected changes in refinery activity
among the 48 contiguous United States, whereas the reductions presented in Table 5-11 also
include refineries in Alaska and Hawaii (see Section 5.1.1).
53 https://www.eia.gov/outlooks/aeo/data/browser/#/?id=ll-AE02023&cases=ref2023&sourcekey=0
54 The calculations are provided on docketed spreadsheet "LMDV FRM AQM petroleum adjustment factors with
exports for refinery corrections.xlsx".
40
-------
Table 5-10 Corrected Adjustment Factors with Addition of Net Exported Refined Fuels
Pollutant
Uncorrected
Corrected
NOx
0.86
0.89
PM2.5
0.86
0.88
S02
0.87
0.89
voc
0.87
0.89
aHere, 'uncorrected' maintains the error that was made prior to AQM.
These data represent what was used as AQM inputs.
bThe corrected emissions were recalculated with the inclusion of net
exports of refined fuels to address the overestimating error.
Pollutant Uncorrected3
Corrected13
Difference % difference
Refineries Emissions Only
PM2.5
2,516
2,112
404
16%
NOx
10,643
8,913
1,730
16%
S02
3,119
2,616
503
16%
VOC
7,336
6,155
1,181
16%
aHere, 'uncorrected' maintains the error that was made prior to AQM. These data represent what was used as AQM inputs.
bThe corrected emissions were recalculated with the inclusion of net exports of refined fuels to address the overestimating
error.
Table 5-
-12 Corrected Emissions Impacts with Addition of Net Exported Refined Fuels
Sector (48 states)
Uncorrected
Corrected
Difference
% difference
PM2.5
Onroad Total
-8,326
-8,326
-
-
Upstream Total
-1,393
-996
397
28%
EGU
1,039
1,039
-
-
Refinery
-2,467
-2,070
397
16%
Crude Production Wells
-102
-102
-
-
Natural Gas Production Wells
137
137
-
-
PM25
Total
-9.719
-9.322
397
4%
NOx
Onroad Total
-84,692
-84,692
-
-
Upstream Total
-9,643
-7,942
1,701
18%
EGU
1,605
1,605
-
-
Refinery
-10,468
-8,767
1,701
16%
Crude Production Wells
-4,778
-4,778
-
-
Natural Gas Production Wells
3,999
3,999
-
-
NOx
Total
-94.335
-92.634
1.701
2%
S02
Onroad Total
-2,334
-2,334
-
-
Upstream Total
-2,929
-2,435
494
17%
EGU
1,946
1,946
-
-
Refinery
-3,067
-2,573
494
16%
Crude Production Wells
-1,867
-1,867
-
-
Natural Gas Production Wells
59
59
-
-
41
-------
Sector (48 states)
Uncorrected
Corrected
Difference
% difference
so2
Total
-5.263
-4.769
494
9%
voc
Onroad Total
-165,159
-165,159
-
-
Upstream Total
-29,029
-27,869
1,160
4%
EGU
467
467
-
-
Refinery
-7,205
-6,045
1,160
16%
Crude Production Wells
-33,343
-33,343
-
-
Natural Gas Production Wells
11,052
11,052
-
-
voc
Total
-194.188
-193.028
1.160
0.6%
5.1.7.3 Uniform application of adjustment factor
Lastly, because we are unable to predict the potential impact that this rule will have on
individual refineries, we have used an adjustment factor method that applies the projected impact
of reduced demand evenly across all relevant refineries, as a scalar of emissions.
5.2 Crude production well and pipeline emissions
5.2.1 Reference Case Crude Production Well Site and Pipeline Inventories for 2050/2055
The emission inventories for crude production wells and associated pipelines in the 2016v3
emissions modeling platform for the year 2026 are projected to the year 2050 using AEO 2023
reference case production forecast data in the year 2050 relative to that in the year 2026. These
reference case crude production well and pipeline inventories were assumed to remain constant
from 2050 to 2055.
5.2.2 Policy Scenario and Associated Crude Demand
The reference case 2050/2055 crude production well and pipeline inventories for 2055 needed
to be adjusted to reflect the impact of the policy scenario which reduced the domestic demand for
liquid fuel, see Table 5-7.55 The total reduced gallons of refined fuel consumed for each scenario
were adjusted to account for refinery efficiency, using factors from Forman, et al (2014), to get
reduced gallons of crude-equivalent finished fuel.56'57 Then the gallons of reduced crude-
equivalent finished fuel were converted to gallons of reduced crude using the energy density of
crude and gasoline and diesel fuel.58
55 The calculations are provided on docketed spreadsheet "LMDV FRM AQM petroleum adjustment factors.xlsx".
56 Forman et al, 2014 dx.doi.org/10.1021/es501035a
57 The conversion of crude oil to products may be more efficient that this value as this value represents the overall
refinery efficiency, not the efficiency for converting crude oil into products.
58 Energy densities came fromEIA, https://www.eia.gov/energyexplained/units-and-calculators/
42
-------
5.2.3 Projected Change in U.S. Crude Production Activity Related to Decreased Domestic
Demand
It was necessary to project how the reduced crude demand associated with the policy scenario
would affect U.S. crude production well and pipeline emissions since U.S. crude demand is also
satisfied by imports, not just domestic production. We projected how the change in crude
demand would affect U.S. crude production based on a comparison generated for the NPRM
AQM analysis using the AEO 2021 Reference case and Low Economic Growth case, and we
retained this factor for the FRM AQM analysis.59 The reduced domestic demand (gallons of
crude) is multiplied by 8% to estimate the reduction in domestically produced crude.60
5.2.4 Generation of Crude Production Well and Pipeline Adjustment Factors
The reduced gallons of crude that would be domestically produced was subtracted from the
total crude produced in the AEO 2023 reference case and used to create an adjustment factor,
0.98, to be applied to the crude production well and pipeline inventories.
Equation 1
AEO2023 reference case Bgal crude produced domestically in 2050
CrudeProdUCtion reduced crude produced domestically due to additional EV penetration
adjustment factor AEO2023 reference case Bgal crude produced domestically in 2050
5.2.5 Limitations of Modeling Impacts on Crude Production Wells and Pipeline Pumps
Because we are unable to predict the potential impact that this rule will have on individual
production wells, we have used an adjustment factor method that applies the projected impact of
reduced demand evenly across all relevant production sites and pipeline pumps, as a scalar of
emissions.
5.3 Natural gas production wells and pipeline pumps emissions
5.3.1 Reference Case Natural Gas Production Well Site and Pipeline Inventories for 2050/2055
Emission inventories for natural gas production wells and associated pipelines in the 2016v3
emissions modeling platform were projected from 2026 to 2050 using AEO 2023 reference case
production forecast data. We assumed no change in refinery emissions between 2050 and 2055.
59 US EPA, 2023. Illustrative Air Quality Analysis for the Light and Medium Duty Vehicle Multipollutant Proposed
Rule - Technical Support Document (TSD). April 2023, EPA-420-D-23-002.
60 An error was made in the NPRM analysis and the AEO 2021 reference case and AEO 2021 low economic growth
case comparison was done for 2030-2050 instead of 2027-2050. It should have started in 2027 as that is the first year
that the rule is implemented. The impact of the error means that the reduced domestic demand should have been
multiplied by 7% instead of 8%. The impact of this error was small enough that it did not change the adjustment
factor that was calculated for crude production wells and pipeline pumps.
43
-------
5.3.2 Policy Scenario and Associated Natural Gas Demand
The reference case natural gas production well and pipeline inventories needed to be adjusted
to reflect the impact of the LMDV policy scenario, which will increase the domestic demand for
electricity, leading to more demand for natural gas.61 Natural gas use projections (trillion cubic
feet) from IPM are presented in Table 5-13, and AEO 2023 reference case projections of the
amount of produced natural gas going to EGUs are presented in Table 5-14.
Table 5-13 IPM projections of Natural Gas Usage, trillion cubic feet, in 2050
Natural Gas Usage (Tcf) |
Reference Case 6.10
LMDV Policy Case 6 50
Table 5-14 Projections of Natural Gas, in trillion cubic feet, in 2050 Reference Case, from AEO 2023
Table 13 _ _
: Natural Gas (Tcf) |
Total Dry Gas Production 42.07
! Consumption of Natural Gas by EGUs 7.74
5.3.3 Generation of Natural Gas Production Well and Pipeline Adjustment Factors
Based on the increased natural gas usage by EGUs indicated in Table 5-13, the LMDV policy
case has 6.6% more natural gas usage than the reference case. The AEO projections from Table
5-14 indicate that 18% of the natural gas projected to be produced domestically in 2050 goes
towards EGUs. The growth factor applied to the reference case natural gas well site and pipeline
pump emission inventories to get the policy scenario natural gas well and pipeline pump
emission inventories was 1.01, see Equation 2.
Equation 2
Growth factor = (1-0.18) + (0.18*1.07)
5.3.4 Limitations of Modeling Impacts on Natural Gas Production Wells and Pipeline Pumps
Because we are unable to predict the potential impact that this rule will have on individual
production wells, we have used an adjustment factor method that applies the projected impact of
increased demand evenly across all relevant production sites and pipeline pumps, as a scalar of
emissions.
61 The calculations are provided on docketed spreadsheet "LMDV FRM AQM petroleum adjustment factors.xlsx"
44
-------
6 Inventory Summary Tables
This section includes summary tables of emission inventories used in the AQM analysis and
described in this document.
Table 6-1 Modeled PM2.5, NOx, SO2, and VOC Annual Emissions Used in AQ Modeling (short tons)
Pollutant
Sector
2016 Base
Year
2055 Cases
Reference Regulatory
Regulatory
Impactb
PM2.5
Onroad Total3
114,519
34,667
26,342
-8,326
Upstream Total3
167,795
64,115
62,722
-1,393
EGU
133,570
26,420
27,459
1,039
Refinery
19,958
18,867
16,399
-2,467
Crude Production Wells + Pipeline Pumps
3,393
5,102
5,000
-102
Natural Gas Production Wells + Pipeline Pumps
10,875
13,726
13,863
137
PM25
Total
282.315
98,782
89.063
-9.719
NOx
Onroad Total3
3,722,735
403,861
319,169
-84,692
Upstream Total3
2,067,563
814,881
805,238
-9,643
EGU
1,319,734
95,934
97,539
1,605
Refinery
78,332
80,188
69,720
-10,468
Crude Production Wells + Pipeline Pumps
161,605
238,895
234,117
-4,778
Natural Gas Production Wells + Pipeline Pumps
507,891
399,863
403,862
3,999
NOx
Total
5.790.298
1.218.742
1.124.407
-94.335
S02
Onroad Total3
25,009
6,458
4,124
-2,334
Upstream Total3
1,637,501
142,170
139,241
-2,929
EGU
1,565,675
17,117
19,063
1,946
Refinery
30,065
25,846
22,779
-3,067
Crude Production Wells + Pipeline Pumps
37,095
93,330
91,464
-1,867
Natural Gas Production Wells + Pipeline Pumps
4,665
5,876
5,935
59
S02
Total
1.662.510
148.628
143.365
-5.263
VOC
Onroad Total3
1,380,318
502,643
337,484
-165,159
Upstream Total3
2,415,830
2,852,174
2,823,145
-29,029
EGU
33,763
17,023
17,490
467
Refinery
67,853
62,842
55,637
-7,205
Crude Production Wells + Pipeline Pumps
1,229,169
1,667,134
1,633,791
-33,343
Natural Gas Production Wells + Pipeline Pumps
1,085,046
1,105,175
1,116,227
11,052
VOC
Total
3.796.149
3,354,817
3.160.629
-194.188
aSectors are for the 48 contiguous United States
bCalculated as the difference between the 2055 Reference Case and the 2055 Regulatory Case emissions values
45
-------
Table 6-2 Modeled 48-state Onroad Emissions (short tons)
Pollutant
2016 Base
Year
2055 Cases
Reference Regulatory
2016 Base vs. 2055
Reference
Difference % Change
Reference vs.
Regulatory
Difference % Change
pm25
114,519
34,667
26,342
79,852
70%
8,326
24%
NOx
3,722,735
403,861
319,169
3,318,874
89%
84,692
21%
S02
25,009
6,458
4,124
18,551
74%
2,334
36%
VOC
1,380,318
502,643
337,484
877,675
64%
165,159
33%
CO
19,218,852
5,035,912
3,248,848
14,182,940
74%
1,787,063
35%
Acrolein
1,480
205
120
1,275
86%
85
41%
Acetaldehyde
13,989
3,285
2,043
10,704
77%
1,242
38%
Benzene
26,255
7,722
4,574
18,533
71%
3,148
41%
1,3-Butadiene
3,694
852
459
2,842
77%
393
46%
Ethylbenzene
20,312
8,046
5,365
12,265
60%
2,682
33%
Formaldehyde
19,539
2,420
1,628
17,120
88%
791
33%
Naphthalene
2,527
316
184
2,210
87%
133
42%
Table 6-3 Modeled 48-state Nonroad Emissions (short tons)
Pollutant
Year
2016
2055
pm25
106,184
55,891
NOx
1,108,985
667,652
S02
1,451
1,248
VOC
1,155,551
954,103
CO
11,257,608
15,083,974
Acrolein
2,067
590
Acetaldehyde
11,099
5,481
Benzene
28,803
28,095
1,3-Butadiene
4,547
4,838
Ethylbenzene
20,239
17,065
Formaldehyde
28,249
13,210
Naphthalene
1,928
1,399
Table 6-4 Modeled 48-state Fugitive Dust Emissions (short tons)
Year
Pollutant
2016
2055
PM25
880.002
921.877
46
-------
7 Air Quality Modeling Methodology
7.1 Air Quality Model - CMAQ
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.62 The air quality modeling completed for the
final rulemaking used the 2016v3 platform with the most recent multi-pollutant CMAQ code
available at the time of air quality modeling (CMAQ version 5.4).63 The 2016 CMAQ runs
utilized the CB6r3 chemical mechanism (Carbon Bond with linearized halogen chemistry) for
gas-phase chemistry, and AER07 (aerosol model with non-volatile primary organic aerosol) for
aerosols. The CMAQ model is regularly peer-reviewed; CMAQ versions 5.2 and 5.3 beta were
most recently peer-reviewed in 2019 for the U.S. EPA.64
7.2 CMAQ Domain and Configuration
The CMAQ modeling analyses used a domain covering the continental United States
(CONUS) and large portions of Canada and Mexico, as shown in Figure 7-1, 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 7-1 provides some basic geographic information
regarding the CMAQ domains and Table 7-2 provides the vertical layer structure for the CMAQ
domain.
Table 7-1 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 7-2)
62 More information available at: fattps://www. epa. gov/e maq.
63Model code for CMAQ v5.4 is available from the Community Modeling and Analysis System (CMAS) at:
http://www.cmascenter.org.
64 The Sixth External Peer Review of the Community Multiscale Air Quality (CMAQ) Modeling System. Available
online at: https://www.epa.gov/sites/production/files/2019-
08/documents/sixth_cmaq_peer_review_comment_report_6.19.19.pdf.
47
-------
Table 7-2 Vertical layer structure for CMAQ domain
Vertical
Layers
Sigma P
Pressure
(mb)
Approximate
Height (m)
35
0.0000
50.00
17,556
34
0.0500
97.50
14,780
33
0.1000
145.00
12,822
32
0.1500
192.50
11,282
31
0.2000
240.00
10,002
30
0.2500
287.50
8,901
29
0.3000
335.00
7,932
28
0.3500
382.50
7,064
27
0.4000
430.00
6,275
26
0.4500
477.50
5,553
25
0.5000
525.00
4,885
24
0.5500
572.50
4,264
23
0.6000
620.00
3,683
22
0.6500
667.50
3,136
21
0.7000
715.00
2,619
20
0.7400
753.00
2,226
19
0.7700
781.50
1,941
18
0.8000
810.00
1,665
17
0.8200
829.00
1,485
16
0.8400
848.00
1,308
15
0.8600
867.00
1,134
14
0.8800
886.00
964
13
0.9000
905.00
797
12
0.9100
914.50
714
11
0.9200
924.00
632
10
0.9300
933.50
551
9
0.9400
943.00
470
8
0.9500
952.50
390
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
48
-------
12US2 domain
x,y origin: -2412000i
col: 396 row:246 j
Figure 7-1 Map of the CMAQ 12 km modeling domain (noted by the purple box)
7.3 CMAQ Inputs
The key inputs to the CMAQ model include emissions from anthropogenic and biogenic
sources, meteorological data, and initial and boundary conditions.
The emissions inputs are summarized in earlier sections of this document.
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.65,66 The WRF
Skamarock, W.C.. et al. (2008) A Description of the Advanced Research WRF Version 3.
https://opensky.ucar.edU/islandora/object/technotes:500.
68 USEPA (2019). Meteorological Model Performance for Annual 2016 Simulation WRF v3.8
https://nepis.epa. gov/Exe/ZvPDF.cgi/P100YD39.PDF?Dockey=P 100YD39.PDF.
49
-------
Model is a state-of-the-science mesoscale numerical weather prediction system developed for
both operational forecasting and atmospheric research applications.67 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.68
The boundary and initial species concentrations were provided by a northern hemispheric
CMAQ modeling platform for the year 2016.69'70 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.71
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.
7.4 CMAQ Model Performance Evaluation
An operational model performance evaluation for ozone, PM2.5 and its related speciated
components, specific air toxics (i.e., formaldehyde, acetaldehyde, and benzene), as well as nitrate
and sulfate deposition were conducted using 2016 state/local monitoring sites data in order to
estimate the ability of the CMAQ modeling system to replicate the base year concentrations for
the 12 km CONUS domain (Section 7.2, Figure 7-1). Included in this evaluation are statistical
measures of model versus observed data that were paired in space and time on a daily or weekly
basis, depending on the sampling frequency of each network (i.e., measured data). For certain
time periods with missing ozone, PM2.5, air toxic, and nitrate and sulfate deposition observations
we excluded the CMAQ predictions from those time periods in our calculations. It should be
noted when pairing 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 are defined in Section 7.4.2). Statistics were calculated for individual monitoring sites
and for each of the nine National Oceanic and Atmospheric Administration (NOAA) climate
67 https://www. mmm.ticar.edu/models/wrf.
68 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/.
69 Henderson, B., et al. (2018) Hemispheric-CMAQ Application and Evaluation for 2016, Presented at 2019 CMAS
Conference, available https://cmascenter.Org/conference//20.l.8/slides/0850 henderson hemispheric-
cmaq application 2018.pptx.
70 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/10.5194/a -20.1.7.
71 USEPA (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.
50
-------
regions of the 12-km U.S. modeling domain (Figure 7-2).72 The regions include the Northeast,
Ohio Valley, Upper Midwest, Southeast, South, Southwest, Northern Rockies, Northwest and
West73'74 as were originally identified in Karl and Koss (1984).75 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 April through September ozone season.76 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.
72 NOAA, National Centers for Environmental Information scientists have identified nine climatically consistent
regions within the contiguous U.S., http://www.ncdc.noaa.gov/monitoring-references/maps/us-climate-regions.php.
73 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.
74 Note most monitoring sites in the West region are located in California (see Figure 7-2), therefore statistics for the
West will be mostly representative of California ozone air quality.
75 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.
76 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.
51
-------
U.S. Climate Regions
Figure 7-2 NO A A Nine Climate Regions (source: http://www.ncdc.noaa.gov/momtoring-
references/maps/us-climate-regions.php#references)
7.4.1 Monitoring Networks
The model evaluation for ozone was based upon comparisons of model predicted 8-hour daily
maximum concentrations to the corresponding ambient measurements for 2016 at monitoring
sites in the EPA Air Quality System (AQS) and the Clean Air Status and Trends Network
(CASTNet). The observed ozone data were measured and reported on an hourly basis. The PM2.5
evaluation focuses on concentrations of PM2.5 total mass and its components including sulfate
(SO4), nitrate (NO3), total nitrate (TNO3), ammonium (NH4), elemental carbon (EC), and organic
carbon (OC) as well as wet deposition for nitrate and sulfate. The PM2.5 performance statistics
were calculated for each season (e.g., "winter" is defined as December, January, and February).
PM2.5 ambient measurements for 2016 were obtained from the following networks: Chemical
Speciation Network (CSN), Interagency Monitoring of PROtected Visual Environments
(IMPROVE), Clean Air Status and Trends Network (CASTNet), and National Acid Deposition
Program/National Trends (NADP/NTN). NADP/NTN collects and reports wet deposition
measurements as weekly average data. The pollutant species included in the evaluation for each
monitoring network are listed in Table 7-3. For PM2.5 species that are measured by more than
one network, we calculated separate sets of statistics for each network. The CSN and IMPROVE
networks provide 24-hour average concentrations on a 1 in every 3-day, or 1 in every 6-day
sampling cycle. The PM2.5 species data at CASTNet sites are weekly integrated samples. In this
analysis we use the term "urban sites" to refer to CSN sites; "suburban/rural sites" to refer to
CASTNet sites; and "rural sites" to refer to IMPROVE sites.
52
-------
Table 7-3 PM2.5 monitoring networks and pollutants species included in the CMAQ performance
evaluation
Ambient
Monitoring
Networks
Particulate
Species
Wet Deposition
Species
PM2 5 Mass
so4
N03
TN03a
EC
oc
nh4
S04
N03
IMPROVE
X
X
X
X
X
CASTNet
X
X
X
CSN
X
X
X
X
X
X
NADP
X
X
a TNO3 = (N03 + HNCb)
The air toxics evaluation focuses on specific species relevant to this rulemaking, i.e.,
formaldehyde, acetaldehyde, and benzene. Similar to the PM2.5 evaluation, the air toxics
performance statistics were calculated for each season to estimate the ability of the CMAQ
modeling system to replicate the base year concentrations for the 12 km continental U.S. domain.
Toxic measurements for 2016 were obtained from the air toxics archive,
https://www.epa.gov/amtic/amtic-air-toxics-data-ambient-monitoring-archive. While most of the
data in the archive are from the AQS database including the National Air Toxics Trends Stations
(NATTS), additional data (e.g., special studies) are included in the archive but not reported in the
AQS.
7.4.2 Model Performance Statistics
The Atmospheric Model Evaluation Tool (AMET) was used to conduct the evaluation
described in this document.77 There are various statistical metrics available and used by the
science community for model performance evaluation. For this evaluation of the 2016 CMAQ
modeling platform, we have selected the mean bias, mean error, normalized mean bias, and
normalized mean error to characterize model performance, which are consistent with the
recommendations in Simon et al. (2012)78 and the photochemical air quality modeling
guidance.79
Mean bias (MB) is the average difference in the predicted and observed values, calculated as
the sum of the difference (predicted-observed) divided by the total number of replicates (n). MB
is given in units of ppb and is defined as:
77 Appel, K.W., Gilliam, R.C., Davis, N., Zubrow, A., and Howard, S.C.: Overview of the Atmospheric Model
Evaluation Tool (AMET) vl.l for evaluating meteorological and air quality models, Environ. Modell. Softw.,26, 4,
434-443, 2011. (http://www.cmascenter.org/).
78 Simon, H., Baker, K., Phillips, S., 2012: Compilation and interpretation of photochemical model performance
statistics published between 2006 and 2012. Atmospheric Environment 61, 124-139.
79 U.S. Environmental Protection Agency (US EPA), Modeling Guidance for Demonstrating Attainment of Air
Quality Goals for Ozone, PM25, and Regional Haze. November 2018, U.S. EPA, Research Triangle Park, NC,
27711, 454/R-18-009, 205pp. https://www.epa.gov/sites/default/files/2020-10/documents/o3-pm-rh-
modeling_guidance-2018 .pdf.
53
-------
MB = ~Hi(P ~ 0) , where P = predicted and O = observed concentrations
Mean error (ME) calculates the absolute value of the difference (predicted - observed)
divided by the total number of replicates (n). ME is given in units of ppb and is defined as:
ME = iฃI|P-0|
Normalized mean bias (NMB) is used to facilitate a range of concentration magnitudes. This
statistic normalizes the difference (predicted - observed) by the sum of observed values. NMB is
a useful model performance indicator because it avoids over-inflating the observed range of
values, especially at low concentrations. NMB is given in percentage units and is defined as:
i(p-o)
NMB = -J *100
n
I(O)
1
Normalized mean error (NME) is similar to NMB, in that the performance statistic is a
normalization of the mean error. NME is calculated as the absolute value of the difference
(predicted - observed) over the sum of observed values. NME is given in percentage units and is
defined as:
n
fJ\p-o\
NME = ^ *100
n
I(o)
1
The "acceptability" of model performance was judged by comparing our CMAQ 2016
performance results to the range of performance found in recent regional ozone and PM2.5 model
54
-------
applications.80'81'82'83'84'85'86'87'88'89 These other modeling studies represent a wide range of
modeling analyses that cover various models, model configurations, domains, years and/or
episodes, chemical mechanisms, and aerosol modules. Overall, the ozone and PM2.5 model
performance results for the 2016 CMAQ simulations are within the range found in other recent
peer-reviewed and regulatory applications. The model performance results, as described in this
document, demonstrate that our applications of CMAQ using this 2016 modeling platform
provide a scientifically credible approach for assessing ozone and PM2.5 concentrations for the
purposes of this final rulemaking.
7.4.3 Evaluation for 8-hour Daily Maximum Ozone
The 8-hour ozone model performance bias and error statistics for each climate region, for
each season defined above, and for each monitor network (AQS and CASTNet) are provided in
Table 7-4. As indicated by the statistics in Table 7-4, bias and error for 8-hour daily maximum
ozone are low in each climate region. Spatial plots of the MB, ME, NMB, and NME for
individual monitors are shown in Figure 7-3 through Figure 7-6. The statistics shown in these
figures were calculated over the ozone season using paired data on days with observed 8-hour
ozone > 60 ppb. Figure 7-3 shows MB for 8-hour ozone > 60 ppb during the ozone season in the
range of ฑ15 ppb at the majority of ozone AQS and CASTNet measurement sites. At both AQS
and CASTNet sites, NMB is within the range of ฑ20 percent (Figure 7-5). ME for 8-hour
80 National Research Council (NRC), 2002. Estimating the Public Health Benefits of Proposed Air Pollution Regulations,
Washington, DC: National Academies Press.
81 Appel, K.W., Roselle, S.J., Gilliam, R.C., and Pleim, J.E, 2010: Sensitivity of the Community Multiscale Air Quality
(CMAQ) model v4.7 results for the eastern United States to MM5 and WRF meteorological drivers. Geoscientific Model
Development, 3, 169-188.
82 Foley, K.M., Roselle, S.J., Appel, K.W., Bhave, P.V., Pleim, J.E., Otte, T.L., Mathur, R., Sarwar, G., Young, J.O.,
Gilliam, R.C., Nolte, C.G., Kelly, J.T., Gilliland, A.B., and Bash, J.O., 2010: Incremental testing of the Community
multiscale air quality (CMAQ) modeling system version 4.7. Geoscientific Model Development, 3, 205-226.
83 Hogrefe, G., Civeroio, K.L., Hao, W., Ku, J-Y., Zalewsky, E.E., and Sistla, G., Rethinking the Assessment of
Photochemical Modeling Systems in Air Quality Planning Applications. Air & Waste Management Assoc.,
58:1086-1099, 2008.
84 Phillips, S., K. Wang, C. Jang, N. Possiel, M. Strum, T. Fox, 2007. Evaluation of 2002 Multi-pollutant Platform: ^'r
Toxics, Ozone, and Particulate Matter, 7th Annual CMAS Conference, Chapel Hill, NC, October 6-8, 2008.
(http ://www. cmascenter. org/conference/2008/agenda. cfm).
85 Simon, H., Baker, K.R., and Phillips, S., 2012. Compilation and interpretation of photochemical model performance
statistics published between 2006 and 2012. Atmospheric Environment 61, 124-139.
http://dx.doi.Org/10.1016/j.atinosenv.2012.07.012.
86 Tesche, T.W., Morris, R., Tonnesen, G., McNally, D., Boylan, J., Brewer, P., 2006. CMAQ/CAMx annual 2002
performance evaluation over the eastern United States. Atmospheric Environment 40, 4906-4919.
87 U.S. Environmental Protection Agency; Technical Support Document for the Final Clean Air Interstate Rule: Air
Quality Modeling; Office of Air Quality Planning and Standards; RTP, NC; March 2005 (CAIR Docket
OAR-2005-0053-2149).
88 U.S. Environmental Protection Agency, Proposal to Designate an Emissions Control Area for Nitrogen Oxides,
Shttps://19january2017snapshot.epa.gov/sites/production/files/2016-09/documents/420r09007.pdfent. EPA-420-R-007,
329pp., 2009. (https://nepis.epa. gov/Exe/ZyPDF.cgi/P1003E8M.PDF?Dockev=P 1003E8M.PDF).
89 U.S. Environmental Protection Agency, 2010, Renewable Fuel Standard Program (RFS2) Regulatory Impact
Analysis EPA-420-R-10-006. February 2010. Sections 3.4.2.1.2 and 3.4.3.3. Docket EPA-HQ-OAR-2009-0472-11332.
55
-------
maximum ozone > 60 ppb, as seen in Figure 7-4, is 20 ppb or less at most of the sites across the
modeling domain.
Table 7-4 Daily Maximum 8-hour Ozone Performance Statistics by Climate Region, by Season, and by
Monitoring Network for the 2016 CMAQ Model Simulation
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(PPb)
(PPb)
(%)
(%)
Winter
11,462
-2.4
4.7
-7.3
14.4
AQS
Spring
15,692
-5.8
7.2
-13.1
16.3
Summer
16,686
1.4
6.3
3.2
13.9
Northeast
Fall
13,780
0.8
4.9
2.4
14.1
Winter
1,238
-3.1
4.7
-8.9
13.6
CASTNet
Spring
1,336
-6.4
7.4
-14.2
16.5
Summer
1,315
0.7
5.8
1.7
13.6
Fall
1,306
0.9
4.8
2.7
14.2
Winter
4,178
-0.6
4.4
-1.9
14.5
AQS
Spring
15,498
-3.3
5.9
-7.2
12.9
Summer
20,495
3.7
7.0
8.1
15.4
Ohio Valley
Fall
14,025
2.1
5.7
5.4
13.1
Winter
1,574
-1.1
4.3
-3.4
13.2
CASTNet
Spring
1,600
-4.0
6.1
-8.7
13.1
Summer
1,551
2.9
6.4
6.4
14.6
Fall
1,528
-0.2
4.9
-0.4
12.3
Winter
1,719
-1.1
4.5
-3.6
14.4
AQS
Spring
6,892
-6.0
7.4
-13.3
16.6
Summer
9,742
0.5
5.8
1.2
13.8
Upper
Fall
6,050
2.4
4.6
7.5
14.6
Midwest
Winter
435
-2.2
4.5
-6.7
13.4
CASTNet
Spring
434
-7.5
8.2
-16.7
18.2
Summer
412
-4.6
5.2
-3.8
12.5
Fall
426
0.2
4.3
0.6
13.7
Southeast
AQS
Winter
7,128
-3.4
5.2
-9.5
14.5
Spring
14,569
-3.9
6.0
-8.5
12.9
56
-------
Climate
Region
Monitor
Network
Season
No. of
Obs
MB
(PPb)
ME
(PPb)
NMB
(%)
NME
(%)
Summer
15,845
3.1
5.9
7.9
15.0
Fall
12,583
0.6
4.9
1.6
12.0
Winter
887
-3.9
5.2
-10.4
14.0
CASTNet
Spring
947
-5.6
6.8
-11.7
14.3
Summer
926
2.5
5.8
6.4
14.8
Fall
928
-0.9
5.3
-2.1
12.9
Winter
11,432
-3.1
5.5
-9.2
16.4
AQS
Spring
13,093
-2.7
6.6
-6.3
15.0
Summer
12,829
1.7
6.3
4.3
16.4
South
Fall
12,443
-0.3
5.1
-0.7
13.0
Winter
523
-3.3
5.2
-9.2
14.3
CASTNet
Spring
532
-3.8
6.7
-8.5
14.7
Summer
508
0.3
7.2
0.7
18.5
Fall
528
-0.2
4.7
-0.6
12.1
Winter
9,990
-4.6
6.4
-11.8
16.3
AQS
Spring
11,381
-7.7
8.5
-15.1
16.5
Summer
12,027
-6.7
8.2
-12.4
15.3
Southwest
Fall
11,097
-2.3
4.7
-5.7
11.4
Winter
757
-6.8
7.1
-15.1
15.9
CASTNet
Spring
810
-8.2
8.6
-15.5
16.3
Summer
812
-5.7
6.9
-10.6
12.9
Fall
791
-2.8
4.2
-6.4
9.6
Winter
4,719
-2.8
5.0
13.5
-9.5
AQS
Spring
4,975
-5.3
6.5
-12.2
14.9
Northern
Rockies
Summer
5,054
-2.6
5.3
-5.6
11.4
Fall
4,876
0.1
4.4
0.2
13.0
Winter
666
-3.8
6.1
-9.6
15.6
CASTNet
Spring
696
-7.0
7.7
-15.1
16.5
Summer
693
-3.9
5.6
-8.1
11.6
57
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(PPb)
(PPb)
(%)
(%)
Fall
605
-1.2
4.9
-3.1
13.1
Winter
677
-3.3
6.0
-10.2
18.6
AQS
Spring
1,288
-6.7
8.2
-16.5
20.4
Summer
2,444
-1.8
6.2
-4.7
16.5
Northwest
Fall
1,236
0.6
5.2
2.0
16.5
Winter
30
-3.8
5.0
-10.3
13.5
CASTNet
Spring
-
-
-
-
-
Summer
-
-
-
-
-
Fall
63
-1.3
4.4
-4.2
14.0
Winter
14,539
-4.3
6.3
-12.5
18.3
AQS
Spring
17,191
-7.7
8.4
-16.8
18.3
Summer
18,132
-7.3
9.6
-13.7
18.0
West
Fall
16,211
-4.9
7.1
-11.3
16.5
Winter
506
-3.6
5.3
-9.2
13.5
CASTNet
Spring
519
-7.6
8.0
-15.8
16.6
Summer
526
-10.2
11.0
-16.8
18.2
Fall
530
-5.1
6.3
-10.8
13.5
58
-------
03_8hrmax MB (ppb) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for 20160401 to 20160930
a CASTNET Daily AQS Daily
Figure 7-3 Mean Bias (ppb) of 8-hour daily maximum ozone greater than 60 ppb over the period April-
September 2016 at AQS and CASTNet monitoring sites in the modeling domain
03_Shrmax ME (ppb) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for 20160401 to 20160930
i CASTNET Daily AQS Daily
Figure 7-4 Mean Error (ppb) of 8-hour daily maximum ozone greater than 60 ppb over the period April-
September 2016 at AQS and CASTNet monitoring sites in the modeling domain
59
-------
03_8hrmax NMB (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for 20160401 to 20160930
units = %
coverage limit - 75%
a CASTNET Daily AQS Daily
Figure 7-5 Normalized Mean Bias (%) of 8-hour daily maximum ozone greater than 60 ppb over the
period April-September AQS and CASTNet 2016 at monitoring sites in the modeling domain
Q3_8hrmax NME (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for 20160401 to 20160930
coverage limit = 75%
k Ifif ซ* /
. v f-vkrj-
N & - ^ l _ t . Ay. a .,ฃ4
40
30
20
10
a CASTNET Daily AQS Daily
Figure 7-6 Normalized Mean Error (%) of 8-hour daily maximum ozone greater than 60 ppb over the
period April-September AQS and CASTNet 2016 at monitoring sites in the modeling domain
60
-------
7.4.4 Seasonal Evaluation of PM2.5 Component Species
The evaluation of 2016 model predictions for PM2.5 covers the performance for the individual
PM2.5 component species (i.e., sulfate, nitrate, organic carbon, elemental carbon, and
ammonium). Performance results are provided for each PM2.5 species. As indicated above, for
each species we present tabular summaries of bias and error statistics by climate region for each
season. These statistics are based on the set of observed-predicted pairs of data for the particular
season at monitoring sites within the nine NOAA climate regions. Separate statistics are
provided for each monitoring network, as applicable for the particular species measured. For
sulfate and nitrate we also provide a more refined temporal and spatial analysis of model
performance that includes spatial maps that show the MB, ME, NMB, and NME =by site,
aggregated by season.
7.4.4.1 Seasonal Evaluation for Sulfate
The model performance bias and error statistics for sulfate for each climate region and each
season by monitor network are provided in Table 7-5. Spatial plots of the NMB and NME by
season for individual monitors are shown in Figure 7-7 through Figure 7-22.
Table 7-5 Sulfate Performance Statistics by Climate Region, by Season, and by Monitoring Network for
the 2016 CMAQ Model Simulation
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Winter
431
-0.1
0.2
-18.1
32.6
IMPROVE
Spring
477
-0.1
0.2
-19.1
28.7
Summer
486
-0.2
0.3
-29.7
38.6
Fall
456
-0.1
0.2
-17.9
34.2
Winter
716
-0.1
0.4
-6.5
40.3
Northeast
CSN
Spring
768
-0.0
0.3
-5.2
35.2
Summer
782
-0.3
0.4
-29.5
36.4
Fall
736
-0.0
0.3
-3.3
37.9
Winter
221
-0.3
0.3
-33.2
33.5
CASTNet
Spring
242
-0.3
0.3
-32.9
33.4
Summer
252
-0.4
0.4
-41.4
41.7
Fall
242
-0.3
0.3
-32.6
33.3
Winter
220
-0.3
0.4
-25.2
35.3
Ohio Valley
IMPROVE
Spring
244
-0.4
0.4
38.2
-28.8
Summer
239
-0.6
0.7
-38.4
44.3
Fall
227
-0.4
0.5
-31.5
35.9
61
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Winter
546
-0.3
0.5
-20.1
36.1
CSN
Spring
562
-0.1
0.4
34.6
-4.9
Summer
553
-0.3
0.6
-20.8
36.3
Fall
541
-0.1
0.4
-11.5
32.2
Winter
212
-0.5
0.5
-36.4
36.9
CASTNet
Spring
228
-0.5
0.5
-37.7
38.2
Summer
224
-0.7
0.7
-41.5
41.5
Fall
226
-0.5
0.5
-36.8
36.8
Winter
200
-0.1
0.2
-14.2
30.4
IMPROVE
Spring
208
-0.1
0.2
-12.4
31.6
Summer
210
-0.2
0.3
-31.7
39.5
Fall
215
-0.1
0.2
-18.7
36.9
Winter
326
0.0
0.3
4.2
34.6
Upper
Midwest
CSN
Spring
354
0.1
0.3
12.8
36.9
Summer
314
-0.1
0.4
-7.0
36.8
Fall
310
0.1
0.3
18.0
42.8
Winter
59
-0.3
0.3
-31.1
31.6
CASTNet
Spring
63
-0.2
0.2
-21.3
22.3
Summer
63
-0.3
0.3
-31.7
31.9
Fall
57
-0.2
0.2
-30.0
30.6
Winter
342
-0.2
0.4
-20.1
37.3
IMPROVE
Spring
379
-0.4
0.5
-34.9
37.7
Summer
394
-0.6
0.6
-47.0
48.2
Fall
366
-0.3
0.3
-28.3
32.9
Southeast
Winter
512
-0.0
0.3
-2.5
35.9
CSN
Spring
551
-0.2
0.4
-20.2
31.5
Summer
523
-0.4
0.4
-34.7
39.4
Fall
505
-0.1
0.3
-13.6
27.2
CASTNet
Winter
150
-0.4
0.4
-37.2
38.2
62
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Spring
164
-0.6
0.6
-45.6
45.7
Summer
164
-0.7
0.7
-53.5
53.5
Fall
154
-0.5
0.5
-40.3
40.3
Winter
240
-0.0
0.2
-6.3
32.1
IMPROVE
Spring
273
-0.3
0.4
-28.9
41.5
Summer
252
-0.8
0.8
-52.3
54.2
Fall
264
-0.3
0.4
-29.4
37.2
Winter
326
-0.1
0.5
-5.0
42.5
South
CSN
Spring
351
-0.5
0.7
-31.4
47.8
Summer
336
-0.7
0.8
-42.2
49.5
Fall
329
-0.3
0.5
-21.2
36.6
Winter
92
-0.4
0.4
-34.1
34.8
CASTNet
Spring
102
-0.6
0.6
-45.3
45.5
Summer
96
-1.0
1.0
-56.9
56.9
Fall
102
-0.5
0.5
-39.6
39.7
Winter
910
0.1
0.2
41.2
70.7
IMPROVE
Spring
991
0.1
0.2
35.6
52.1
Summer
985
-0.3
0.3
-45.9
52.7
Fall
962
-0.1
0.2
-24.1
43.7
Winter
246
-0.1
0.4
-10.6
69.3
Southwest
CSN
Spring
255
0.2
0.2
44.6
56.1
Summer
250
-0.3
0.4
-41.6
50.2
Fall
260
-0.0
0.2
-8.6
42.4
Winter
101
-0.1
0.1
24.6
51.2
CASTNet
Spring
115
0.1
0.1
18.8
29.3
Summer
114
-0.3
0.3
-44.8
46.5
Fall
115
-0.1
0.2
-27.2
37.8
Northern
Rockies
IMPROVE
Winter
542
0.1
0.2
16.7
57.2
Spring
573
0.1
0.2
17.6
43.1
63
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Summer
603
-0.0
0.1
-4.3
36.4
Fall
574
0.0
0.1
2.8
41.4
Winter
139
0.2
0.4
42.0
67.3
CSN
Spring
151
0.1
0.3
26.9
51.6
Summer
153
0.0
0.2
6.9
42.2
Fall
136
0.1
0.3
30.5
57.4
Winter
126
-0.1
0.1
-20.0
37.2
CASTNet
Spring
139
-0.0
0.1
-1.0
21.5
Summer
138
-0.1
0.1
-23.7
27.8
Fall
129
-0.1
0.1
-14.2
27.3
Winter
427
0.1
0.1
50.1
79.5
IMPROVE
Spring
505
0.1
0.2
33.2
50.5
Summer
519
-0.0
0.2
-5.7
45.3
Fall
499
0.0
0.1
15.8
59.0
Winter
156
0.2
0.3
72.9
>100
Northwest
CSN
Spring
161
0.3
0.3
64.9
71.3
Summer
166
0.0
0.3
4.7
46.9
Fall
161
0.2
0.3
58.4
83.4
Winter
12
0.0
0.1
22.2
40.8
CASTNet
Spring
13
0.0
0.1
14.0
21.4
Summer
13
-0.0
0.0
-8.9
13.4
Fall
13
-0.0
0.1
-5.0
25.6
Winter
565
0.1
0.2
64.4
97.0
IMPROVE
Spring
608
0.0
0.3
8.1
51.4
Summer
603
-0.2
0.3
-33.9
47.0
West
Fall
576
-0.1
0.2
-13.8
46.1
Winter
340
0.1
0.3
18.4
62.3
CSN
Spring
352
-0.1
0.4
-12.6
45.3
Summer
349
-0.7
0.7
-47.0
51.1
64
-------
Climate
Region
Monitor
Network
Season
No. of
Obs
ll
ME
(ug/m3)
NMB
(%)
NME
(%)
Fall
330
-0.2
0.4
-25.4
43.7
CASTNet
Winter
69
0.0
0.2
16.2
58.1
Spring
73
-0.2
0.3
-25.9
39.1
Summer
75
-0.5
0.5
-53.3
54.0
Fall
77
-0.2
0.3
-36.8
43.3
for December to February 2016
IMPROVE ฑ CSN ฆ CASTNET Weekly
Figure 7-7 Mean Bias (fig/m3) of sulfate during winter 2016 at monitoring sites in the modeling domain
65
-------
S04 ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
IMPROVE a CSN ฆ CASTNET Weekly
Figure 7-8 Mean Error (fig/m3) of sulfate during winter 2016 at monitoring sites in the modeling domain
units = %
coverage limit = 75%
> 100
80
60
40
20
0
-20
-40
-60
-80
< -100
12US2 for December to February 2016
S04NMB
IMPROVE a CSN ฆ CASTNET Weekly
Figure 7-9 Normalized Mean Bias (%) of sulfate during winter 2016 at monitoring sites in the modeling
domain
66
-------
Units =%
coverage limit = 75%
-I
>100
-
90
80
70
60
a
50
40
ฆ
30
20
-
m
.
IMPROVE ^ CSN
for December to February 2016
ฆ CASTNET Weekly
S04NME
Figure 7-10 Normalized Mean Error (%) of sulfate (luring winter 2016 at monitoring sites in the
modeling domain
units - ug/m3
coverage limit = 75%
IMPROVE a CSN ฆ CASTNET Weekly
Figure 7-11 Mean Bias (jug/m3) of sulfate during spring 2016 at monitoring sites in the modeling domain
67
S04MB
2016
-------
IMPROVE a CSN
CASTNET Weekly
Figure 7-12 Mean Error (jig/m3) of sulfate during spring 2016 at monitoring sites in the modeling
domain
units - %
coverage limit = 75%
>100
80
60
40
20
0
-20
-40
-60
-80
< -100
IMPROVE a CSN ฆ CASTNET Weekly
Figure 7-13 Normalized Mean Bias (%) of sulfate during spring 2016 at monitoring sites in the modeling
domain
S04
2016
68
-------
S04 NME (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for March to May 2016
coverage limit = 75%
100
CASTNET Weekly
Figure 7-14 Normalized Mean Error (%) of sulfate during spring 2016 at monitoring sites in the
modeling domain
S04 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
- ug/m3
coverage limit = 76%
IMPROVE * CSN ฆ CASTNET Weekly
Figure 7-15 Mean Bias (jig/m3) of sulfate during summer 2016 at monitoring sites in the modeling
domain
69
-------
>2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
Figure 7-16 Mean Error (jig/m3) of sulfate during summer 2016 at monitoring sites in the modeling
domain
S04 ME (ug/m3) for run CMAQ 2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
IMPROVE a CSN ฆ CASTNET Weekly
units - %
coverage limit = 75%
IMPROVE * CSN ฆ CASTNET Weekly
Figure 7-17 Normalized Mean Bias (%) of sulfate (luring summer 2016 at monitoring sites in the
modeling domain
70
S04 NMB
-------
S04 NME (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
CASTNET Weekly
Figure 7-18 Normalized Mean Error (%) of sulfate during summer 2016 at monitoring sites in the
modeling domain
S04 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
- ug/m3
coverage limit = 75%
CASTNET Weekly
Figure 7-19 Mean Bias (fig/m3) of sulfate during fall 2016 at monitoring sites in the modeling domain
71
-------
S04 ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
U**' IV.
u -
units - ug/m3
coverage limit = 75%
CASTNET Weekly
Figure 7-20 Mean Error (fig/m3) of sulfate during fall 2016 at monitoring sites in the modeling domain
SQ4 NMB (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
.. i (P.
coverage limit = 75%
k A 4 ' '? " * *ฆ
j , _ ^ 4 J8
IMPROVE
CSN
CASTNET Weekly
Figure 7-21 Normalized Mean Bias (%) of sulfate during fall 2016 at monitoring sites in the modeling
domain
72
-------
to November 2016
S04 NME
units _ %
coverage limit = 75%
I> 100
90
BO
70
IMPROVE * CSN ฆ CASTNET Weekly
Figure 7-22 Normalized Mean Error (%) of sulfate during fall 2016 at monitoring sites in the modeling
domain
7.4.4.2 Seasonal Evaluation for Nitrate
The model performance bias and error statistics for nitrate for each climate region and season
are provided in Table 7-6. This table includes statistics for both particulate nitrate, as measured
at CSN and IMPROVE sites, and total nitrate (TN0.3=N03+HN03), as measured at CASTNet
sites. Spatial plots of the MB, ME, NMB, and NME by season for individual monitors are shown
in Figure 7-23 through Figure 7-54.
Table 7-6 Nitrate and Total Nitrate Performance Statistics by Climate Region, by Season, and by
Monitoring Network for the 2016 CMAQ Model Simulation
Climate
Monitor
No. of
MB
ME
NMB
NME
Region
Network
Sfitisoii
Obs
(ug/mJ)
(ug/mJ)
(%)
(%)
Winter
431
0.8
0.8
>100
>100
IMPROVE
Spring
477
0.1
0.2
24.8
76.5
(N03)
Summer
486
0.0
0.2
29.0
>100
Fall
456
0.2
0.3
63.0
>100
Northeast
Winter
715
1.2
1.4
73.1
84.5
CSN
Spring
770
0.3
0.6
32.3
67.7
(N03)
Summer
778
-0.1
0.2
-36.1
67.6
Fall
737
0.3
0.5
48.2
81.4
CASTNet
Winter
221
0.5
0.5
31.0
34.4
73
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
(TN03)
Spring
242
0.0
0.3
1.5
25.7
Summer
252
0.1
0.3
11.5
29.2
Fall
242
0.2
0.3
20.6
35.2
Winter
220
-0.1
0.7
-7.4
55.2
IMPROVE
Spring
244
-0.2
0.3
-41.5
62.3
(N03)
Summer
239
-0.1
0.2
-30.5
87.2
Fall
227
-0.1
0.3
-27.6
65.9
Winter
543
0.2
1.1
7.8
46.0
Ohio Valley
CSN
Spring
562
0.1
0.6
9.0
68.5
(N03)
Summer
552
0.0
0.3
1.8
87.2
Fall
538
0.1
0.5
16.4
66.2
Winter
212
-0.1
0.5
-5.0
21.4
CASTNet
Spring
228
-0.2
0.4
-13.2
24.3
(TNO3)
Summer
224
0.1
0.4
9.7
30.6
Fall
226
0.1
0.5
6.3
32.5
Winter
200
-0.2
0.7
-16.4
49.0
IMPROVE
Spring
208
-0.2
0.3
-34.0
59.0
(N03)
Summer
210
-0.0
0.1
-9.7
83.1
Fall
215
-0.1
0.2
-29.2
65.4
Winter
326
0.2
1.0
6.4
40.3
Upper
Midwest
CSN
Spring
354
0.1
0.7
11.2
62.0
(N03)
Summer
313
0.0
0.3
1.3
87.4
Fall
307
0.1
0.4
14.4
57.7
Winter
59
-0.3
0.6
-14.1
23.6
CASTNet
Spring
63
-0.1
0.4
-5.7
29.0
(TNO3)
Summer
63
0.1
0.3
6.6
30.3
Fall
57
-0.0
0.3
-3.1
28.1
Southeast
IMPROVE
Winter
342
0.2
0.4
49.5
84.4
(N03)
Spring
379
-0.1
0.2
-36.9
67.6
74
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Summer
394
-0.1
0.1
-28.2
75.7
Fall
366
-0.0
0.2
-13.5
71.9
Winter
573
0.8
0.9
>100
>100
CSN
Spring
643
-0.0
0.3
-10.2
75.8
(NOs)
Summer
608
-0.1
0.2
-25.3
80.4
Fall
560
0.1
0.3
39.9
96.0
Winter
150
0.2
0.5
13.6
40.1
CASTNet
Spring
164
-0.4
0.5
-32.5
39.4
(TNCb)
Summer
164
-0.1
0.3
-14.6
36.2
Fall
154
-0.0
0.5
-1.5
39.7
Winter
240
0.0
0.6
0.2
62.4
IMPROVE
Spring
273
-0.1
0.2
-34.8
70.2
(N03)
Summer
252
-0.1
0.2
-69.3
85.0
Fall
264
-0.1
0.2
-39.3
67.5
Winter
326
0.2
0.6
27.2
68.2
South
CSN
Spring
349
-0.1
0.2
-29.5
70.8
(N03)
Summer
335
-0.1
0.2
-33.3
81.0
Fall
330
-0.0
0.3
84.0
-36.4
Winter
92
-0.1
0.5
-9.0
28.8
CASTNet
Spring
102
-0.3
0.3
-27.2
29.2
(TNCb)
Summer
96
-0.4
0.5
-32.0
39.0
Fall
102
-0.1
0.3
-7.3
29.9
Winter
910
-0.1
0.2
-46.5
75.7
IMPROVE
Spring
991
-0.1
0.1
-56.7
84.9
(N03)
Summer
985
-0.1
0.1
-93.2
95.7
Southwest
Fall
962
-0.1
0.1
-70.2
84.5
CSN
Winter
247
-1.6
1.8
-64.4
72.3
(N03)
Spring
255
-0.2
0.3
-50.6
66.7
Summer
250
-0.2
0.3
-74.3
97.6
75
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Fall
257
-0.3
0.5
-54.1
81.8
Winter
92
-0.1
0.5
-9.0
28.8
CASTNet
Spring
102
-0.3
0.3
-27.2
29.2
(TN03)
Summer
96
-0.4
0.5
-32.0
39.0
Fall
102
-0.1
0.3
-7.3
29.9
Winter
542
-0.2
0.3
-41.0
69.8
IMPROVE
Spring
573
-0.1
0.1
-41.4
73.7
(NOs)
Summer
603
-0.1
0.1
-76.2
85.8
Fall
574
-0.0
0.1
-24.2
83.9
Winter
139
-0.1
0.7
-9.4
55.8
Northern
Rockies
CSN
Spring
151
-0.1
0.3
-25.6
53.6
(N03)
Summer
153
-0.1
0.1
-48.7
74.7
Fall
135
-0.0
0.2
-4.6
62.8
Winter
126
-0.3
0.3
38.0
47.6
CASTNet
Spring
139
-0.1
0.1
-19.6
29.7
(TNCb)
Summer
138
-0.2
0.2
-24.3
27.8
Fall
129
-0.1
0.1
-13.3
28.7
Winter
427
-0.1
0.3
-19.2
97.6
IMPROVE
Spring
505
0.1
0.2
53.2
>100
(N03)
Summer
519
0.1
0.2
73.0
>100
Fall
499
0.1
0.2
36.7
>100
Winter
157
-0.0
1.1
-3.3
93.5
Northwest
CSN
Spring
161
0.9
1.0
>100
>100
(N03)
Summer
166
1.2
1.3
>100
>100
Fall
161
0.7
0.9
>100
>100
Winter
-
-
-
-
-
CASTNet
Spring
-
-
-
-
-
(TNO3)
Summer
-
-
-
-
-
Fall
-
-
-
-
-
76
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Winter
565
-0.1
0.3
-26.1
60.7
IMPROVE
Spring
608
-0.1
0.2
-25.3
58.0
(NOs)
Summer
603
-0.2
0.3
-62.0
86.5
Fall
576
-0.2
0.3
-47.6
71.5
Winter
341
-7.8
2.0
-53.6
61.3
West
CSN
Spring
352
-0.8
0.9
-50.1
57.7
(NOs)
Summer
348
-0.7
0.8
-58.7
67.0
Fall
332
-1.2
1.4
-61.5
73.0
Winter
69
-0.3
0.4
-32.5
47.3
CASTNet
Spring
73
-0.3
0.4
-35.8
37.9
(TNO3)
Summer
75
-0.7
0.8
-43.2
44.8
Fall
77
-0.4
0.5
-37.2
43.0
77
-------
N03 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
IMPROVE a CSN
Figure 7-23 Mean Bias (jug/m3) for nitrate during winter 2016 at monitoring sites in the modeling domain
IMPROVE a CSN
for December to February 2016
N03 ME
Figure 7-24 Mean Error (fig/m3) for nitrate during winter 2016 at monitoring sites in the modeling
domain
78
-------
TN03 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
CASTNET Weekly
Figure 7-25 Mean Bias (fig/m3) for total nitrate during winter 2016 at monitoring sites in the modeling
domain
TN03 ME
-------
>100
80
60
40
20
0
-20
-40
-60
-80
<-100
IMPROVE a CSN
Figure 7-27 Normalized Mean Bias (%) for nitrate during winter 2016 at monitoring sites in the
modeling domain
coverage limit - 75%
>100
90
80
70
60
50
40
30
20
10
0
Figure 7-28 Normalized Mean Error (%) for nitrate during winter 2016 at monitoring sites in the
modeling domain
IMPROVE a CSN
80
-------
TN03NMB
for December to February 2016
units = %
coverage limit = 75%
CASTNET Weekly
Figure 7-29 Normalized Mean Bias (%) for total nitrate (luring winter 2016 at monitoring sites in the
modeling domain
units =
%
covera
e limit = 75%
F
> 100
80
_
60
40
20
J
-20
-
-40
-
-60
-80
<-100
CASTNET Weekly
Figure 7-30 Normalized Mean Error (%) for total nitrate during winter 2016 at monitoring sites in the
modeling domain
TN03 NMB {%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
81
-------
N03 MB
2016
Is - ug/m3
coverage limit = 75%
1.5
0.5
-0.5
-1.5
: -2
IMPROVE a CSN
Figure 7-31 Mean Bias (jug/m3) for nitrate during spring 2016 at monitoring sites in the modeling domain
N03 ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for March to May 2016
units - ug/mS
coverage limit = 75%
IMPROVE * CSN
Figure 7-32 Mean Error (fig/m3) for nitrate during spring 2016 at monitoring sites in the modeling
domain
82
-------
TN03 MB (ug/m3) for run
2016
units = ug.'m3
coverage limit = 75%
ฆ 2
1.5
1
0.5
0
-0.5
-1
-1.5
I <-2
units = ug.'m3
coverage limit = 75%
>2
1.8
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
CASTNET Weekly
Figure 7-33 Mean Bias (jig/m3) for total nitrate during spring 2016 at monitoring sites in the modeling
domain
Figure 7-34 Mean Error (fig/m3) for total nitrate during spring 2016 at monitoring sites in the modeling
domain
CASTNET Weekly
83
-------
N03NMB
2016
units _ %
coverage limit = 75%
IMPROVE * CSN
Figure 7-35 Normalized Mean Bias (%) for nitrate during spring 2016 at monitoring sites in the
modeling domain
units - %
coverage limit = 75%
IMPROVE a CSN
2016
> 100
90
80
70
60
50
40
30
20
10
0
N03
Figure 7-36 Normalized Mean Error (%) for nitrate during spring 2016 at monitoring sites in the
modeling domain
84
-------
TN03 NMB (%) for run
for March to May 2016
units = %
coverage limit = 75%
> 100
80
60
40
20
0
-20
-40
-60
-80
<-100
units = %
coverage limit = 75%
>100
90
80
70
60
50
40
30
20
10
0
CASTNET Weekly
Figure 7-37 Normalized Mean Bias (%) for total nitrate during spring 2016 at monitoring sites in the
modeling domain
Figure 7-38 Normalized Mean Error (%) for total nitrate during spring 2016 at monitoring sites in the
modeling domain
CASTNET Weekly
85
-------
N03 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
IMPROVE a CSN
Figure 7-39 Mean Bias (jig/m3) for nitrate during summer 2016 at monitoring sites in the modeling
domain
N03 ME (ug/rn3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
IMPROVE a CSN
Figure 7-40 Mean Error (jig/m3) for nitrate during summer 2016 at monitoring sites in the modeling
domain
86
-------
TN03 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
CASTNET Weekly
Figure 7-41 Mean Bias (fig/m3) for total nitrate (luring summer 2016 at monitoring sites in the modeling
domain
TN03 ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
CASTNET Weekly
Figure 7-42 Mean Error (jig/m3) for total nitrate during summer 2016 at monitoring sites in the
modeling domain
87
-------
N03NMB
units _ %
coverage limit = 75%
IMPROVE * CSN
Figure 7-43 Normalized Mean Bias (%) for nitrate during summer 2016 at monitoring sites in the
modeling domain
N03 NME (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
IMPROVE a CSN
Figure 7-44 Normalized Mean Error (%) for nitrate during summer 2016 at monitoring sites in the
modeling domain
88
-------
TN03 NMB (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
CASTNET Weekly
Figure 7-45 Normalized Mean Bias (%) for total nitrate during summer 2016 at monitoring sites in the
modeling domain
TN03 NME
for June to August 2016
units = %
coverage limit = 75%
CASTNET Weekly
Figure 7-46 Normalized Mean Error (%) for total nitrate during summer 2016 at monitoring sites in
the modeling domain
89
-------
N03 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
IMPROVE a CSN
Figure 7-47 Mean Bias (jig/m3) for nitrate during fall 2016 at monitoring sites in the modeling domain
units - ug/m3
coverage limit = 75%
IMPROVE a CSN
12US2 for September to November 2016
N03 ME
Figure 7-48 Mean Error (jig/m3) for nitrate during fall 2016 at monitoring sites in the modeling domain
90
-------
TN03 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
CASTNET Weekly
Figure 7-49 Mean Bias (fig/m3) for total nitrate during fall 2016 at monitoring sites in the modeling
domain
TN03 ME (ug'm3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 lor September to November 2016
CASTNET Weekly
Figure 7-50 Mean Error (fig/m3) for total nitrate during fall 2016 at monitoring sites in the modeling
domain
91
-------
N03 NMB (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
coverage limit = 75%
IMPROVE a CSN
100
Figure 7-51 Normalized Mean Bias (%) for nitrate during fall 2016 at monitoring sites in the modeling
domain
coverage limit = 75%
N03NME
November 2016
> 100
90
80
70
60
50
40
30
20
10
0
IMPROVE a CSN
Figure 7-52 Normalized Mean Error (%) for nitrate (luring fall 2016 at monitoring sites in the modeling
domain
92
-------
TN03 NMB <%) for
for September to November 2016
units = %
coverage limit = 75%
> 100
80
60
40
20
0
-20
-40
-60
-80
<-100
Unite = %
coverage limit = 75%
>100
90
80
70
60
50
40
30
20
10
0
CASTNET Weekly
Figure 7-53 Normalized Mean Bias (%) for total nitrate during fall 2016 at monitoring sites in the
modeling domain
Figure 7-54 Normalized Mean Error (%) for total nitrate during fall 2016 at monitoring sites in the
modeling domain
CASTNET Weekly
93
-------
7.4.4.3 Seasonal Ammonium Performance
The model performance bias and error statistics for ammonium for each climate region and
season are provided in Table 7-7. Spatial plots of the MB, ME, NMB, and NME by season for
individual monitors are shown in Figure 7-55 through Figure 7-70.
Table 7-7 Ammonium Performance Statistics by Climate Region, by Season, and by Monitoring Network
for the 2016 CMAQ Model Simulation
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Winter
718
0.6
0.7
>100
>100
CSN
Spring
770
0.2
0.3
82.6
>100
Summer
782
-0.0
0.1
-5.1
60.2
Northeast
Fall
737
0.2
0.3
77.9
>100
Winter
221
0.1
0.1
12.2
28.0
CASTNet
Spring
242
-0.1
0.1
-26.2
34.7
Summer
252
-0.2
0.2
-45.9
46.2
Fall
242
-0.1
0.1
-23.6
37.0
Winter
547
0.2
0.5
26.6
65.0
CSN
Spring
562
0.1
0.3
34.6
80.3
Summer
554
0.0
0.2
6.4
64.8
Ohio Valley
Fall
541
0.1
0.3
14.8
67.8
Winter
212
-0.2
0.2
-21.4
27.9
CASTNet
Spring
228
-0.2
0.3
-40.3
44.4
Summer
224
-0.2
0.2
-40.4
41.5
Fall
226
-0.2
0.2
-36.4
39.9
Winter
326
0.3
0.5
43.5
66.0
CSN
Spring
354
0.2
0.3
43.2
80.6
Summer
314
0.1
0.2
45.9
86.0
Upper
Midwest
Fall
310
0.2
0.3
80.8
>100
Winter
59
-0.2
0.3
-25.5
30.7
CASTNet
Spring
63
-0.1
0.2
-14.9
38.3
Summer
63
-0.1
0.1
-38.2
39.1
Fall
57
-0.1
0.2
-34.5
38.8
94
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Winter
513
0.3
0.4
95.4
>100
CSN
Spring
551
-0.1
0.2
-28.7
60.9
Summer
524
-0.1
0.2
-41.9
66.3
Southeast
Fall
503
-0.0
0.2
-6.8
67.7
Winter
150
-0.0
0.1
-3.1
28.0
CASTNet
Spring
164
-0.2
0.2
-52.4
53.1
Summer
164
-0.2
0.2
-58.1
58.1
Fall
154
-0.2
0.2
-38.9
42.0
Winter
327
0.2
0.3
45.0
90.8
CSN
Spring
351
-0.1
0.3
-42.2
77.5
Summer
336
-0.1
0.2
-38.7
82.9
South
Fall
331
-0.1
0.2
-21.4
62.3
Winter
92
-0.1
0.2
-16.2
36.6
CASTNet
Spring
102
-0.2
0.2
-51.7
56.1
Summer
96
-0.2
0.2
-56.5
57.6
Fall
102
-0.2
0.2
-41.2
45.6
Winter
247
-0.4
0.5
-56.9
82.8
CSN
Spring
255
-0.0
0.1
-16.4
>100
Summer
250
-0.1
0.1
-56.4
99.2
Southwest
Fall
260
-0.1
0.2
-43.0
>100
Winter
101
-0.1
0.1
-42.6
56.0
CASTNet
Spring
115
-0.0
0.1
-35.9
48.4
Summer
114
-0.1
0.1
-64.5
64.5
Fall
115
-0.1
0.1
-50.9
53.9
Winter
143
0.2
0.3
96.9
>100
Northern
Rockies
CSN
Spring
151
0.1
0.2
92.6
>100
Summer
153
0.1
0.1
>100
>100
Fall
139
0.1
0.1
>100
>100
CASTNet
Winter
126
-0.1
0.1
-52.9
56.2
95
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Spring
139
-0.1
0.1
-45.6
51.3
Summer
138
-0.1
0.1
-60.2
60.2
Fall
129
-0.1
0.1
-50.9
54.3
Winter
157
0.1
0.3
31.5
>100
CSN
Spring
161
0.2
0.2
>100
>100
Summer
166
0.2
0.2
>100
>100
Northwest
Fall
161
0.2
0.2
>100
>100
Winter
12
-0.0
0.0
-15.5
33.7
CASTNet
Spring
13
-0.1
0.1
-66.3
66.5
Summer
13
-0.1
0.1
-81.8
81.8
Fall
13
-0.1
0.1
-68.6
68.7
Winter
341
-0.4
0.6
-44.3
71.7
CSN
Spring
352
-0.2
0.3
-47.3
73.6
Summer
349
-0.2
0.3
-61.0
71.3
West
Fall
332
-0.3
0.4
-56.4
79.7
Winter
69
-0.1
0.1
-34.5
56.4
CASTNet
Spring
73
-0.1
0.1
-56.4
58.2
Summer
75
-0.3
0.3
-81.1
81.1
Fall
77
-0.1
0.1
-59.6
62.5
96
-------
NH4 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
CSN a CASTNET Weekly
Figure 7-55 Mean Bias (jag/m3) of ammonium during winter 2016 at monitoring sites in the modeling
domain
NH4 ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
CSN a CASTNET Weekly
Figure 7-56 Mean Error (fig/m3) of ammonium during winter 2016 at monitoring sites in the modeling
domain
97
-------
NH4 NMB (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12l)S2 for December to February 2016
>100
80
60
40
20
0
| -20
J -40
-60
-80
I <-100
units = %
coverage limit = 75%
>100
90
80
70
60
50
40
30
20
10
0
CSN * CASTNET Weekly
Figure 7-57 Normalized Mean Bias (%) of ammonium during winter 2016 at monitoring sites in the
modeling domain
Figure 7-58 Normalized Mean Error (%) of ammonium during winter 2016 at monitoring sites in the
modeling domain
CSN
a CASTNET Weekly
98
-------
NH4 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for March to May 2016
CSN a CASTNET Weekly
Figure 7-59 Mean Bias (ng/m3) of ammonium during spring 2016 at monitoring sites in the modeling
domain
NH4 ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for March to May 2016
CSN a CASTNET Weekly
Figure 7-60 Mean Error (fig/m3) of ammonium during spring 2016 at monitoring sites in the modeling
domain
99
-------
NH4 NMB (%) far run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 (or March to May 2016
>100
80
60
40
20
0
| -20
J -40
-60
-80
I <-100
units =
14
coverage limit = 75%
E
>100
-
90
_
80
70
60
50
-
40
-
30
-
20
10
_
0
CSN * CASTNET Weekly
Figure 7-61 Normalized Mean Bias (%) of ammonium during spring 2016 at monitoring sites in the
modeling domain
Figure 7-62 Normalized Mean Error (%) of ammonium during spring 2016 at monitoring sites in the
modeling domain
CSN
* CASTNET Weekly
100
-------
NH4 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
CSN a CASTNET Weekly
Figure 7-63 Mean Bias (jag/m3) of ammonium during summer 2016 at monitoring sites in the modeling
domain
NH4 ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
CSN a CASTNET Weekly
Figure 7-64 Mean Error (fig/m3) of ammonium during summer 2016 at monitoring sites in the modeling
domain
101
-------
NH4 NMB (%) for run CMAQ_2Q16gh_cb6r5_ae7nvpoa_12US2 lor June to August 2016
>100
80
60
40
20
0
| -20
J -40
-60
-80
I <-100
units = %
coverage limit = 75%
i
>100
-
90
-
80
70
60
-
50
-
40
-
30
-
20
10
_
0
CSN a CASTNET Weekly
Figure 7-65 Normalized Mean Bias (%) of ammonium during summer 2016 at monitoring sites in the
modeling domain
Figure 7-66 Normalized Mean Error (%) of ammonium during summer 2016 at monitoring sites in the
modeling domain
CSN
* CASTNET Weekly
102
-------
NH4 MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
CSN a CASTNET Weekly
Figure 7-67 Mean Bias (jag/m3) of ammonium during fall 2016 at monitoring sites in the modeling
domain
NH4 ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
CSN a CASTNET Weekly
Figure 7-68 Mean Error (fig/m3) of ammonium during fall 2016 at monitoring sites in the modeling
domain
103
-------
units = %
coverage limit = 75%
>100
80
60
40
20
0
-20
-40
-60
-80
<-100
CSN * CASTNET Weekty
Figure 7-69 Normalized Mean Bias (%) of ammonium during fall 2016 at monitoring sites in the
modeling domain
units = %
coverage limit ฆ 75%
>100
90
80
70
60
50
40
30
20
10
0
Figure 7-70 Normalized Mean Error (%) of ammonium during fall 2016 at monitoring sites in the
modeling domain
CSN
* CASTNET Weekty
104
-------
7.4.4.4 Seasonal Elemental Carbon Performance
The model performance bias and error statistics for elemental carbon for each of climate
region and season are provided in Table 7-8. The statistics show clear at urban and rural sites in
most climate regions. Spatial plots of the MB, ME, NMB, and NME by season for individual
monitors are shown in Figure 7-71 through Figure 7-86.
Table 7-8 Elemental Carbon Performance Statistics by Climate Region, by Season, and by Monitoring
Network for the 2016 CMAQ Model Simulation
Climate
Region
Monitor
Network
Season
No. of
Obs
MB
(ug/m3)
ME
(ug/m3)
NMB
(%)
NME
(%)
Winter
429
0.1
0.1
46.2
60.5
IMPROVE
Spring
478
0.0
0.1
13.4
44.7
Summer
479
-0.0
0.1
-6.5
39.1
Northeast
Fall
456
0.0
0.1
9.5
43.6
Winter
722
0.1
0.4
21.7
56.3
CSN
Spring
785
-0.0
0.3
-3.2
44.7
Summer
788
-0.1
0.2
-13.2
41.1
Fall
780
0.1
0.3
14.8
50.0
Winter
217
0.0
0.1
7.6
46.2
IMPROVE
Spring
242
-0.1
0.1
-23.9
49.9
Summer
241
-0.1
0.1
-36.1
40.1
Ohio Valley
Fall
232
-0.1
0.1
-28.7
38.3
Winter
535
0.1
0.2
17.8
46.8
CSN
Spring
571
-0.1
0.2
-15.1
39.4
Summer
532
-0.1
0.2
-20.0
38.7
Fall
535
-0.0
0.2
-7.0
35.1
Winter
222
0.1
0.1
37.0
53.7
IMPROVE
Spring
239
-0.0
0.1
-17.3
45.2
Upper
Midwest
Summer
236
-0.1
0.1
-30.9
44.4
Fall
243
-0.0
0.1
-12.7
44.0
Winter
334
0.2
0.2
53.8
73.7
CSN
Spring
347
-0.0
0.2
-0.1
48.6
Summer
332
-0.0
0.2
-9.7
46.4
105
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Fall
338
0.0
0.2
7.0
47.5
Winter
398
-0.0
0.1
-8.1
49.0
IMPROVE
Spring
446
-0.2
0.2
-46.8
58.3
Summer
442
-0.1
0.1
-29.9
48.7
Southeast
Fall
422
-0.1
0.1
-31.1
40.8
Winter
436
-0.0
0.2
-3.7
41.5
CSN
Spring
478
-0.1
0.2
-25.9
42.9
Summer
445
-0.0
0.2
-10.3
49.6
Fall
430
-0.1
0.3
-19.8
41.1
Winter
240
-0.0
0.1
-5.2
41.1
IMPROVE
Spring
272
-0.0
0.1
-14.3
52.7
Summer
242
-0.0
0.1
-30.2
42.8
South
Fall
262
-0.1
0.1
-33.5
42.6
Winter
272
-0.0
0.2
-5.1
40.6
CSN
Spring
297
-0.1
0.2
-16.5
38.7
Summer
251
-0.0
0.2
-3.4
52.1
Fall
238
-0.0
0.2
-2.3
45.4
Winter
890
-0.1
0.1
-34.3
58.1
IMPROVE
Spring
981
-0.0
0.1
-2.2
65.6
Summer
962
-0.0
0.1
-22.9
57.1
Southwest
Fall
945
-0.0
0.1
-24.7
58.9
Winter
228
0.0
0.4
3.1
42.1
CSN
Spring
254
0.2
0.2
48.5
61.1
Summer
237
0.1
0.1
25.7
48.9
Fall
240
0.1
0.3
18.7
49.1
Winter
557
0.0
0.0
12.5
75.0
Northern
Rockies
IMPROVE
Spring
594
-0.0
0.1
-17.2
76.3
Summer
616
0.0
0.1
16.1
76.6
Fall
585
-0.0
0.1
-19.6
61.3
106
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Winter
141
-0.0
0.2
-0.7
96.8
CSN
Spring
145
-0.0
0.1
-12.8
58.6
Summer
161
-0.0
0.1
-18.0
45.7
Fall
146
-0.0
0.2
-13.5
70.9
Winter
434
0.0
0.1
54.0
>100
IMPROVE
Spring
505
0.1
0.1
>100
>100
Summer
504
0.1
0.2
79.8
>100
Northwest
Fall
474
0.1
0.2
>100
>100
Winter
140
0.3
0.6
42.1
84.2
CSN
Spring
150
0.7
0.8
>100
>100
Summer
158
1.0
1.0
>100
>100
Fall
155
0.8
1.0
>100
>100
Winter
540
-0.0
0.1
-12.3
62.5
IMPROVE
Spring
600
0.0
0.1
17.5
69.2
Summer
601
-0.0
0.1
-10.1
65.4
West
Fall
565
0.0
0.1
2.2
60.8
Winter
286
0.0
0.5
4.6
42.6
CSN
Spring
294
0.2
0.3
49.9
61.3
Summer
290
0.2
0.2
42.1
54.6
Fall
277
0.2
0.4
36.6
55.1
107
-------
EC MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
IMPROVE * CSN
Figure 7-71 Mean Bias (jig/m33) of elemental carbon during winter 2016 at monitoring sites in the
modeling domain
IMPROVE * CSN
for December to February 2016
EC ME
Figure 7-72 Mean Error (fig/m3) of elemental carbon during winter 2016 at monitoring sites in the
modeling domain
108
-------
EC NMB (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
coverage limit = 75%
IMPROVE * CSN
Figure 7-73 Normalized Mean Bias (%) of elemental carbon during winter 2016 at monitoring sites in
the modeling domain
EC NMB (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
units = %
coverage limit = 75%
IMPROVE * CSN
Figure 7-74 Normalized Mean Error (%) of elemental carbon during winter 2016 at monitoring sites in
the modeling domain
109
-------
EC MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for March to May 2016
IMPROVE a CSN
Figure 7-75 Mean Bias (jig/m3) of elemental carbon during spring 2016 at monitoring sites in the
modeling domain
EC ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for March to May 2016
IMPROVE a CSN
Figure 7-76 Mean Error (fig/m3) of elemental carbon during spring 2016 at monitoring sites in the
modeling domain
110
-------
EC NMB (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for March to May 2016
coverage limit = 75%
IMPROVE * CSN
Figure 7-77 Normalized Mean Bias (%) of elemental carbon during spring 2016 at monitoring sites in
the modeling domain
units - %
coverage limit = 75%
> 100
90
80
70
60
50
40
30
20
10
0
Figure 7-78 Normalized Mean Error (%) of elemental carbon during spring 2016 at monitoring sites in
the modeling domain
IMPROVE a CSN
111
-------
EC MB
units - ug/m3
coverage limit = 75%
ฆ IMPROVE * CSN
Figure 7-79 Mean Bias (jig/m3) of elemental carbon during summer 2016 at monitoring sites in the
modeling domain
EC ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
IMPROVE a CSN
Figure 7-80 Mean Error (fig/m3) of elemental carbon during summer 2016 at monitoring sites in the
modeling domain
112
-------
EC NMB
2016
units _ %
coverage limit = 75%
IMPROVE * CSN
Figure 7-81 Normalized Mean Bias (%) of elemental carbon during summer 2016 at monitoring sites in
the modeling domain
units - %
coverage limit = 76%
EC NME
12US2 for June to August 2016
> 100
90
80
70
60
50
40
30
20
10
0
IMPROVE a CSN
Figure 7-82 Normalized Mean Error (%) of elemental carbon during summer 2016 at monitoring sites
in the modeling domain
113
-------
EC MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
IMPROVE a CSN
Figure 7-83 Mean Bias (fig/m3) of elemental carbon during fall 2016 at monitoring sites in the modeling
domain
EC ME (ug/m3) for run CMAQ 2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
IMPROVE a CSN
Figure 7-84 Mean Error (fig/m3) of elemental carbon during fall 2016 at monitoring sites in the modeling
domain
114
-------
EC NMB
for September to November 2016
units _ %
coverage limit = 75%
ฆ IMPROVE * CSN
Figure 7-85 Normalized Mean Bias (%) of elemental earbon during fall 2016 at monitoring sites in the
modeling domain
units - %
coverage limit = 75%
EC NME
November 2016
> 100
90
80
70
60
50
40
30
20
10
0
IMPROVE a CSN
Figure 7-86 Normalized Mean Error (%) of elemental carbon during fall 2016 at monitoring sites in the
modeling domain
115
-------
7.4.4.5 Seasonal Organic Carbon Performance
The model performance bias and error statistics for organic carbon for each climate region
and season are provided in Table 7-9. The statistics in this table indicate a tendency for the
modeling platform to observed organic carbon concentrations during most seasons and climate
regions except in the Northern Rockies and the Western U.S. Spatial plots of the MB, ME,
NMB, and NME by season for individual monitors are shown in Figure 7-87 through Figure
7-102.
Table 7-9 Organic Carbon Performance Statistics by Climate Region, by Season, and by Monitoring
Network for the 2016 CMAQ Model Simulation
Climate
Region
Monitor
Network
Season
No. of
Obs
MB
(ug/m3)
ME
(ug/m3)
NMB
(%)
NME
(%)
Winter
427
-0.1
0.3
-12.7
34.2
IMPROVE
Spring
477
-0.3
0.3
-40.8
46.3
Summer
482
-0.5
0.6
-43.9
49.1
Northeast
Fall
459
-0.2
0.4
-26.9
40.4
Winter
722
-0.1
0.8
-7.0
48.1
CSN
Spring
785
-0.4
0.7
-24.5
42.4
Summer
788
-0.4
0.7
-20.6
38.0
Fall
780
-0.1
0.8
-7.99
40.7
Winter
217
-0.4
0.6
-37.7
60.5
IMPROVE
Spring
242
-0.5
0.7
-42.0
64.3
Summer
242
-0.1
0.6
-8.5
43.3
Ohio Valley
Fall
232
-0.6
0.9
-31.8
50.0
Winter
535
-0.6
0.7
-35.0
42.0
CSN
Spring
571
-0.7
0.7
-41.8
45.9
Summer
431
-0.2
0.8
-12.9
41.3
Fall
532
-0.8
1.0
-32.8
42.3
Winter
226
-0.1
0.2
-20.8
40.8
IMPROVE
Spring
238
-0.5
0.6
-58.1
63.8
Upper
Midwest
Summer
237
-0.6
0.7
-51.6
57.8
Fall
243
-0.4
0.4
-42.4
49.8
CSN
Winter
333
-0.1
0.5
-7.3
42.3
Spring
347
-0.6
0.8
-40.1
51.5
116
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Summer
331
-0.6
0.7
-34.3
42.8
Fall
337
-0.4
0.6
-29.4
39.6
Winter
398
-0.5
0.7
-46.2
63.1
IMPROVE
Spring
447
-5.5
5.6
-88.7
90.1
Summer
445
-0.4
0.7
-29.0
49.5
Southeast
Fall
423
-0.8
1.1
-40.9
54.8
Winter
436
-0.9
0.9
-42.5
46.1
CSN
Spring
478
-0.8
0.9
-40.4
44.4
Summer
445
-0.1
0.7
-5.8
34.2
Fall
430
-0.9
1.4
-30.4
47.9
Winter
239
-0.4
0.5
-50.4
55.6
IMPROVE
Spring
272
-0.7
0.7
-66.1
68.1
Summer
250
-0.5
0.6
-42.4
49.8
South
Fall
264
-0.6
0.6
-49.4
54.8
Winter
272
-0.9
1.0
-46.2
50.6
CSN
Spring
297
-0.7
0.7
-46.0
51.4
Summer
251
-0.3
0.7
-17.0
47.7
Fall
237
-0.6
0.9
-28.5
44.9
Winter
881
-0.5
0.5
-71.7
73.2
IMPROVE
Spring
981
-0.3
0.3
-70.2
74.4
Summer
978
-0.7
0.7
-76.8
80.0
Southwest
Fall
964
-0.4
0.5
-67.7
74.3
Winter
228
-1.2
1.4
-46.3
56.8
CSN
Spring
254
-0.4
0.6
-38.8
54.5
Summer
237
-0.8
0.9
-59.9
60.9
Fall
240
-0.7
0.8
-42.4
50.1
Northern
Rockies
Winter
549
-0.2
0.2
-50.4
61.2
IMPROVE
Spring
590
-0.5
0.5
-77.0
79.6
Summer
631
-0.9
0.9
-71.7
77.2
117
-------
Climate
Monitor
Season
No. of
MB
ME
NMB
NME
Region
Network
Obs
(ug/m3)
(ug/m3)
(%)
(%)
Fall
600
-0.4
0.5
-68.7
72.8
Winter
140
-0.5
0.6
-54.9
65.6
CSN
Spring
145
-0.6
0.6
-68.2
71.1
Summer
161
-1.1
1.1
-72.1
72.1
Fall
146
-0.6
0.7
-61.7
64.4
Winter
407
-0.1
0.3
-25.9
78.0
IMPROVE
Spring
497
-0.2
0.4
-33.9
81.7
Summer
494
-0.2
0.6
-29.3
76.2
Northwest
Fall
516
-0.6
1.0
-50.8
79.7
Winter
139
-0.5
1.4
-21.9
57.9
CSN
Spring
150
0.6
1.2
42.4
86.3
Summer
155
0.4
1.3
20.5
67.5
Fall
158
0.8
1.5
55.6
100
Winter
552
-0.4
0.4
-63.4
66.8
IMPROVE
Spring
599
-0.4
0.4
-68.7
70.9
Summer
608
-1.2
1.2
-69.3
71.2
West
Fall
574
-0.7
0.7
-63.0
66.0
Winter
285
-1.8
1.9
-48.2
50.5
CSN
Spring
294
-0.6
0.7
-41.0
44.8
Summer
289
-1.5
1.5
-60.8
61.1
Fall
277
-1.3
1.4
-44.6
48.8
118
-------
OC MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
IMPROVE a CSN
Figure 7-87 Mean Bias (jig/m3) of organic carbon during winter 2016 at monitoring sites in the modeling
domain
IMPROVE * CSN
for December to February 2016
OC ME
Figure 7-88 Mean Error (fig/m3) of organic carbon during winter 2016 at monitoring sites in the
modeling domain
119
-------
OC NMB (%) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
IMPROVE a CSN
Figure 7-89 Normalized Mean Bias (%) of organic carbon during winter 2016 at monitoring sites in the
modeling domain
units = %
coverage limit = 75%
>100
90
80
70
60
50
40
30
20
10
0
IMPROVE a CSN
12US2 for December to February 2016
OC NME (%) for run
Figure 7-90 Normalized Mean Error (%) of organic carbon during winter 2016 at monitoring sites in
the modeling domain
120
-------
OC MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
IMPROVE a CSN
Figure 7-91 Mean Bias (fig/m3) of organic carbon (luring spring 2016 at monitoring sites in the modeling
domain
OC ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for December to February 2016
IMPROVE ^ CSN
Figure 7-92 Mean Error (ng/m3) of organic carbon during spring 2016 at monitoring sites in the
modeling domain
121
-------
OC NMB
2016
units - %
coverage limit = 75%
> 100
90
80
70
60
50
40
30
20
10
0
> 100
80
60
40
20
0
-20
-40
-60
-80
< -100
IMPROVE a CSN
Figure 7-93 Normalized Mean Bias (%) of organic carbon during spring 2016 at monitoring sites in the
modeling domain
Figure 7-94 Normalized Mean Error (%) of organic carbon during spring 2016 at monitoring sites in
the modeling domain
IMPROVE a CSN
122
-------
OC MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
IMPROVE a CSN
Figure 7-95 Mean Bias (fig/m3) of organie carbon during summer 2016 at monitoring sites in the
modeling domain
OC ME (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
IMPROVE a CSN
Figure 7-96 Mean Error (fig/m3) of organic carbon during summer 2016 at monitoring sites in the
modeling domain
123
-------
OC NMB (%) for run CMAQ 2016gh_cb6r5_ae7nvpoa_12US2 for June to August 2016
IMPROVE * CSN
Figure 7-97 Normalized Mean Bias (%) of organic carbon during summer 2016 at monitoring sites in
the modeling domain
units - %
coverage limit = 75%
IMPROVE a CSN
> 100
90
80
70
60
50
40
30
20
10
0
OCNME
Figure 7-98 Normalized Mean Error (%) of organic carbon during summer 2016 at monitoring sites in
the modeling domain
124
-------
OC MB (ug/m3) for run CMAQ_2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
IMPROVE a CSN
Figure 7-99 Mean Bias (jig/m3) of organic carbon during fall 2016 at monitoring sites in the modeling
domain
OC ME (ug/m3) for run CMAQ 2016gh_cb6r5_ae7nvpoa_12US2 for September to November 2016
IMPROVE a CSN
Figure 7-100 Mean Error (fig/m3) of organic carbon during fall 2016 at monitoring sites in the modeling
domain
125
-------
OCNMB
for September to November 2016
units _ %
coverage limit = 75%
IMPROVE * CSN
Figure 7-101 Normalized Mean Bias (%) of organic carbon during fall 2016 at monitoring sites in the
modeling domain
coverage limit = 75%
> 100
90
80
50
40
30
20
,ฐ
'o
IMPROVE * CSN
Figure 7-102 Normalized Mean Error (%) of organic carbon during fall 2016 at monitoring sites in the
modeling domain
7.4.5 Seasonal Hazardous Air Pollutants Performance
A seasonal operational model performance evaluation for specific hazardous air pollutants
(i.e., formaldehyde, acetaldehyde, and benzene) was conducted in order to estimate the ability of
November 2016
OC NME
126
-------
the CMAQ modeling system to replicate the base year concentrations for the 12 km CONUS
domain. The seasonal model performance results for the 12 km modeling domain are presented
below in Table 7-10. Toxic measurements included in the evaluation were taken from the 2016
air toxics archive, https://www.epa.gov/amtic/amtic-air-toxics-data-ambient-monitoring-archive.
While most of the data in the archive are from the AQS database including the National Air
Toxics Trends Stations (NATTS), additional data (e.g., special studies) are included in the
archive but not reported in the AQS. Similar to PM2.5 and ozone, the evaluation principally
consists of statistical assessments of model versus observed data that were paired in time and
space on a daily basis.
Model predictions of annual formaldehyde, acetaldehyde, and benzene showed relatively
small to moderate bias and error percentages when compared to observations. Model
performance for HAPs is not as good as model performance for ozone and PM2.5. Technical
issues in the HAPs data consist of (1) uncertainties in monitoring methods; (2) limited
measurements in time/space to characterize ambient concentrations ("local in nature"); (3)
ambient data below method detection limit (MDL); (4) commensurability issues between
measurements and model predictions; (5) emissions and science uncertainty issues may also
affect model performance; and (6) limited data for estimating intercontinental transport that
effects the estimation of boundary conditions (i.e., boundary estimates for some species are much
higher than predicted values inside the domain).
As with the national, annual PM2.5 and ozone CMAQ modeling, the "acceptability" of model
performance was judged by comparing our CMAQ 2016 performance results to the limited
performance found in recent regional multi-pollutant model applications.90'91 Overall, the mean
bias and error (MB and ME), as well as the normalized mean bias and error (NMB and NME)
statistics shown below in Table 7-10 indicate that CMAQ-predicted 2016 toxics (i.e., observation
vs. model predictions) are within the range of recent regional modeling applications.
Table 7-10 Hazardous Air Toxics Performance Statistics by Season for the 2016 CMAQ Model
Simulation
Air Toxic Species
Season
No. of
Obs.
MB
(ug/m3)
ME
(ug/m3)
NMB
(%)
NME
(%)
Formaldehyde
Winter
1,184
-1.0
1.1
-61.0
63.1
Spring
1,914
-1.3
1.3
-60.1
61.6
Summer
2,318
-1.4
1.5
-43.4
47.7
Fall
1,886
-1.1
1.2
-48.0
53.3
Acetaldehyde
Winter
1,818
-0.4
0.4
-52.6
57.1
Spring
1,920
-0.5
0.5
-57.5
60.6
90 Phillips, S., K. Wang, C. Jang, N. Possiel, M. Strum, T. Fox, 2007: Evaluation of 2002 Multi-pollutant Platform:
Air Toxics, Ozone, and Particulate Matter, 7th Annual CMAS Conference, Chapel Hill, NC, October 6-8, 2008.
91 Wesson, K., N. Fann, M. Morris, T. Fox, and B. Hubbell 2010: A Multi-pollutant, Risk-based Approach to the
Air Quality Management: Case Study for Detroit, Atmospheric Pollution Research, 1 (4) (2010), pp. 296-
304, 10.5094/APR.2010.037.
127
-------
Summer
2,316
-0.3
0.5
-31.1
49.9
Fall
1,870
-0.4
0.5
-43.9
53.1
Benzene
Winter
3,991
-0.0
0.1
-18.0
42.0
Spring
4,479
-0.1
0.1
-31.9
47.7
Summer
5,907
-0.0
0.1
-21.2
54.6
Fall
4,572
-0.1
0.1
-29.2
48.5
7.4.6 Seasonal Nitrate and Sulfate Deposition Performance
Seasonal nitrate and sulfate wet deposition performance statistics for the 12 km Continental
U.S. domain are provided in Table 7-11 and Table 7-12. The model predictions for seasonal
nitrate deposition generally show under prediction for the continental U.S. NADP sites (NMB
values range from -0.6% to -83.7%). Sulfate deposition performance shows similar under
predictions (NMB values range from -1.3% to 81.7%). The errors for both annual nitrate and
sulfate are relatively moderate with most values ranging from 33% to 92% which reflect scatter
in the model predictions versus observation comparison.
Table 7-11 Nitrate Wet Deposition Performance Statistics by Climate Region, by Season, and by
Climate
Region
Season
No. of
Obs
!i
ME
(ug/m3)
NMB
(%)
NME
(%)
Northeast
Winter
600
-0.1
0.1
-41.0
54.1
Spring
649
-0.0
0.1
-12.1
44.5
Summer
681
-0.0
0.1
-21.7
51.1
Fall
679
-0.0
0.1
-0.6
48.9
Ohio Valley
Winter
297
-0.0
0.1
-5.2
49.7
Spring
300
-0.0
0.1
-9.9
33.9
Summer
309
-0.1
0.1
-32.1
51.8
Fall
288
0.0
0.1
5.1
52.0
Upper
Midwest
Winter
275
-0.0
0.1
-40.5
63.9
Spring
277
-0.0
0.1
-28.1
46.7
Summer
292
-0.1
0.1
-34.6
49.0
Fall
301
-0.0
0.1
-17.7
47.8
Southeast
Winter
359
-0.0
0.0
-3.8
51.3
Spring
376
-0.0
0.1
-14.8
45.8
128
-------
Climate
Region
Season
No. of
Obs
!i
ME
(ug/m3)
NMB
(%)
NME
(%)
Summer
413
-0.1
0.1
-32.7
50.7
Fall
385
-0.0
0.0
-13.0
59.9
South
Winter
236
0.0
0.0
9.3
56.2
Spring
263
-0.0
0.1
-15.6
44.9
Summer
281
-0.1
0.1
-39.5
56.0
Fall
280
-0.0
0.0
-20.9
53.7
Southwest
Winter
300
-0.0
0.0
-78.5
83.1
Spring
322
-0.0
0.1
-70.8
81.6
Summer
292
-0.0
0.1
-39.7
56.9
Fall
334
-0.0
0.0
-47.6
72.4
Northern
Rockies
Winter
216
-0.0
0.0
-68.7
87.3
Spring
251
-0.0
0.1
-43.9
68.0
Summer
226
-0.0
0.1
-41.2
52.7
Fall
237
-0.0
0.0
-37.1
63.6
Northwest
Winter
121
-0.0
0.0
-0.5
51.7
Spring
141
-0.0
0.0
-7.0
59.3
Summer
138
-0.0
0.0
-1.4
73.1
Fall
145
0.0
0.0
22.7
66.1
West
Winter
151
-0.0
0.0
-27.1
57.0
Spring
151
0.0
0.0
7.3
79.0
Summer
161
-0.0
0.0
-83.7
93.1
Fall
160
-0.0
0.0
-15.0
76.2
Table 7-12 Sulfate Wet Deposition Performance Statistics by Climate Region, by Season, and by
Climate
Region
Season
No. of
Obs
MB
(ug/m3)
!i
NMB
(%)
NME
(%)
Northeast
Winter
600
-0.1
0.1
-51.1
59.8
Spring
681
-0.0
0.1
-21.3
56.5
Summer
679
-0.0
0.1
-26.7
53.3
129
-------
Climate
Region
Season
No. of
Obs
!i
ME
(ug/m3)
NMB
(%)
NME
(%)
Fall
649
-0.0
0.1
-30.6
47.9
Ohio Valley
Winter
297
-0.0
0.1
-36.7
53.4
Spring
300
-0.0
0.1
-26.2
38.8
Summer
309
-0.0
0.1
-26.6
51.5
Fall
288
-0.0
0.0
-20.3
52.3
Upper
Midwest
Winter
275
-0.0
0.0
-46.7
61.3
Spring
292
-0.0
0.1
-28.1
50.3
Summer
277
-0.0
0.0
-37.3
51.4
Fall
301
-0.0
0.1
-41.0
55.8
Southeast
Winter
359
-0.0
0.1
-34.3
52.5
Spring
376
-0.0
0.1
-34.2
54.9
Summer
413
-0.0
0.1
-33.2
54.1
Fall
385
-0.0
0.0
-27.2
62.5
South
Winter
236
-0.0
0.0
-26.4
51.1
Spring
263
-0.1
0.1
-48.2
57.1
Summer
281
-0.1
0.1
-46.4
64.7
Fall
280
-0.0
0.0
-42.4
62.2
Southwest
Winter
300
-0.0
0.0
-81.7
86.0
Spring
322
-0.0
0.0
-71.3
81.4
Summer
292
-0.0
0.0
-38.9
60.0
Fall
334
-0.0
0.0
-67.9
76.3
Northern
Rockies
Winter
216
-0.0
0.0
-74.8
86.8
Spring
251
-0.0
0.0
-55.3
61.1
Summer
226
-0.0
0.0
-32.4
54.0
Fall
237
-0.0
0.0
-52.6
66.1
Northwest
Winter
121
0.0
0.0
80.1
62.8
Spring
141
-0.0
0.0
-8.4
53.2
Summer
138
0.0
0.0
18.0
89.3
130
-------
Climate
Region
Season
No. of
Obs
!i
ME
(ug/m3)
NMB
(%)
NME
(%)
Fall
145
0.0
0.0
22.9
77.4
West
Winter
151
0.0
0.0
46.7
92.9
Spring
151
0.0
0.0
27.2
93.0
Summer
161
-0.0
0.0
-80.7
93.0
Fall
160
-0.0
0.0
-1.3
84.0
7.5 Model Simulation Scenarios
As part of our analysis for this rulemaking, the CMAQ modeling system was used to calculate
annual PM2.5 concentrations, 8-hour maximum average ozone season concentrations, annual
NO2, SO2, and CO concentrations, annual and seasonal (summer and winter) air toxics
concentrations, and annual nitrogen and sulfur deposition for each of the following emissions
scenarios:
2016 base year
2055 reference
2055 light and medium duty regulatory scenario
We use the predictions from the CMAQ model in a relative sense by combining the 2016
base-year predictions with predictions from each future-year scenario and applying these
modeled ratios to ambient air quality observations to estimate 8-hour ozone concentrations
during the ozone season (May - Sept), daily and annual PM2.5 concentrations, and visibility
impairment for each of the 2055 scenarios. The ambient air quality observations are average
conditions, on a site-by-site basis, for a period centered around the model base year (i.e., 2014-
2018).
The projected annual PM2.5 concentrations were calculated using the Speciated Modeled
Attainment Test (SMAT) approach that utilizes a Federal Reference Method (FRM) mass
construction methodology which results in reduced nitrates (relative to the amount measured by
routine speciation networks), higher mass associated with sulfates (reflecting water included in
FRM measurements), and a measure of organic carbonaceous mass that is derived from the
difference between measured PM2.5 and its non-carbon components. This characterization of
PM2.5 mass also reflects crustal material and other minor constituents. The resulting
characterization provides a complete mass balance. It does not have any unknown mass that is
sometimes presented as the difference between measured PM2.5 mass and the characterized
chemical components derived from routine speciation measurements. However, the assumption
that all mass difference is organic carbon has not been validated in many areas of the U.S. The
SMAT methodology uses the following PM2.5 species components: sulfates, nitrates, ammonium,
organic carbon mass, elemental carbon, crustal, water, and blank mass (a fixed value of 0.5
|ig/m3). More complete details of the SMAT procedures can be found in the report "Procedures
131
-------
for Estimating Future PM2.5 Values for the CAIR Final Rule by Application of the (Revised)
Speciated Modeled Attainment Test (SMAT)." For this analysis, several datasets and techniques
were updated. These changes are fully described within the technical support document for the
Final Transport Rule AQM TSD.
Additionally, we conducted an analysis to compare the absolute differences between the
future year reference and regulatory scenario for annual and seasonal acetaldehyde, benzene,
formaldehyde, and naphthalene, as well as annual NO2, SO2, CO, and nitrate/sulfate deposition.
These data were not compared in a relative sense due to the limited observational data available.
8 Additional Results of Air Quality Analysis
EPA conducted an air quality modeling analysis of a regulatory scenario involving light- and
medium-duty "onroad" vehicle emission reductions and corresponding changes in "upstream"
emission sources like EGU (electric generating unit) emissions and refinery emissions.
The RIA includes maps that present the impact of the LMDV regulatory scenario on projected
ozone, PM2.5, NO2, SO2, CO, and air toxics concentrations, and projected nitrogen and sulfur
deposition. In this TSD we present annual reference and LMDV regulatory scenario maps for
ozone, PM2.5, CO, NO2, SO2, air toxics, and nitrogen and sulfur deposition as well as seasonal
difference maps for air toxics and visibility levels at Mandatory Class I Federal Areas.
8.1 Annual 2055 Reference, LMDV Regulatory, and Onroad-Only Scenario Maps
The following section presents maps of projected ambient concentrations of PM2.5, ozone,
CO, NO2, SO2, acetaldehyde, benzene, 1,3-butadiene, formaldehyde, and naphthalene, and total
nitrogen and sulfur deposition in the 2055 reference case and the 2055 LMDV regulatory
scenario and the 2055 onroad-only scenario.
132
-------
>18.0
16.0
14.0
12.0
m
10.0 ฃ
D
8,0
6.0
4,0
<2.0
Figure 8-1 Projected Annual Average PM2.5 Concentrations in 2055 Reference Case (ug/m3)
133
-------
>18.0
16.0
14.0
12.0
m
10.0 ฃ
D
8,0
6.0
4,0
<2.0
Figure 8-2 Projected Annual Average PM2.5 Concentrations in 2055 LMDV Regulatory Scenario (ug/m3)
>18.0
16.0
14.0
12.0
m
10.0 -I
o>
13
8,0
6.0
4,0
<2.0
Figure 8-3 Projected Annual Average PM2.5 Concentrations in 2055 Onroad-Only Scenario (ug/m3)
134
-------
Figure 8-4 Projected Ozone Season (Apr-Sept) 8-hour Maximum Average Ozone Concentrations in 2055
Reference case (ppb)
Figure 8-5 Projected Ozone Season (Apr-Sept) 8-hour Maximum Average Ozone Concentrations in 2055
LMDV Regulatory Scenario (ppb)
135
-------
Figure 8-6 Projected Ozone Season (Apr-Sept) 8-hour Maximum Average Ozone Concentrations in 2055
Onroad-Only Scenario (ppb)
<12
Figure 8-7 Projected Annual Average CO Concentrations in 2055 Reference Case (ppb)
CO 2016 Annual avg 2055gh ref
136
-------
CO 2016 Annual ava 2055ah ctl
fR r
L -*ซ.
Wlax: 449-8623 Min: 64,7835
>120
Figure 8-8 Projected Annual Average CO Concentrations in 2055 LMDV Regulatory Scenario (ppb)
>120
Figure 8-9 Projected Annual Average CO Concentrations in 2055 Onroad-Only Scenario (ppb)
137
-------
r .
N02 2016 Annual avg 2055gh ref
A
*
*A:
tr
Wax: 28.8266 Min: 0
Vป .
\
ฆ0^76 \
Figure 8-10 Projected Annual Average NO2 Concentrations in 2055 Reference Case (ppb)
N02 2016 Annual avg 2055gh_onronly
f ฆ
ซV
&
b
'*A:
V *
*
L
?
j t'fc
ฆ>: 1
ft. <ฆ
. s
f
1 *
t V
M
Wax: 28.8217 Min: 0.0276
? *to'
>10.0
9.0
8.0
7.0
6.0 >
_Q
Q.
5.0 a
4.0
3.0
2.0
<1.0
>10.0
9.0
8.0
7.0
6.0 >
-O
Q.
5.0
4.0
3.0
2.0
<1.0
Figure 8-11 Projected Annual Average NO? Concentrations in 2055 LMDV Regulatory Scenario (ppb)
138
-------
F . ,"*
N02 2016 Annual avg 2055gh ctl
V . ( ; V
sr.-
s . -
*
M
."v
>10.0
9.0
8.0
7.0
6.0
5.0
4.0
3.0
2.0
<1.0
(ppb)
>1.00
0.90
0.80
0.70
0.60 >
JD
Q.
0.50 CL
0.40
0.30
0.20
<0.10
Figure 8-13 Projected Annual Average SOi Concentrations in 2055 Reference Case (ppb)
Figure 8-12 Projected Annual Average NO2 Concentrations in 2055 Oil road-Only Scenario
SQ2 2016 Annual avg 2055gh_ref
139
-------
Figure 8-14 Projected Annual Average SO2 Concentrations in 2055 LMDV Regulatory Scenario (ppb)
Figure 8-15 Projected Annual Average SO2 Concentrations in 2055 On road-Only Scenario (ppb)
140
-------
ALD2 UGM3 2016 Annual avg 2055gh ref
>1.50
1.35
1.20
1.05
0.90 m
ฃ
Ol
0.75
0.60
0.45
0.30
<0.15
to* xa
Max: 2.346 Min: 0.0376
Figure 8-16 Projected Annual Average Acetaldehyde Concentrations in 2055 Reference Case (ug/m3)
ALD2 UGM3 2016 Annual "avq 2055qh ctl
>1.50
1.35
1.20
1.05
0.90 m
E
ฆJ 0.75 S1
0.60
0.45
0.30
<0.15
Figure 8-17 Projected Annual Average Acetaldehyde Concentrations in 2055 LMDV Regulatory
Scenario (ug/m3)
141
-------
ALD2 UGM3 2016 Annual avg 2055gh onronly
>1.50
1.35
1.20
1.05
0.90 m
ฃ
Ol
0.75
0.60
0.45
0.30
<0.15
Max: 2.3455 Min: 0.0375
Figure 8-18 Projected Annual Average Acetaldehyde Concentrations in 2055 Onroad-Only Scenario
(ug/m3)
BENZENE 2016 Annual avg 2055gh ref
>0.20
0.18
0.16
0.14
0.12 >
-Q
Q.
o.io
0.08
0.06
0.04
<0.02
CT- W ^ Tl
Max: 3.5703 Min: 0.00&7 ' fr
Figure 8-19 Projected Annual Average Benzene Concentrations in 2055 Reference Case (ug/m3)
142
-------
BENZENE 2016 Annual avg 2055gh ctl
>0.20
0.18
0.16
0.14
0.12 >
.a
o.io
0.08
0.06
.0.04
ฆ V H / JIHWT* n<0.02
Max: 3.5701 Min: 0 nn<>7 ifr 1
Figure 8-20 Projected Annual Average Benzene Concentrations in 2055 LMDV Regulatory Scenario
(ug/m3)
BENZENE 2016 Annual avg 2055gh onronly
>0.20
0.18
0.16
0.14
0.12 >
XI
0.10 ฐ-
0.08
0.06
0.04
<0.02
_ \9 W VTl
Max: 3.5702 Min: 0.00^7 |fcji
Figure 8-21 Projected Annual Average Benzene Concentrations in 2055 Qnroad-Qnly Scenario (ug/m3)
143
-------
BUTADIENE13 2016 Annual avg 2055gh_ref
>0.020
0.018
0.016
0.014
0.012 m
E
0.010 5
0.008
0.006
0.004
<0.002
Case (ppb)
>0.020
0.018
0.016
0.014
0.012 m
0.010 =
0.008
0.006
0.004
<0.002
Figure 8-23 Projected Annual Average 1,3-Butadiene Concentrations in 2055 LMDV Regulatory
Scenario (ppb)
Figure 8-22 Projected Annual Average 1,3-Butadiene Concentrations in 2055 Reference
BUTADIENES 20X6 Annual avg 2055gh_ctl
144
-------
I >0.020
0.018
0.016
0.014
0.012 m
E
0.010 ?
10.008
ฆ 0.006
0.004
<0.002
Figure 8-24 Projected Annual Average 1,3-Butadiene Concentrations in 2055 Onroad-Qnly Scenario
(ppb)
BUTADIENE13 2016 Annua! avg 2055gh_onronly
145
-------
FORM 2016 Annual avg 2055gh_ref
>
.o
Q.
Q.
I >2.0
1.8
1.6
1.4
1.2
1.0
0.8
0.6
0.4
<0.2
Max: 10.0744 MirrJU^
Figure 8-25 Projected Annual Average Formaldehyde Concentrations in 2055 Reference Case (ug/m3)
FORM 2016 Annual avq 2055qh ctl
I >2.0
18
1.6
I.4
II,2
L
0.8
0.6
ฆ 0.4
<0.2
>
XI
Q.
CL
Figure 8-26 Projected Annual Average Formaldehyde Concentrations in 2055 LMDV Regulatory
Scenario (ug/m3)
146
-------
>2.0
1.8
1.6
1.4
1.2 >
n
Q.
1.0 ฐ-
0.8
0.6
0.4
<0.2
Figure 8-27 Projected Annual Average Formaldehyde Concentrations in 2055 Onroad-Only Scenario
(ug/m3)
I >0.050
0.045
0.040
0.035
0.030 m
E
10.025 ?
0.020
0.015
0.010
<0.005
Figure 8-28 Projected Annual Average Naphthalene Concentrations in 2055 Reference Case (ug/m3)
147
-------
NAPHTHALENE 2016 Annual avg 2055gh_ctl
>0.050
0.045
0.040
0.035
0.030 m
E
0.025 S1
0.020
0.015
wwmw 0010
.At "X. <0.005
Max: 0.5802 Min: 0.0 ' "9._
Figure 8-29 Projected Annual Average Naphthalene Concentrations in 2055 LMDV Regulatory Scenario
(ug/m3)
NAPHTHALENE 2016 Annual avg 2055gh_onronly
>0.050
0.045
0.040
0.035
0.030 m
E
CJ>
0.025 ^
I
0.020
0.015
0.010
<0.005
Figure 8-30 Projected Annual Average Naphthalene Concentrations in 2055 Onroad-Only Scenario
(ug/m3)
148
-------
TP-N-TOT 2016 Annual sum 2055gh_ref
<1.0
Figure 8-31 Projected Annual Nitrogen Deposition in 2055 Reference Case (kg N/ha)
TD_N TOT 2016 Annual sum 2055gh_ctl
Figure 8-32 Projected Annual Nitrogen Deposition in 2055 LMDV Regulatory Scenario (kg N/ha)
149
-------
TDNTOT 2016 Annual sum 2055gh_onronly
<1.0
Figure 8-33 Projected Annual Nitrogen Deposition in 2055 Onroad-Only Scenario (kg N/ha)
Figure 8-34 Projected Annual Sulfur Deposition in 2055 Reference Case (kg S/ha)
150
-------
Figure 8-35 Projected Annual Sulfur Deposition in 2055 LMDV Regulatory Scenario (kg S/ha)
Figure 8-36 Projected Annual Sulfur Deposition in 2055 Onroad-Only Scenario (kg S/ha)
8.2 Seasonal Air Toxics Maps
The following section presents maps of projected January and July monthly ambient
concentrations for acetaldehyde, benzene, 1,3-butadiene, formaldehyde, and naphthalene in the
151
-------
2055 reference case and the 2055 LMDV regulatory scenario and the 2055 onroad-only scenario,
as well as maps of projected January and July monthly average changes in ambient
concentrations in 2055.
ALD2 UGM3 2016 January avg 2055gh_ref
>1.50
1.35
1.20
1.05
0.90 m
C7I
0.75 ^
0.60
0.45
0.30
<0.15
Figure 8-37 Projected January Average Acetaldehyde Concentrations in 2055 Reference Case (ug/m3)
152
-------
ALD2 UGM3 2016 January avg 2055gh_ctl
I
>1.50
1.35
1.20
1.05
0.90 m
E
CT
0.75 =>
0.60
0.45
0.30
<0.15
Figure 8-38 Projected January Average Acetaldehyde Concentrations in 2055 LMDV Regulatory
Scenario (ug/m3)
ALD2 UGM3 2016 January avg 2055gh onronly
>1.50
1.35
1.20
1.05
0.90 m
E
CJ1
0.75 3
I
0.60
0.45
0.30
<0.15
Figure 8-39 Projected January Average Acetaldehyde Concentrations in 2055 Onroad-Only Scenario
(ug/m3)
153
-------
ALD2 UGM3 2016 July avg 2055gh_ref
I >1.50
1.35
1.20
1.05
0.90 ro
E
Ol
- 0.75 3
10.60
0.45
0.30
<0.15
Figure 8-40 Projected July Average Acetaldehyde Concentrations in 2055 Reference Case (ug/m3)
154
-------
ALD2 UGM3 2016 July avg 2055gh_ctl
>1.50
0.90 ro
0.60
<0.15
Figure 8-41 Projected July Average Acetaldehyde Concentrations in 2055 LMDV Regulatory Scenario
(ug/m3)
I >1.50
1.35
1.20
1.05
-0.90 m
E
0.75 ง"
<0.15
Figure 8-42 Projected July Average Acetaldehyde Concentrations in 2055 Onroad-Only Scenario
(ug/m3)
155
-------
BENZENE 2016 January avg 2055gh_ref
>0.20
0.18
0.16
0.14
0.12
0.10
0.08
0.06
0.04
<0.02
Figure 8-43 Projected January Average Benzene Concentrations in 2055 Reference Case (ppb)
156
-------
BENZENE 2016 January avg 2055gh_ctl
>
Q.
Q.
>0.20
0.18
0.16
0.14
0.12
0.10
0.08
0.06
0.04
<0.02
Figure 8-44 Projected January Average Benzene Concentrations in 2055 LMDV Regulatory Scenario
(ppb)
BENZENE 2016 January avg 2055gh onronly
>0.20
0.18
0.16
0.14
0.12 >
_Q
o.io ฐ-
0.08
0.06
0.04
<0.02
Figure 8-45 Projected January Average Benzene Concentrations in 2055 Qnroad-Only Scenario (ppb)
157
-------
BENZENE 2016 July avg 2055gh_ref
ft
1
Max: 4.2798 Min
0.000
Figure 8-46 Projected July Average Benzene Concentrations in 2055 Reference Case (ppb)
158
-------
BENZENE 2016 July avg 2055gh_ctl
>0.20
0.18
0.16
0.14
0.12
0.10
0.08
0.06
0.04
<0.02
>
.a
Q.
Q.
Scenario (ppb)
>0.20
0.18
0.16
0.14
0.12 >
-Q
0.10 ^
0.08
0.06
0.04
<0.02
Figure 8-48 Projected July Average Benzene Concentrations in 2055 Onroad-Only Scenario (ppb)
Figure 8-47 Projected July Average Benzene Concentrations in 2055 LMDV Regulatory
BENZENE 2016 July avg 2055gh_onronly
159
-------
BUTADIENE13 2016 January avg 2055gh ref
I >0.020
0.018
0.016
0.014
0.012 m
E
10.010 =
0.008
0.006
0.004
<0.002
Figure 8-49 Projected January Average 1,3-butadiene Concentrations in 2055 Reference Case (ug/m3)
160
-------
BUTADIENE13 2016 January avg 2055gh ctl
Max: 1.0631 Min: 0.0
I >0.020
0.018
0.016
0.014
0.012 m
E
10.010 =
0.008
0.006
0.004
<0.002
Regulatory
I >0.020
0.018
0.016
0.014
0.012 m
E
10.010 3
0.008
0.006
0.004
<0.002
Figure 8-51 Projected January Average 1,3-butadiene Concentrations in 2055 Onroad-Only Scenario
(ug/m3)
Figure 8-50 Projected January Average 1,3-butadiene Concentrations in 2055 LMDV
Scenario (ug/mJ)
BUTADIENE13 2016 January avq 2055qh onronly
Max: 1.0631 Min; 0.0
161
-------
BUTADIENE13 2016 July avg 2055gh ref
Jj
< ซ* '
-<5
IX
* ' "9^
f & '
x r- t 1
7" . V
c. r?>
. * T
J
-V,
"** . " > ' ฆ J
* \
i i :;-.4.
i." . >.ฆ*ซVI -if "
a4d^>V
Max: 1.5676 Min: 0.0
Figure 8-52 Projected July Average 1,3-butadiene Concentrations in 2055 Reference Case (ug/m3)
162
-------
BUTADIENE13 2016 July avg 2055gh ctl
Jj *ฆ
f-,v
Lvc3
if- ' "9^
4 # ^ -
:--,r
% * rป i ฆ y,, >ฆฃ e:*4' .
h* ^ t . : r:i#& /
- ฆ >< /J" ';ฆ . f
/
Max: 1.5677 Min: 0.0
/
>0.020
0.018
0.016
0.014
0.012
Max: 1.5677 Min: 0.0
fe
*
V1*
*ป' ปฆ / ; " ';T.
1
v<
>0.020
0.018
0.016
0.014
0.012 m
0.010 3
0.008
0.006
0.004
<0.002
Figure 8-54 Projected July Average 1,3-butadiene Concentrations in 2055 Onroad-Only Scenario
(ug/m3)
163
-------
164
-------
f r,.i
ซr . ฆ
FORM 2016 January avg 2055gh_ctl
>2.0
1.8
'1.6
1.4
1.2
1.0
0.8
0.6
'0.4
<0.2
>
-Q
Q.
Max: 12.7867 Min: 0.0895
Figure 8-56 Projected January Average Formaldehyde Concentrations in 2055 LMDV Regulatory
Scenario (ppb)
FORM 2016 January avg 2055gh onronly
>2.0
r
I.6
II.4
1.2 >
JD
Q.
1.0 Q-
10.8
0.6
10.4
<0.2
Max: 12.7865 Min: 0.0895
Figure 8-57 Projected January Average Formaldehyde Concentrations in 2055 Onroad-Only Scenario
(ppb)
165
-------
166
-------
Figure 8-59 Projected July Average Formaldehyde Concentrations in 2055 LMDV Regulatory Scenario
(ppb)
Figure 8-60 Projected July Average Formaldehyde Concentrations in 2055 Onroad-Only Scenario (ppb)
167
-------
NAPHTHALENE 2016 January avg 2055gh_ref
>0.050
0.045
0.040
0.035
0.030
-------
r-?-m
NAPHTHALENE 2016 January avg 2055gh ctl
>0.050
0.045
0.040
0.035
0.030 0.050
0.045
0.040
0.035
0.030 m
Ol
0.025 3
0.020
0.015
0.010
<0.005
Max: 0.3878 Min: 0.0
Figure 8-63 Projected January Average Naphthalene Concentrations in 2055 Qnroad-Only Scenario
(ug/m3)
169
-------
NAPHTHALENE 2016 July avg 2055gh ref
^XTV^
'ฆ rp .
c
t-
r
/ h
Max: 3.2062 Min: 0.0 L
* :.* <**
5* .
*
A : !%ฆ
i * 4 i
i *j ' ฆซ ...jv
; . .. > A
>/-'>- ?
-W h-j
I T\2M Lr*f-~ -ฆ
* 4 > j"
9 ;v Ir^ "
> .. .. r
J >ฆ "ฆ ฆ;ฆ'*. r
is
/
sv
a
ฆป. i
Figure 8-64 Projected July Average Naphthalene Concentrations in 2055 Reference Case (ug/m3)
170
-------
I >0.050
0.045
0.040
0.035
0.030 m
E
O)
10.025 a
0.020
0.015
0.010
<0.005
Figure 8-65 Projected July Average Naphthalene Concentrations in 2055 LMDV Regulatory Scenario
(ug/m3)
>0.050
0.045
0.040
0.035
0.030 m
E
Ol
0.025 =
0.020
0.015
0.010
<0.005
Figure 8-66 Projected July Average Naphthalene Concentrations in 2055 Onroad-Only Scenario (ug/m3)
NAPHTHALENE 2016 July avg 2055gh ctl
w
63F' M. I ' ra&
,f ฆ
'ฆ rp
t- '
r
.*ฆ .
*
m
* J V
y.pjt f
* <\
* * * ^ .
.'5 'j ' ฆ >
ฆ
ฆtf
&.
*
Max: 3.2062 Min: 0.0
A
/ 9'
NAPHTHALENE 2016 July avg 2055gh onronly
^ L ; " nib.
ฆ 2* ' ฆ
~ * * .
i.
1:.v ,
m
$.
*
vS
. ! - r"' c*Vy-'
4 - ป
ฆ -"'.'i r-T *
ฆ%
Max: 3.2062 Min: 0.0
171
-------
Max: 0.0168 Min: -0.051;
diff ALD2 UGM3 2016 January avg 2055gh_ctl - 2055gh_ref
I>6.00e-03
4.50e-03
3.00e-03
1.50e-03
0.00e+00
-1.50e-03
taJ-3.00e-03
-4.50e-03
<-6.00e-03
Figure 8-67 Projected Changes in Average Acetaldehyde Concentrations in January 2055 due to LMDV
Regulatory Scenario
diff ALD2 UGM3 2016 July avq 2055qh ctl - 2055gh ref
I
I
>6.00e-03
4.50e-03
3.00e-03
1.50e-03
r
0.00e+00 i
I
-1.50e-03
-3.00e-03
-4.50e-03
<-6.00e-03
Figure 8-68 Projected Changes in Average Acetaldehyde Concentrations in July 2055 due to LMDV
Regulatory Scenario
172
-------
diff ALD2_UGM3 2016 January avg 2055gh onronly - 2055gh ref
I>6.00e-03
4.50e-03
3.00e-03
1.50e-03
m
0.00e+00 -I
01
3
-1.50e-03
U-3.00e-03
I-4.50e-03
<-6.00e-03
Figure 8-69 Projected Changes in Average Acetaldehyde Concentrations in January 2055 from "Onroad
Only" Emissions Changes
diff ALD2UGM3 2016 July avg 2055gh_onronly - 2055gh_ref
#
b
n
ฆw
i *
Max: 0.001 Min: -0.033fi
L
4
I
I
>6.00e-03
4.50e-03
3.00e-03
1.50e-03
r
0.00e+00 i
I
-1.50e-03
-3.00e-03
4.50e-03
<-6.00e-03
Figure 8-70 Projected Changes in Average Acetaldehyde Concentrations in July 2055 from "Onroad
Only" Emissions Changes
173
-------
diff BENZENE 2016 January avg 2055gh ctl - 2055gh_ref
*
&
> /.
r i
fir.
Max: 0.0065 Min: -0.0392
: -0.039:
I
>2.00e-03
n 1.50e-03
1.00e-03
5.00e-04
>
0.00e+00 ฆง_
Q.
-5.00e-04
tad -1.00e-03
-1.50e-03
<-2.00e-03
Figure 8-71 Projected Changes in Average Benzene Concentrations in January 2055 due to LMDV
Regulatory Scenario
diff BENZENE 2016 July avg 2055gh_ctl - 2055gh_ref
J
I
\v
} * ' -'w* ฆ'
ฆ * T
>2.00e-03
1.50e-03
1.00e-03
5.00e-04
0.00e+00 i
-5.00e-04
-1.00e-03
-1.50e-03
<-2.00e-03
Max: 0.013 Min: 0.0443
ฆL
Figure 8-72 Projected Changes in Average Benzene Concentrations in July 2055 due to LMDV
Regulatory Scenario
174
-------
Max: -0.0 Min: -0.027:
A
- 2055gh ref
diff BENZENE 2016 January avg
'
>2.00e-03
1.50e-03
1.00e-03
5.00e-04
>
0.00e+00 ฆง_
Q.
-5.00e-04
-1.00e-03
I-1.50e-03
<-2.00e-03
Figure 8-73 Projected Changes in Average Benzene Concentrations in January 2055 from "Qnroad
Only" Emissions Changes
diff BENZENE 2016 July avg 2055gh onronly - 2055gh ref
I
>2.00e-03
n 1.50e-03
1.00e-03
5.00e-04
0.00e+00 i
-5.00e-04
-1.00e-03
-1.50e-03
<-2.00e-03
Figure 8-74 Projected Changes in Average Benzene Concentrations in July 2055 from "Onroad Only"
Emissions Changes
175
-------
diff BUTADIENE13 2016 January avg 2055gh_ctl - 2055gh_ref
' tfl JllL
Max: le-04 Min:
-0.01^4
I
>2.00e-04
1.50e-04
1.00e-04
5.00e-05
0.00e+00
-5.00e-05
-1.00e-04
-1.50e-04
<-2.00e-04
Figure 8-75 Projected Changes in Average 1,3-Butadiene Concentrations in January 2055 due to LMDV
Regulatory Scenario
diff BUTADIENE13 2016 July avg 2055gh_ctl - 2055gh_ref
i
i
*
Max: le-04 Min: -O.OOi
I
>2.00e-04
1.50e-04
1.00e-04
5.00e-05
0.00e+00
-5.00e-05
J-1.00e-04
-1.50e-04
<-2.00e-04
I
Figure 8-76 Projected Changes in Average 1,3-Butadiene Concentrations in July 2055 due to LMDV
Regulatory Scenario
176
-------
cliff BUTADIENE13 2016 January avg 2055gh onronly - 2055gh ref
I
I
>2.00e-04
1.50e-04
1.00e-04
5.00e-05
0.00e+00
-5.00e-05
-1.00e-04
-1.50e-04
<-2.00e-04
Figure 8-77 Projected Changes in Average 1,3-Butadiene Concentrations in January 2055 from "Onroad
Only" Emissions Changes
diff BUTADIENE13 2016 July avg 2055gh_onronly - 2055gh_ref
i
Max: le-04 Min:
-0.00^3
I
>2.00e-04
1.50e-04
1.00e-04
5.00e-05
0.00e+00
-5.00e-05
y-1.00e-04
-1.50e-04
<-2.00e-04
Figure 8-78 Projected Changes in Average 1,3-Butadiene Concentrations in July 2055 from "Onroad
Only" Emissions Changes
177
-------
Figure 8-79 Projected Changes in Average Formaldehyde Concentrations in January 2055 due to LMDV
Regulatory Scenario
diff FORM 2016 July avg 2055qh ctl - 2055gh ref
Max: 0.1816 Min: -0.1934
>8.00e-03
H6.00e-03
4.00e-03
2.00e-03
0.00e+00 i
-2.00e-03
-4.00e-03
-6.00e-03
<-8.00e-03
2055 due to LMDV
Figure 8-80 Projected Changes in Average Formaldehyde Concentrations in July
Regulatory Scenario
178
-------
diff FORM 2016 January avg 2055gh_onronly - 2055gh_ref
A
r
Max: -0.0 Min: -0.0246'
I
>8.00e-03
6.00e-03
4.00e-03
2.00e-03
0.00e+00
Q.
-2.00e-03
-4.00e-03
-6.00e-03
<-8.00e-03
4.00e-03
2.00e-03
0.00e+00 ฃ
o.
-2.00e-03
U-4.00e-03
Max: le-04 Min: -0.03
6.00e-03
<-8.00e-03
Figure 8-82 Projected Changes in Average Formaldehyde Concentrations in July 2055 from "Onroad
Only" Emissions Changes
Figure 8-81 Projected Changes in Average Formaldehyde Concentrations in January 2055 from
"Onroad Only" Emissions Changes
diff FORM 2016 July avg 2055gh_onronly - 2055gh_ref
>8.00e-03
6.00e-03
179
-------
Max: 0.0 Min: -0.0031
diff
Figure 8-83 Projected Changes in Average Naphthalene Concentrations in January 2055 due to LMDV
Regulatory Scenario
diff NAPHTHALENE 2016 July avg 2055gh_ctl - 2055gh_ref
rt
.yf
M
ฆ*?m
ฆv
1 L
1 1
I
*. ~
ฆ
4
Max: 0.0 Min: -0.0016
1
ฆk
ป
>2.00e-04
Hl.50e-04
1.00e-04
5.00e-05
0.00e+00
-5.00e-05
J-1.00e-04
-1.50e-04
<-2.00e-04
Figure 8-84 Projected Changes in Average Naphthalene Concentrations in July 2055 due to LMDV
Regulatory Scenario
180
-------
cliff NAPHTHALENE 2016 January avg 2055gh onronly - 2055gh ref
h
' '
t
Max: -0.0 Min: -0.0031'
I
I
>2.00e-04
1.50e-04
1.00e-04
5.00e-05
0.00e+00
-5.00e-05
d-1.00e-04
-1.50e-04
<-2.00e-04
Figure 8-85 Projected Changes in Average Naphthalene Concentrations in January 2055 from "Qnroad
Only" Emissions Changes
diff NAPHTHALENE 2016 July avg 2055gh_onronly - 2055gh_ref
ป.
j
.i
. ifwmL 'J
ฆ >r
* <* mL v . % i
>' : :
%
i
ซ y
/
A
Max: 0.0 Min: -0.0016
I
P\
XI
I
>2.00e-04
1.50e-04
1.00e-04
5.00e-05
0.00e+00
-5.00e-05
J-X.00e-04
-1.50e-04
<-2.00e-04
I
Figure 8-86 Projected Changes in Average Naphthalene Concentrations in July 2055 from "Onroad
Only" Emissions Changes
8.3 Projected Visibility in Mandatory Class I Federal Areas
Air quality modeling was used to project visibility conditions in 145 Mandatory Class I
Federal areas across the U.S. with and without the rule in 2055. Fhe results show that in 2055,
the rule will improve projected visibility on the 20% most impaired days in 138 of the modeled
181
-------
areas (95%) and will lead to no change for the other 7 modeled areas (5%). The average visibility
on the 20 percent most impaired days at all modeled Mandatory Class I Federal areas is projected
to improve by 0.04 deciviews, or 0.34 percent, in 2055. The greatest improvement in visibility
would occur in Mammoth Cave Area in Kentucky, where visibility is projected to improve by
1.26 percent (0.20 deciviews) in 2055 due to the rule.
Table 8-lProjected Visibility in Mandatory Class I Federal areas in 2055 in AQM Reference and
Regulatory cases
Class I Area Name
State
2016
Baseline
Visibility
(dv) on
20%
Most
Impaired
Days
2055
Reference
Visibility
(dv) on
20%
Most
Impaired
Days
2055
LMDV
Regulatory
Scenario
Visibility
(dv) on
20% Most
Impaired
Days
2055
Onroad-
Only
Scenario
Visibility
(dv) on
20%
Most
Impaired
Days
Natural
Background
(dv) on 20%
Most
Impaired
Days
Sipsey Wilderness
Alabama
19.03
14.72
14.59
14.59
9.62
Chiricahua NM
Arizona
7.29
6.87
6.86
6.86
4.18
Chiricahua Wilderness
Arizona
9.41
8.74
8.72
8.72
4.93
Galiuro Wilderness
Arizona
9.41
8.74
8.72
8.72
4.93
Grand Canyon NP
Arizona
9.41
8.74
8.72
8.72
4.93
Mazatzal Wilderness
Arizona
6.87
6.38
6.37
6.36
4.16
Mount Baldy Wilderness
Arizona
9.47
8.90
8.88
8.88
5.22
Petrified Forest NP
Arizona
8.16
7.50
7.48
7.48
4.21
Pine Mountain Wilderness
Arizona
9.47
8.90
8.88
8.88
5.22
Saguaro NM
Arizona
10.75
10.20
10.16
10.16
5.14
Superstition Wilderness
Arizona
10.45
9.81
9.78
9.78
5.14
Sycamore Canyon Wilderness
Arizona
11.96
11.58
11.57
11.57
4.68
Caney Creek Wilderness
Arkansas
18.29
14.04
13.96
13.98
9.54
Upper Buffalo Wilderness
Arkansas
17.95
14.18
14.08
14.09
9.41
Agua Tibia Wilderness
California
16.34
14.93
14.84
14.86
7.66
Ansel Adams Wilderness (Minarets)
California
10.98
10.28
10.26
10.26
6.06
Caribou Wilderness
California
10.23
9.66
9.64
9.64
6.10
Cucamonga Wilderness
California
13.19
11.66
11.51
11.54
6.12
Desolation Wilderness
California
9.31
8.82
8.80
8.81
4.91
Dome Land Wilderness
California
15.14
14.32
14.29
14.30
6.19
Emigrant Wilderness
California
11.57
11.08
11.06
11.06
6.29
Hoover Wilderness
California
7.65
7.31
7.30
7.30
4.90
John Muir Wilderness
California
10.98
10.28
10.26
10.26
6.06
Joshua Tree NM
California
12.87
12.16
12.12
12.12
6.09
Kaiser Wilderness
California
10.98
10.28
10.26
10.26
6.06
Kings Canyon NP
California
18.43
17.40
17.36
17.37
6.29
Lassen Volcanic NP
California
9.67
9.29
9.28
9.28
6.18
182
-------
Class I Area Name
State
2016
Baseline
Visibility
(dv) on
20%
Most
Impaired
Days
2055
Reference
Visibility
(dv) on
20%
Most
Impaired
Days
2055
LMDV
Regulatory
Scenario
Visibility
(dv) on
20% Most
Impaired
Days
2055
Onroad-
Only
Scenario
Visibility
(dv) on
20%
Most
Impaired
Days
Natural
Background
(dv) on 20%
Most
Impaired
Days
Lava Beds NM
California
10.23
9.66
9.64
9.64
6.10
Mokelumne Wilderness
California
9.31
8.82
8.80
8.81
4.91
Pinnacles NM
California
14.10
13.38
13.36
13.35
6.94
Redwood NP
California
14.11
13.09
13.06
13.07
6.80
San Gabriel Wilderness
California
12.65
12.42
12.42
12.42
8.59
San Gorgonio Wilderness
California
13.19
11.66
11.51
11.54
6.12
San Jacinto Wilderness
California
14.45
12.54
12.42
12.44
6.20
San Rafael Wilderness
California
14.45
12.54
12.42
12.44
6.20
Sequoia NP
California
18.43
17.40
17.36
17.37
6.29
South Warner Wilderness
California
9.67
9.29
9.28
9.28
6.18
Thousand Lakes Wilderness
California
10.23
9.66
9.64
9.64
6.10
Ventana Wilderness
California
14.10
13.38
13.36
13.35
6.94
Yo Semite NP
California
11.57
11.08
11.06
11.06
6.29
Black Canyon of the Gunnison NM
Colorado
6.55
6.16
6.15
6.15
3.97
Eagles Nest Wilderness
Colorado
4.98
4.53
4.52
4.52
3.02
Flat Tops Wilderness
Colorado
4.98
4.53
4.52
4.52
3.02
Great Sand Dunes NM
Colorado
8.02
7.52
7.50
7.50
4.45
La Garita Wilderness
Colorado
6.55
6.16
6.15
6.15
3.97
Maroon Bells-Snowmass Wilderness
Colorado
4.98
4.53
4.52
4.52
3.02
Mesa Verde NP
Colorado
6.51
5.78
5.76
5.75
4.20
Mount Zirkel Wilderness
Colorado
5.47
4.89
4.86
4.86
3.16
Rawah Wilderness
Colorado
5.47
4.89
4.86
4.86
3.16
Rocky Mountain NP
Colorado
8.41
7.39
7.35
7.35
4.94
Weminuche Wilderness
Colorado
4.98
4.53
4.52
4.52
3.02
West Elk Wilderness
Colorado
6.55
6.16
6.15
6.15
3.97
Chassahowitzka
Florida
17.41
15.43
15.38
15.38
9.03
Everglades NP
Florida
14.90
14.07
14.06
14.06
8.33
St. Marks
Florida
17.39
15.25
15.22
15.22
9.13
Cohutta Wilderness
Georgia
17.37
13.58
13.49
13.50
9.88
Okefenokee
Georgia
17.39
15.66
15.63
15.63
9.45
Wolf Island
Georgia
17.39
15.66
15.63
15.63
9.45
Craters of the Moon NM
Idaho
8.50
7.54
7.48
7.47
4.97
Sawtooth Wilderness
Idaho
8.61
8.34
8.33
8.33
4.70
Selway-Bitterroot Wilderness
Idaho
8.37
8.13
8.13
8.13
5.45
183
-------
Class I Area Name
State
2016
Baseline
Visibility
(dv) on
20%
Most
Impaired
Days
2055
Reference
Visibility
(dv) on
20%
Most
Impaired
Days
2055
LMDV
Regulatory
Scenario
Visibility
(dv) on
20% Most
Impaired
Days
2055
Onroad-
Only
Scenario
Visibility
(dv) on
20%
Most
Impaired
Days
Natural
Background
(dv) on 20%
Most
Impaired
Days
Mammoth Cave NP
Kentucky
21.02
15.87
15.67
15.68
9.80
Breton
Louisiana
18.97
17.33
17.29
17.30
9.23
Acadia NP
Maine
14.54
13.36
13.31
13.31
10.39
Moosehorn
Maine
13.32
12.41
12.37
12.37
9.98
Roosevelt Campobello International Park
Maine
13.32
12.41
12.37
12.37
9.98
Isle Royale NP
Michigan
15.54
14.05
13.97
13.98
10.17
Seney
Michigan
17.57
15.37
15.23
15.23
11.11
Boundary Waters Canoe Area
Minnesota
13.96
12.53
12.46
12.47
9.09
Voyageurs NP
Minnesota
14.18
12.86
12.81
12.81
9.37
Hercules-Glades Wilderness
Missouri
18.72
15.02
14.91
14.92
9.30
Mingo
Missouri
20.13
16.18
16.00
16.01
9.18
Anaconda-Pintler Wilderness
Montana
8.37
8.13
8.13
8.13
5.45
Bob Marshall Wilderness
Montana
10.06
9.86
9.86
9.86
5.53
Cabinet Mountains Wilderness
Montana
9.87
9.61
9.59
9.59
5.64
Gates of the Mountains Wilderness
Montana
7.47
7.33
7.32
7.32
4.53
Glacier NP
Montana
13.77
13.39
13.34
13.34
6.90
Medicine Lake
Montana
15.30
15.25
15.18
15.20
5.95
Mission Mountains Wilderness
Montana
10.06
9.86
9.86
9.86
5.53
Red Rock Lakes
Montana
7.52
7.16
7.15
7.15
3.97
Scapegoat Wilderness
Montana
10.06
9.86
9.86
9.86
5.53
UL Bend
Montana
10.93
10.97
10.96
10.96
5.87
Jarbidge Wilderness
Nevada
7.97
7.75
7.74
7.74
5.23
Great Gulf Wilderness
New Hampshire
13.07
11.43
11.38
11.38
9.78
Presidential Range-Dry River Wilderness
New Hampshire
13.07
11.43
11.38
11.38
9.78
Brigantine
New Jersey
19.31
16.45
16.28
16.27
10.68
Bandelier NM
New Mexico
8.44
7.74
7.70
7.70
4.59
Bosque del Apache
New Mexico
10.47
9.62
9.59
9.60
5.39
Carlsbad Caverns NP
New Mexico
12.64
12.60
12.58
12.59
4.83
Gila Wilderness
New Mexico
7.58
7.10
7.08
7.08
4.20
Pecos Wilderness
New Mexico
5.95
5.35
5.33
5.33
3.50
Salt Creek
New Mexico
14.97
14.70
14.63
14.66
5.49
San Pedro Parks Wilderness
New Mexico
6.43
5.87
5.86
5.85
3.33
Wheeler Peak Wilderness
New Mexico
9.95
9.56
9.54
9.54
4.89
White Mountain Wilderness
New Mexico
5.95
5.35
5.33
5.33
3.5
184
-------
Class I Area Name
State
2016
Baseline
Visibility
(dv) on
20%
Most
Impaired
Days
2055
Reference
Visibility
(dv) on
20%
Most
Impaired
Days
2055
LMDV
Regulatory
Scenario
Visibility
(dv) on
20% Most
Impaired
Days
2055
Onroad-
Only
Scenario
Visibility
(dv) on
20%
Most
Impaired
Days
Natural
Background
(dv) on 20%
Most
Impaired
Days
Linville Gorge Wilderness
North Carolina
16.42
12.62
12.57
12.56
9.70
Shining Rock Wilderness
North Carolina
15.49
11.68
11.63
11.64
10.25
Swanquarter
North Carolina
16.30
13.40
13.29
13.29
10.01
Lostwood
North Dakota
16.18
16.35
16.28
16.29
5.87
Theodore Roosevelt NP
North Dakota
14.06
13.29
13.20
13.21
5.94
Wichita Mountains
Oklahoma
18.12
15.40
15.31
15.32
6.92
Crater Lake NP
Oregon
7.98
7.68
7.67
7.67
5.16
Diamond Peak Wilderness
Oregon
7.98
7.68
7.67
7.67
5.16
Eagle Cap Wilderness
Oregon
11.19
10.19
10.15
10.15
6.58
Gearhart Mountain Wilderness
Oregon
7.98
7.68
7.67
7.67
5.16
Hells Canyon Wilderness
Oregon
12.33
11.37
11.31
11.31
6.57
Kalmiopsis Wilderness
Oregon
11.97
11.57
11.55
11.55
7.78
Mount Hood Wilderness
Oregon
9.27
8.78
8.76
8.76
6.59
Mount Jefferson Wilderness
Oregon
11.28
10.85
10.84
10.84
7.30
Mount Washington Wilderness
Oregon
7.98
7.68
7.67
7.67
5.16
Mountain Lakes Wilderness
Oregon
11.28
10.85
10.84
10.84
7.30
Strawberry Mountain Wilderness
Oregon
11.19
10.19
10.15
10.15
6.58
Three Sisters Wilderness
Oregon
11.28
10.85
10.84
10.84
7.30
Cape Romain
South Carolina
17.67
15.32
15.28
15.28
9.78
Badlands NP
South Dakota
12.33
11.45
11.40
11.40
6.09
Wind Cave NP
South Dakota
10.53
9.44
9.43
9.41
5.64
Great Smoky Mountains NP
Tennessee
17.21
13.54
13.45
13.45
10.05
Joyce-Kilmer-Slickrock Wilderness
Tennessee
17.21
13.54
13.45
13.45
10.05
Big Bend NP
Texas
14.06
13.31
13.30
13.31
5.33
Guadalupe Mountains NP
Texas
12.64
12.60
12.58
12.59
4.83
Arches NP
Utah
6.76
5.80
5.77
5.76
4.13
Bryce Canyon NP
Utah
6.60
6.03
6.01
6.01
4.08
Canyonlands NP
Utah
6.76
5.80
5.77
5.76
4.13
Capitol Reef NP
Utah
7.18
6.56
6.54
6.54
4.00
ZionNP
Utah
8.76
8.31
8.31
8.30
5.18
Lye Brook Wilderness
Vermont
14.75
12.55
12.44
12.44
10.24
James River Face Wilderness
Virginia
17.89
13.83
13.71
13.71
9.47
Shenandoah NP
Virginia
17.07
12.06
11.96
11.95
9.52
Alpine Lake Wilderness
Washington
12.74
11.71
11.63
11.64
7.27
185
-------
Class I Area Name
State
2016
Baseline
Visibility
(dv) on
20%
Most
Impaired
Days
2055
Reference
Visibility
(dv) on
20%
Most
Impaired
Days
2055
LMDV
Regulatory
Scenario
Visibility
(dv) on
20% Most
Impaired
Days
2055
Onroad-
Only
Scenario
Visibility
(dv) on
20%
Most
Impaired
Days
Natural
Background
(dv) on 20%
Most
Impaired
Days
Glacier Peak Wilderness
Washington
9.98
9.57
9.54
9.55
6.89
Goat Rocks Wilderness
Washington
7.98
7.63
7.62
7.62
6.14
Mount Adams Wilderness
Washington
12.66
12.11
12.08
12.08
7.66
Mount Rainier NP
Washington
9.98
9.57
9.54
9.55
6.89
North Cascades NP
Washington
11.90
11.72
11.71
11.71
6.90
Olympic NP
Washington
9.46
9.05
9.04
9.04
5.96
Pasayten Wilderness
Washington
7.98
7.63
7.62
7.62
6.14
Dolly Sods Wilderness
West Virginia
17.65
12.26
12.17
12.17
8.92
Otter Creek Wilderness
West Virginia
17.65
12.26
12.17
12.17
8.92
Bridger Wilderness
Wyoming
6.77
6.34
6.33
6.33
3.92
Fitzpatrick Wilderness
Wyoming
6.77
6.34
6.33
6.33
3.92
Grand Teton NP
Wyoming
7.52
7.16
7.15
7.15
3.97
North Absaroka Wilderness
Wyoming
7.17
6.80
6.79
6.79
4.55
Teton Wilderness
Wyoming
7.52
7.16
7.15
7.15
3.97
Washakie Wilderness
Wyoming
7.17
6.80
6.79
6.79
4.55
Yellowstone NP
Wyoming
7.52
7.16
7.15
7.15
3.97
a The level of visibility impairment in an area is based on the light-extinction coefficient and a unitless visibility
index, called a "deciview", which is used in the valuation of visibility. The deciview metric provides a scale for
perceived visual changes over the entire range of conditions, from clear to hazy. Under many scenic conditions, the
average person can generally perceive a change of one deciview. The higher the deciview value, the worse the
visibility. Thus, an improvement invisibility is a decrease in deciview value.
186
------- |