United Slates
0*5#	Environmental Protected
^1	Agflncy
Technical Guidance on Tracking Visibility Progress
for the Second Implementation Period of the
Regional Haze Program

-------

-------
EPA-454/R-18-010
December 2018
Technical Guidance on Tracking Visibility Progress for the Second Implementation Period
of the Regional Haze Program
U.S. Environmental Protection Agency
Office of Air Quality Planning and Standards
Air Quality Assessment Division
Research Triangle Park, North Carolina

-------
Table of Contents
1.0 Purpose of this guidance	1
1.1	Regional Haze Background	2
1.2	Statutory Provisions and Regulatory Requirements	2
2.0 Ambient Data Analysis	4
2.1	Recommendations for estimating daily natural and anthropogenic visibility fractions
and light extinction budgets and calculating the 20% most impaired and 20% clearest
days	5
2.2	Calculating the baseline, current, and natural visibility conditions	14
3.0 Adjustment of the Uniform Rate of Progress for Impacts from International Anthropogenic
Emissions	17
3.1	URP Adjustment in the Regional Haze Rule	17
3.2	Year Selection for Estimating International Contribution	18
3.3	Estimating the Anthropogenic International Visibility Impacts	20
3.4	EPA Review of an International URP Adjustment	22
References	24
APPENDIX A	26
APPENDIX B	28

-------
1.0 Purpose of this guidance
The purpose of this guidance document is to support states in the development of regional
haze state implementation plans (SIPs) to protect visibility in certain national parks and
wilderness areas, known as mandatory Class I Federal areas,1 in particular the SIPs that are due
to be submitted to the Environmental Protection Agency (EPA) by July 31, 2021, for the second
implementation period. The required content of these SIPs is specified in 40 CFR 51.308(f),
which was revised in 2017.2 As called for in the agency's "Regional Haze Reform Roadmap"
(September 11, 2018), this guidance document describes EPA's recommended methods on two
aspects of the regional haze program:
1)	Visibility Tracking Metrics: The 2017 Regional Haze Rule revisions require a revised
approach to tracking visibility improvements over time within the Uniform Rate of
Progress (URP) framework.3 Under these 2017 Regional Haze Rule revisions, in the
second and future implementation periods, states must select the "20 percent most
impaired days" each year at each Class I area based on daily anthropogenic
impairment.4 This guidance document describes a recommended methodology to
develop the baseline and current visibility conditions, and natural conditions on the
most impaired and clearest days.
2)	International Emissions: The 2017 Regional Haze Rule includes a provision that allows
states to propose an adjustment to the URP to account for impacts from anthropogenic
sources outside the United States, if the adjustment has been developed through
scientifically valid data and methods. This guidance document describes recommended
tools and methods to develop optional adjustments to the URP endpoint to account for
international anthropogenic emissions impacts.
This document provides recommendations on these two aspects of SIP development under the
Regional Haze Rule and is for use by states in developing SIP submissions and for EPA Regional
offices in acting on them. This document does not substitute for provisions or requirements of
the Clean Air Act (CAA), nor is it a rule itself. Thus, it does not impose binding, enforceable
requirements on any party. States retain the discretion to develop regional haze SIP revisions
that differ from this guidance so long as they are consistent with the CAA and the implementing
regulations - a core principle of cooperative federalism.
1For brevity, mandatory Class I Federal areas will often be referred to as "Class I areas" in the remainder of this
document.
2Final Rule: Protection of Visibility: Amendments to Requirements for State Plans, 82 FR 3078, January 10, 2017.
3"URP framework" refers to the interrelated Regional Haze Rule requirements regarding the quantification of
historical and projected visibility conditions using specific metrics, the quantification of natural conditions, the
quantification of the uniform progress that would achieve natural visibility conditions for the 20 percent most
anthropogenically impaired days in 2064, the URP glidepath, the setting of reasonable progress goals (RPGs) for
the end of the implementation period, and the comparison of the RPG for the 20 percent most anthropogenically
impaired days to the URP glidepath.
Previously, states and EPA tracked visibility progress on the 20 percent worst visibility days.
1

-------
EPA generally expects that SIPs which follow this guidance are likely to meet the related
applicable statutory and regulatory requirements. Final decisions by EPA to approve a particular
SIP revision can only be made based on the requirements of the statute and the Regional Haze
Rule, and on whether the SIP submission is the product of reasoned decision making. In
addition, final EPA decisions can only be made following a state's final submission of the SIP
revision to EPA in accordance with all applicable requirements, including appropriate notice and
opportunity for public review and comment. Only final actions taken to approve or disapprove
SIP submissions would be final actions for purposes of CAA section 307(b). Therefore, this
guidance is not judicially reviewable. This guidance does not change or substitute for any law,
regulation, or other legally binding requirement and is not legally enforceable. Due to case-
specific circumstances, following the recommendations in this document does not ensure that
the related aspects of a SIP will be approvable in all instances, as this guidance may not apply to
the facts and circumstances underlying a particular SIP.
We encourage states to discuss with their EPA Regional office early in their SIP development
the approach they anticipate taking and how the interpretations and recommendations in this
guidance may relate to their SIPs.
1.1	Regional Haze Background
Consistent with the CAA, "regional haze" is defined at 40 CFR 51.300 as "visibility impairment
that is caused by the emission of air pollutants from numerous anthropogenic sources located
over a wide geographic area. Such sources include, but are not limited to, major and minor
stationary sources, mobile sources, and area sources." This visibility impairment is a result of
anthropogenic particles and gases in the atmosphere that scatter and absorb (i.e., extinguish)
light, thus acting to reduce overall visibility. The primary cause of regional haze is light
extinction by particulate matter (PM). For purposes of the Regional Haze Rule, light extinction is
estimated from measurements of PM and its chemical components (sulfate, nitrate, organic
mass by carbon (OMC), light absorbing carbon (LAC), fine soil (FS), sea salt, and coarse material
(CM)), assumptions about relative humidity at the monitoring site, and the use of a commonly
accepted algorithm (Pitchford, et al., 2007). These estimates of light extinction are
logarithmically transformed to deciview units. The Regional Haze Rule established the deciview
haze index as the principal metric for expressing visibility on any particular day. The deciview
haze index is calculated from light extinction values and expresses uniform changes in the
degree of haze in terms of common increments across the entire range of visibility conditions,
from pristine to extremely hazy.
The PM measurements used in the regional haze program are collected by the IMPROVE
(Interagency Monitoring for PROtected Visual Environments) monitoring network. The Regional
Haze Rule requires states to submit a series of SIPs to protect visibility in Class I areas.
1.2	Statutory Provisions and Regulatory Requirements
In section 169A of the 1977 Amendments to the CAA, Congress established a program for
protecting and restoring visibility in certain national parks, wilderness areas, and other Class I
2

-------
areas due to their "great scenic importance."5 This section of the CAA establishes as a national
goal the "prevention of any future, and the remedying of any existing, impairment of visibility in
Class I areas which impairment results from manmade air pollution." This section also required
EPA to issue regulations requiring states to adopt implementation plans containing emission
limits as may be necessary to make reasonable progress towards meeting this goal.
In 2017, EPA issued a final rule revising portions of the visibility protection rule promulgated in
1980 and the Regional Haze Rule promulgated in 1999.6 The revised rule covers EPA's review of
periodic SIPs developed for the second and subsequent implementation periods, among other
requirements.
The Regional Haze Rule established the concept of state-set reasonable progress goals (RPG) for
the 20 percent most anthropogenically impaired days as a regulatory construct promulgated to
implement the statutory requirements for visibility protection. These RPGs reflect the visibility
conditions that are projected to be achieved by the end of the applicable implementation
period as a result of its own and other states' long-term strategies, as well as the
implementation of other requirements of the CAA.
The 2017 Regional Haze Rule requires states to determine the baseline (2000-2004) visibility
condition for the 20 percent most impaired days and requires that the long-term strategy and
RPG must provide for improvement in visibility for the most impaired days, relative to the
baseline period. Specifically, states must determine the rate of improvement in visibility that
would need to be maintained during each implementation period in order to reach natural
conditions by 2064 for the 20 percent most impaired days, given the starting point of the 2000-
2004 baseline visibility condition.7 The "glidepath," or URP, is the amount of visibility
improvement that would be needed to stay on a linear path from the baseline period to natural
conditions.
The URP is calculated according to the following formula:
URP = [(2000-2004 visibility^* most impaired (natural visibility)2o*most/mPa/red]/60	(Eqn. 1)
An example diagram of the URP (in this case for GRSM1 in Great Smoky Mountains National
Park) for the entire 2000-2064 period is shown in Figure 1. In this diagram, the URP (orange
line) connects the 2000-2004 baseline period with the 2064 endpoint at the estimate of natural
visibility conditions.
5H.R. Rep. No. 294, 95th Cong. 1st Sess. at 205 (1977).
645 FR 80084 (December 2,1980) and 64 FR 35714 (July 1,1999)
7See 40 CFR 51.308(f)(1).
3

-------
5-yr running averages used to assess
current visibility impairment (dv)
30.0-
27.0
Uniform rate of progress line ("glidepath"),
starting at 2000-2004 and ending at 2064 (dv)
24.0-
21.0
18.0-
Annual average of the
20% most impaired days (dv)
15.0
O 12.0-
9.0 -
Baseline visibility conditions
on the 20% clearest days
6.0
Natural conditions on the 20% most impaired days
3.0 -
0.0
2050
2060
2000
2010
2020
2030
2040
Figure 1. Example diagram showing the important parameters used to calculate the visibility
metrics for the Regional Haze Rule
The 2017 Regional Haze Rule also requires states to determine the baseline (2000-2004)
visibility condition for the 20 percent clearest days and requires that the long-term strategy and
RPG ensure no degradation in visibility for the clearest days since the baseline period.
2.0 Ambient Data Analysis
Among the requirements described in 40 CFR 51.308(f)(l)(i)-(vi), states must calculate the
following tracking metrics using available ambient monitoring data (states typically use data
collected by the IMPROVE monitoring program):
•	Baseline, current, and natural visibility conditions for the 20 percent most
anthropogenically impaired and 20 percent clearest days. These six conditions must
be quantified in deciviews.
•	The URP between the baseline visibility condition for the most impaired days and
the natural visibility condition for the most impaired days. The URP must be
quantified as the visibility improvement (in deciviews per year) that would need to
be maintained during each implementation period in order to attain natural visibility
conditions by 2064. The rule also allows states to propose an optional adjustment to
the URP to account for impacts from anthropogenic sources outside the U.S. and
from certain wildland prescribed fires.
The 1999 rule text defined "most impaired days" and "least impaired days" by referring to
highest and lowest levels of "visibility impairment" caused by manmade air pollution. The 1999
final rule preamble stated that the least and most impaired days were to be selected as the
monitored days with the lowest and highest actual deciview levels, respectively, without
4

-------
distinguishing between the natural and anthropogenic contributions to reduced visibility.8 In
2003, EPA issued guidance describing in detail the steps for selecting and calculating light
extinction on the "worst" and "best" visibility days (EPA, 2003a). Consistent with the 1999 final
rule preamble, the 2003 guidance recommended that states determine the most and least
impaired days based on which days had the highest and lowest overall deciview values, rather
than determining and selecting the days with the highest and lowest anthropogenic
impairment. However, because natural haze due to wildfires or dust storms can be larger than
anthropogenic impairment for some Class I areas (particularly in the western U.S.), this
approach resulted in some days with large natural sources of haze being included in the 20
percent most impaired days metric.
The 2017 Regional Haze Rule defines visibility impairment or anthropogenic visibility
impairment as "any humanly perceptible difference due to air pollution from anthropogenic
sources between actual visibility and natural visibility on one or more days. Because natural
visibility can only be estimated or inferred, visibility impairment also is estimated or inferred
rather than directly measured."9 In this definition, the Regional Haze Rule's definition of
visibility impairment is synonymous with anthropogenic impairment. A metric that reflects both
the fraction of the actual light extinction that is above natural levels (in Mm"1) as well as the
logarithmic relationship between light extinction and perceived visibility is, thus, a logical basis
for selecting the 20 percent most anthropogenically impaired days. One such metric is the
difference (the "delta deciviews") between the total deciview value that exists (or is projected
to exist) and the deciview value that would have existed if there were only natural sources
causing reduced visibility. This is the metric that EPA recommends be used. We recommend
that states use Equation 2 to calculate anthropogenic visibility impairment:
^ ^anthropogenic visibility impairment - dvtota|-dvnatura|	(Eqn. 2)
where dvtotai is the overall deciview value for a day, and dvnaturai is the natural portion of the
deciview value for a day The Regional Haze Rule does not specify how dvtotai and dvnaturai are to
be determined; that is the subject of some of EPA's recommendations in this document.
2.1 Recommendations for estimating daily natural and anthropogenic visibility fractions
and light extinction budgets and calculating the 20% most impaired and 20%
clearest days
The first step in determining dvnatura| is to split the daily light extinction into natural and
anthropogenic fractions. Because these are not directly measured, a statistical or
computational method must be used to estimate these fractions. This guidance document
presents EPA's current recommendation for estimating these fractions; data for this
recommended approach, as well as the results of applying the recommended approach,10 will
8See 64 FR 35728.
9See 40 CFR 51.301.
10A state that wishes to follow the EPA-recommended approach may download these completed results and will
not have to itself execute the 7 steps discussed in this section. These completed results are available at
httpi//vista,cira,colostate,edu/lmprove/rhr-summary-data.
5

-------
be provided by the IMPROVE program to states for their use on an annual basis. EPA may
provide refinements to this method or additional recommended methods through additional
guidance as such methods become available.
In general, the recommended approach to splitting daily light extinction into natural and
anthropogenic fractions is to estimate the natural contribution to daily light extinction and then
attribute the remaining light extinction to anthropogenic sources. The natural contributions are
grouped into two types - "episodic" and "routine." Episodic natural contributions are those
that occur relatively infrequently. These may differ in number and size from year to year and
likely result from extreme events. Routine natural contributions are those that occur on all or
most days in a year or season and are more consistent from year to year. Large wildfires and
strong dust storms are examples of episodic natural contributions from extreme events, while
biogenic secondary aerosol is an example of a routine natural contribution.11 It is useful to
make this distinction because the values used by most states in the first implementation period
to represent natural visibility conditions, the "NC-II" estimates,12 are generally recognized as
representing long-term averages influenced by routine natural sources but not episodic natural
sources (EPA, 2003b). As explained below, the annual average NC-II estimates are used in the
recommended method described in this section, but in a manner that is consistent with the
premise that they represent only the influences of routine natural sources.
The recommended steps (1 through 7) to estimate natural and anthropogenic light extinction
and the 20 percent most impaired days for the year are detailed below, using an example for
Mesa Verde National Park (MEVE1). Note that the values throughout this example are unique
to MEVE1 and have been included for illustrative purposes only. Each Class I area is treated
individually, and these values do not apply to any site other than MEVE1. A flow chart
summarizing these steps is shown in Figure 2.
nThe EPA recognizes that natural emissions can also include volcanic emissions. The approach described in this
guidance document does not attempt to account for haze formed from natural volcanic emissions. We encourage
states with Class I areas affected by volcanic emissions to work with their EPA Regional office to determine an
appropriate approach for determining which days are the 20 percent most anthropogenically impaired days.
12"NC-II" refers to a set of estimates of natural conditions for each Class I area contained in Regional Haze Rule
Natural Level Estimates Using the Revised IMPROVE Aerosol Reconstructed Light Extinction Algorithm, available at
httpi//vista,cira,colostate,edu/improve/publications/graylit/032 NaturalCondllpaper/Copeland etal NaturalCondi
tionsll Description,pdf. As called for in the agency's "Regional Haze Reform Roadmap" (September 11, 2018, the
agency may be updating the natural visibility conditions estimates in spring 2019, as necessary.
6

-------
Step 1	Step 2	Step 3
Assign daily
carbon and dust
tight extinction
above thresholds
to natural
(episodic)
Calculate lowest
95th percentile
carbon and dust
light extinction
from 2000-2014
to identify
extreme events
Reallocate
remaining daily
carbon and dust
light extinction
to original
components
Step 7	Step 6	Step 5 \ Step 4
Sort each year's
days by the daily
anthropogenic
impairment and
select the 20%
most impaired
days
Calculate the daily
anthropogenic
impairment using
daily estimates
of anthopogenic
and natural light
extinction
Consider the
remaining light
extinction from
all components
besides sea-salt
as anthropogenic
Calculate natural
(routine) fraction
of unassigned
components based
on non-episodic
annual averages,
NC-II estimates
Figure 2. Flow chart of the 7 steps involved in calculating the 20% most impaired days.
Step 1: Establish light extinction thresholds to identify extreme events
Analysis from the first implementation period showed that smoke from wildfires (mainly
composed of OMC and LAC) and dust storms (mainly composed of CM and FS) are major
contributors to light extinction at many Class I areas (Tombach, 2008). For each Class I area,
using data from the IMPROVE monitor associated with the area, identify for each year the 95th
percentile 24-hour carbon (OMC + LAC) light extinction (Spracklen, et al., 2007) (Jaffe, et al.,
2008). Choose the year between 2000 and 2014 with the lowest such value. This year
represents the "low wildfire" year of this period. Also, choose the year with the lowest 95th
percentile 24-hour dust (CM + FS) light extinction. This year represents the "low dust storm"
year of this period. The 95th percentile carbon and dust values for these years will serve as the
threshold values used to identify impacts on carbon and dust light extinction from extreme
episodic events in that year and other years. At Class I areas where episodic influences vary
significantly from year to year, it will not be unusual for more than five percent of the
monitored days to be affected by extreme episodic events in years other than the "low
wildfire" and "low dust storm" years. Thus, this approach allows a different number of high
carbon days or high dust days in different years to be identified as ones with extreme episodic
impacts, but all the days that are identified will have carbon or dust concentrations at least as
high as the respective threshold. EPA believes this method for calculating threshold values for
identifying episodic light extinction is reasonable and practical to apply to the large set of
IMPROVE data, and our investigations have indicated that the results (i.e., the days selected as
the 20 percent most impaired) would not be substantially different if slightly different
percentile values were used for this purpose. However, some areas with a high frequency of
episodic wildfire smoke or dust impacts even in the "low wildfire" or "low dust storm year"
could use a lower percentile value. At other sites, the year representing the lowest wildfire or
dust storm thresholds may have no episodic impacts and such sites could use a higher
7

-------
percentile value. States may use other reasonable thresholds for determining impacts from
extreme natural events if they explain why another method is appropriate for their individual
Class I areas.
The 95th percentile value will be the 0.95xn measured value sorted from lowest to highest. If
0.95xn is not an integer value, the 95th percentile value is the monitored value such that it and
all lower values are more than 95 percent of the sample. For MEVE1 in 2003, there were 105
complete values for carbon. For MEVE1 in 2003, 0.95x105 is 99.75, so the 95th percentile value
would be the 100th value counting up from the lowest value, out of 105 (the sixth value
counting down from the highest value). In 2003, the 100th highest carbon value is 25.36 Mm"1.
Repeat this process to get a 95th percentile value for each year from 2000 to 2014 for carbon
and dust. The results for each of these years for MEVE1 are shown in Table 1.
Table 1. 95th percentile values for carbon and dust light extinction (in units of Mm1) from
2000-2014 at MEVE1.
Year
Annual 95th percentile carbon
light extinction
Annual 95th percentile dust
light extinction
2000
12.68
7.732
2001
7.002
6.686
2002
16.14
19.60
2003
25.36
16.45
2004
5.937
5.498
2005
9.640
5.658
2006
7.813
5.326 (lowest)
2007
11.72
5.685
2008
7.545
9.257
2009
10.55
10.35
2010
7.109
13.30
2011
5.289
9.726
2012
10.66
8.930
2013
5.396
8.223
2014
5.054 (lowest)
9.281
The years 2014 and 2006 have the lowest carbon and dust 95th percentile values for MEVE1,
respectively. The year 2014 was in this sense the "low wildfire" year at MEVE1, such that the
95th percentile value for carbon in 2014 becomes the threshold for identifying extreme wildfire-
affected days in any year, with the same concept applying to dust from natural sources and the
year 2006. The 95th percentile value of carbon in 2014 was 5.054 Mm"1, and the 95th percentile
value of dust in 2006 was 5.326 Mm"1. National maps of the light extinction thresholds to
identify extreme carbon and dust events are shown in Figures 3 and 4.
8

-------
carbon e3 (Mm1)
Figure 3. Site-specific extinction thresholds of carbon from extreme episodic events
Figure 4. Site-specific extinction thresholds of dust from extreme episodic events
9

-------
Step 2: Assign the portions of daily carbon and dust light extinction that are in excess of these
thresholds to "natural (episodic)"
The IMPROVE light extinction data for a specific day (May 12, 2003) at MEVE1 are shown in
Table 2. The light extinction from carbon (25.36 Mm"1) on this day is greater than the threshold
of 5.054 Mm"1. Therefore, 25.36 minus 5.054 or 20.31 Mm 1 is assigned to natural (episodic).
Carbon light extinction in the amount of 5.054 Mm 1 remains to be split between natural
(routine) and anthropogenic in Step 4 below, after the combined value of carbon is reallocated
to OMC and LAC (Step 3). The dust-related light extinction of 2.699 Mm 1 is less than the
threshold value of 5.326 Mm"1, therefore no dust-related light extinction is assigned to natural
(episodic). However, the 2.699 Mm 1 value for dust-related light extinction does need to be split
between natural (routine) and anthropogenic in Step 4 below, after the combined value of dust
is reallocated to FS and CM (Step 3). A summary of the episodic thresholds and the light
extinction assigned to natural (episodic) is shown in Table 2.
Step 3: Reallocate the daily combined carbon and dust light extinction remaining after
assigning values over the threshold values to "natural (episodic)" into OMC, LAC, FS, and CM.
Separate the combined carbon back into OMC and LAC and separate the combined dust back
into FS and CM based on the original percentages of the individual PM species to the grouped
light extinction. For example, at MEVE1 on May 12, 2003, the total carbon light extinction was
25.36 Mm"1, with OMC light extinction of 21.78 and LAC Mm"1 light extinction of 3.576 Mm"1.
The total dust light extinction was 2.699 Mm"1 with 1.141 Mm"1 from FS and 1.558 Mm"1 from
CM. Therefore, on May 12, 2003, carbon light extinction was 85.88 percent from OMC and
14.10 percent from LAC; dust light extinction was 42.27 percent from FS and 57.72 percent
from CM. Separate the estimates of natural (episodic) and the remaining light extinction from
carbon back into OMC and LAC and the remaining light extinction from dust back into FS and
CM, using these percentages. Table 2 shows the results of these calculations for this example
day at MEVE1. For example, of the 20.31 Mm1 of carbon assigned to natural (episodic), 17.44
Mm4 (or 85.88 percent) is reallocated to OMC and 2.864 Mm1 (14.10 percent) is reallocated to
LAC. States may use other approaches than the one recommended here.
10

-------
Table 2. Total and speciated light extinction (in units of Mm1) for an example day (May 12,
2003) at MEVE1
PM Species
Total Light
Threshold
Light extinction
Light extinction

Extinction

associated with
natural (episodic)
remaining after
episodic treatment
Sulfate
2.963
NA
0
2.963
Nitrate
0.8055
NA
0
0.8055
Carbon
25.36
5.054
20.31
5.054
(OMC + LAC)
(21.78 + 3.576)

(17.44 + 2.864)
(4.340 + 0.7126)
Dust
2.699
5.326
0
2.699
(FS + CM)
(1.141+1.558)


(1.141+1.558)
Sea salt
0.001469
NA
0
0.001469
Rayleigh
9.0
NA
0
9.0
TOTAL
40.83
NA
20.31
20.52
Step 4: Split the remaining components of light extinction into natural (routine) and
anthropogenic based in part on the NC-II estimates (Copeland, et al., 2008).
In order to split the remaining components of light extinction into natural (routine) and
anthropogenic components, we recommend allowing the natural (routine) to vary daily based
on the NC-II estimates weighted by the ratio of the light extinction remaining after removing
episodic contributions to the non-episodic annual average extinction. At most Class I areas, the
use of this ratio results in higher natural (routine) values in the summer and lower values in the
winter.
Starting with the daily results from Step 3 for all days with complete data in a year, calculate
the annual average light extinction values for each PM species, excluding light extinction
already attributed to episodic events. The non-episodic annual averages for 2003 at MEVE1 are
shown in Table 3.
For all PM species except sea salt (which is treated as entirely natural (routine)), use the
existing NC-II annual average natural light extinction values (which are distinct from the "p90"
values), the daily light extinction values, and the annual averages for the site (for both,
excluding the light extinction already attributed to episodic events) to calculate a daily estimate
of natural (routine). These three input values appear in Table 3 for the May 12, 2003, MEVE1
example. (Table 3 is not intended to show a calculation using these values. The way these input
values are used to complete the calculation of anthropogenic versus natural light extinction for
a given day is described below the table.)
11

-------
Table 3. The remaining light extinction at MEVE1 on May 12, 2003 (in units of Mm1), after
applying thresholds to allocate some light extinction to natural (episodic). The NC-II average
light extinction estimates and the 2003 annual average light extinction excluding the episodic
light extinction are also shown.
PM
Light extinction
NC-II
2003 annual
Natural
Anthropogenic
species
remaining after
average
average non-
(routine) light
light extinction

episodic
natural light
episodic light
extinction
(Step 5)

treatment (Step 3)
extinction
extinction
(Step 4)

Sulfate
2.963
0.5741
4.103
0.4146
2.548
Nitrate
0.8055
0.5829
1.599
0.2936
0.5119
OMC
4.340
1.831
3.194
2.489
1.851
LAC
0.7126
0.2
0.8624
0.1653
0.5467
FS
1.141
0.5
0.8793
0.6490
0.492
CM
1.558
1.726
2.351
1.144
0.4140
Sea salt
0.001469
0.01822
0.02806
0.001469
0
Rayleigh
9.000
9.000
9.000
9.000
0
TOTAL
NA
NA
NA
14.16
6.36
The calculation formula in Step 4 using daily and annual average inputs like those shown in
Table 3 depends on whether, for a given PM species, the annual average light extinction value
(excluding episodic events) for the particular year is greater than or less than the NC-II estimate
of annual average natural light extinction. For a site and PM species with an annual average
light extinction value (excluding episodic events) less than the NC-II estimate, all of the daily
light extinction is assigned to natural (routine). This results in the natural (routine) light
extinction being different each day, with the annual average being less than the NC-II estimate.
For a site and PM species with an annual average light extinction value (excluding episodic
events) greater than the NC-II estimate, the daily estimates of natural (routine) light extinction
are calculated by multiplying the total daily light extinction for each species by the ratio of the
NC-II annual average estimates and the annual average non-episodic light extinction. This
results in the natural (routine) light extinction being different each day and the annual average
of the daily estimates of natural (routine) light extinction equaling the NC-II annual average
value. The daily contributions to natural (routine) are calculated according to Equation 3:
. . daily extinction x NC-II estimate	.
natural(routine)=		(Eqn. 3)
non-episodic annual average
An example for the OMC light extinction on May 12, 2003, at MEVE1, using extinction values
from Table 3, is shown below.
,, . .	4.340x1.831	1
natural(routine)OMC=——— = 2.489 Mm 1
Repeat this calculation for LAC, FS, CM, sulfate, and nitrate light extinction (not shown here).
States may use other reasonable methods for estimating routine natural and anthropogenic
fractions if they explain why another method is appropriate for their individual Class I areas.
12

-------
Step 5: Consider the remaining light extinction from sulfate, nitrate, OMC, LAC, FS, and CM as
"anthropogenic."
Starting with the daily total light extinction measured on each day, subtract the daily natural
(episodic) and daily natural (routine) to find the daily anthropogenic light extinction attributable
to each PM species and overall, i.e., the light extinction budget. For the May 12, 2003, MEVE1
example, the daily total light extinction was 40.83 Mm"1, daily natural light extinction was 34.46
Mm 1 (20.31 episodic + 14.16 routine), and daily anthropogenic light extinction was 6.36 Mm 1
(see Tables 2 and 3).
Step 6: Calculate anthropogenic impairment for each day using the daily estimates of natural
and anthropogenic light extinction, according to Equation 2.
For each day at the Class I area of interest, convert the daily total and natural light extinction to
deciviews and calculate anthropogenic impairment according to Equation 2. At MEVE1, for May
12, 2003, the anthropogenic impairment is calculated as:
40.83	34.46
A dvanthr0p0genic visibility impairment = 10 x In — 10 X In	= 1.695 dv
Step 7: Sort each year's days with complete data by the anthropogenic impairment value and
choose the 20 percent most impaired days based on this value.
Perform these calculations for each day at the Class I area of interest, then rank the days within
each year from high to low by anthropogenic impairment where a rank of 1 is the most
impaired day (i.e., the day with the highest anthropogenic impairment value). At MEVE1, this
day, May 12, 2003, with an anthropogenic impairment value of 1.695 deciviews, is a relatively
low impairment day and was ranked 99 out of 105 total days with complete observations.
Therefore, based on anthropogenic impairment, this day is not one of the 20 percent most
impaired days for 2003.13 Average the deciviews of total haze on the 20 percent most impaired
days for each year to obtain a single value for the associated visibility condition for each year
(for MEVE1 in 2003, which had 105 complete observations, 21 days will be in the 20 percent
most impaired).
States may choose alternative approaches for estimating natural and anthropogenic
contributions to light extinction, but the Regional Haze Rule requires states to choose the 20
percent most impaired days based on anthropogenic impairment. In other words, while Steps 1
through 6 described above are EPA recommendations and states are not precluded from using
other approaches to determine the anthropogenic impairment on each day, the Regional Haze
Rule requires states to follow Step 7 as it is described here.
The 2017 Regional Haze Rule revisions introduced a new term for describing the days with the
lowest light extinction and deciview values: clearest days. These days are not to be selected
13In contrast, if ranking this day based on either total light extinction or overall visibility conditions (the ranking
would be the same with these two metrics), as the EPA's guidance for the first implementation period
recommended, this day would be ranked 14 out of 105 days with complete observations and would be one of the
20 percent of days with the haziest visibility conditions.
13

-------
based on the lowest anthropogenic impairment (as referring to them as the 20 percent least
impaired days as in the 1999 Regional Haze Rule would suggest). These will be the days with the
lowest values of the deciview index. It is unnecessary to split the data into "natural" and
"anthropogenic" fractions. Rather, the days are to be sorted for each year by total deciviews,
and the 20 percent of days with the lowest deciviews are the 20 percent clearest days. When
0.2 multiplied by the number of monitored days in a year with complete data is not an integer,
an "extra" day should not be included in the set of clearest days, which means that the
percentage of days in this set may be a value somewhat below 20 percent.
2.2 Calculating the baseline, current, and natural visibility conditions
The 2017 Regional Haze Rule continues to define the period for establishing baseline visibility
conditions as 2000 to 2004 for the second and future implementation periods.14 Visibility
conditions averaged over these 5 baseline years are the starting point for calculating the URP
and drawing the URP line for all implementation periods of the Regional Haze Rule. It is
important to note that in the 2017 Regional Haze Rule, the term "most impaired days" has a
different meaning than EPA and states gave to that term in the first implementation period. The
"baseline visibility condition (in deciviews) for the 20 percent most impaired days" in a state's
SIP submission for the second implementation period will likely have a different value than the
baseline values used in SIPs for the first implementation period, even if there have been no
revisions to the IMPROVE data for the 2000-2004 period. The differences will be largest at Class
I areas impacted by fire and dust events in the baseline period. If a state chooses an alternative
approach for estimating natural and anthropogenic contributions to light extinction, it is likely
that the baseline visibility condition will have a different value than with the recommended
approach.
The period for calculating current visibility conditions in the 2017 Regional Haze Rule is the 5-
year period ending with the most recently available data. Due to the laboratory, data analysis,
and quality assurance procedures of the IMPROVE program, there is some delay between the
date of the filter collection and the date the data are ready for use in analyses. Current visibility
conditions must be calculated based on the annual average level of visibility impairment for the
20 percent most impaired and the 20 percent clearest days. The current visibility condition for
each set of days is the average of the valid annual values from the 5-year period ending with
the most recently available data set as expressed in deciviews. Five years are averaged to
account for variability in meteorology and emissions. Data completeness requirements for valid
years are described in the 2003 Regional Haze Rule visibility tracking guidance (EPA, 2003a).
Incomplete or missing data from some IMPROVE sites may require the combination or
substitution of data from multiple IMPROVE sites for the ongoing visibility tracking of the
Regional Haze Rule. The appropriate EPA Regional office should be consulted when data
completeness issues arise.
14lt is recommended that the data for the 2000-2004 baseline period be refreshed prior to analysis due to periodic
revisions in the methods for calculating ambient concentrations from measurements made on filters and for filling
in missing or invalidated data.
14

-------
The URP framework requires states to determine a single value for the "natural visibility
condition" for the 20 percent most impaired days. Given the inherent day-to-day variability of
natural processes (e.g., windblown dust, fire, volcanic activity, biogenic emissions, etc.), it
follows that even if there were no anthropogenic sources, visibility would not be constant and
would vary day-to-day. Also, visibility due only to natural sources has never occurred in modern
times and, therefore, has never been directly measured nor could it be directly measured. It
must be estimated. Even if past natural conditions could be known with certainty, future
natural conditions may be different. The steps for estimating natural and anthropogenic
fractions of light extinction recommended in this guidance are based on estimates of natural
visibility conditions for each monitored day in the past, with a given past day having the
potential for both routine and episodic contributions to natural conditions. An additional step is
needed to get the single value for the "natural visibility condition." Under the Regional Haze
Rule, the single value of the natural visibility condition for the 20 percent most impaired days is
used in several ways:
1.	The value of the natural visibility condition is to be compared to the "current
visibility condition," i.e., the most recent 5-year average of actual visibility for the 20
percent most impaired days. (51.308(f)(l)(v)).
2.	The URP can be calculated as the difference between the 2000-2004 baseline
visibility condition and the natural visibility condition for the 20 percent most
impaired days, divided by 60 years. In other words, the "glidepath" can end at the
natural visibility condition in 2064. (51.30S(f)(l)(vi)).15
3.	The future year (2028 for the second implementation period) RPG for the 20 percent
most impaired days is compared to its value on the URP line, which can use the
natural visibility condition estimate as its endpoint. (51.308(f)(l)(vi)).
We are recommending that states set the single value of the natural visibility condition for the
20 percent most impaired days to be equal to the average of the new estimates of daily natural
visibility conditions estimated for the particular days that have been identified as the 20
percent most impaired days from 2000-2014. This method takes advantage of the already
calculated daily "natural (episodic)" and "natural (routine)" estimates of light extinction
produced in steps 1 through 3. These revised natural visibility conditions are consistently lower
in magnitude than the "p90" NC-II haze estimates (representing the average conditions for days
between the 80th percentile and the 100th percentile) and generally more similar in magnitude
to the annual average NC-II haze estimates. (Gantt, et al., 2018) describes in greater detail the
methodology, seasonality, composition, and trends in the natural visibility conditions, and a
summary of the natural visibility condition estimates for each IMPROVE site can be found in
Appendix A.
When following the recommended approach to select the 20 percent most impaired days
based on anthropogenic impairment, days with large impacts from extreme, episodic natural
events such as fires and dust storms are no longer selected. Therefore, these extreme impacts
15lf an adjustment is made to the URP for impacts from international anthropogenic emissions or wildland
prescribed fires, the glidepath would not end at the natural visibility condition.
15

-------
generally will riot be included in estimates of natural visibility conditions that will be compared
with the most impaired days. This addresses past feedback from the first implementation
period that the natural and current visibility conditions were inconsistent because 1) the "p90"
NC-II Natural Haze estimates developed by the Natural Haze Levels II Committee in 2007 failed
to include effects from large episodic natural events (Tombach, 2008) and 2) the 20 percent of
days with the worst overall visibility included these extreme episodic natural influences. In
addition to avoiding selecting historical days dominated by extreme natural events when
calculating the single value of the natural visibility condition that will be used in calculating the
URP, it is important to recognize that the 20 percent most impaired days will be distributed
across seasons of the year differently than the 20 percent haziest days used for the first
implementation period (Gantt, et a!., 2018). Because the revised national visibility condition
value is calculated from estimates of daily natural contributions on the most impaired days, this
recommended natural visibility condition value is more consistent within the framework.
At most Class I areas, these updates to the baseline and natural visibility conditions in
recommended approach result in a time series that is less influenced by natural events and
reflects the substantial visibility improvements that have occurred in many areas of the United
States between 2000 and 2016 (see Figure 5).
Sawtooth
Annual Avg.
5-yr Avg.
Glidepath
® 15 -
2000 2004 2008 2012 2016
Year
deciview deviation
¦
0 -3.0 -2.0 -1,0 0 1.0 2.0 3.0 4.0
Mesa Verde
Guadalupe Mtns
Shenandoah
30.0 -
27.0 -
-------
3,0 Adjustment of the Uniform Rate of Progress for Impacts from
International Anthropogenic Emissions
Visibility at Class I areas is impacted not only by natural and anthropogenic emissions from
within the U.S., but also by natural and anthropogenic international emissions. Due to the fact
that international anthropogenic emissions are beyond the control of states preparing regional
haze SIPs, the Regional Haze Rule allows states to propose an adjustment of the 2064 URP to
account for international anthropogenic impacts, if the adjustment has been developed using
scientifically valid data and methods. The URP can be adjusted by adding an estimate of the
visibility impact of international anthropogenic sources to the value of natural visibility
condition to get an adjusted 2064 endpoint.16
The optional adjustment to the URP for international anthropogenic emissions is in addition to
another optional adjustment relating to certain prescribed fires. Specifically, the rule also
allows states to include an adjustment of the URP to account for impacts from certain wildland
prescribed fires. The information and procedures for prescribed fire adjustments are expected
in practice to be similar to the recommended international anthropogenic adjustment
procedure provided in this guidance. Therefore, this section of the guidance document may be
useful in either case. Note that preliminary EPA regional haze modeling (based on 2011 fire
emissions) (EPA, 2017) indicates that prescribed fire impacts on the 20 percent most impaired
days at most Class I areas are likely to be small and, thus, any adjustment to the URP would also
be small. This section focuses on anthropogenic international emissions and their impacts.
3.1 URP Adjustment in the Regional Haze Rule
The relevant international anthropogenic URP adjustment language in the Regional Haze Rule is
at 40 CFR 51.308(f)(l)(vi):
(B) As part of its implementation plan submission, the State may propose (1) an
adjustment to the uniform rate of progress for a mandatory Class I Federal area to
account for impacts from anthropogenic sources outside the United States and/or (2) an
adjustment to the uniform rate of progress for the mandatory Class I Federal area to
account for impacts from wildland prescribed fires that were conducted with the
objective to establish, restore, and/or maintain sustainable and resilient wildland
ecosystems, to reduce the risk of catastrophic wildfires, and/or to preserve endangered
or threatened species during which appropriate basic smoke management practices
were applied. To calculate the proposed adjustment(s), the State must add the
estimated impact(s) to the natural visibility condition and compare the baseline visibility
16The EPA expects that the revised approach of selecting the most anthropogenically impaired days for purposes of
defining RPGs and tracking progress, which focuses progress tracking on days not affected by large episodic natural
events such as dust storms and wildfires, will also largely resolve any concerns stemming from the same types of
natural emission sources in other countries. Because the recommended method for identifying the most
anthropogenically impaired days is based entirely on information from IMPROVE monitoring sites, it can be
executed without detailed information on the emissions from natural sources outside the U.S.
17
-------
condition for the most impaired days to the resulting sum. If the Administrator
determines that the State has estimated the impact(s) from anthropogenic sources
outside the United States and/or wildland prescribed fires using scientifically valid data
and methods, the Administrator may approve the proposed adjustment(s) to the
uniform rate of progress.
The URP is based on 1) PM species measurements that do not distinguish between PM due to
natural, U.S. anthropogenic, and international anthropogenic emissions, and 2) the "natural
visibility condition" endpoint that should not include any anthropogenic contribution (see
Figure 1). The natural visibility condition endpoint, therefore, assumes no anthropogenic
international contribution in 2064 and the default URP slope reflects a hypothetical uniform
decrease in both U.S. and international anthropogenic haze contributions. The rule provision
that allows states to include an international adjustment allows for the modification of the URP
slope to account for international anthropogenic contributions that states cannot control.
The Regional Haze Rule allows for an adjusted URP but does not prescribe a particular
adjustment methodology. To inform this adjustment to the URP, EPA recommends the use of
chemical transport models (CTMs) as the most broadly applicable method for attributing
pollutant concentrations to emissions sources. Two key issues with using CTMs for this purpose
are addressed below: what year should be used to estimate international anthropogenic
impacts, and how to apply the models to quantify international anthropogenic impacts.
3.2 Year Selection for Estimating International Contribution
Estimating international anthropogenic visibility impact is a function of transport patterns and
emissions. Both meteorology and emissions are year-specific, so the first choice in
photochemical modeling is determining what year to simulate. For example, the estimation
could be based on a current or recent year, the implementation period end year (e.g., 2028,
2038, etc.), or the URP endpoint (2064).
To illustrate the potential impacts of international emissions, Figure 6 shows hypothetical
effects of adjusting the 2064 endpoint to account for international emissions at a hypothetical
Class I area. The URP lines in Figure 6 illustrate how the URP slope is impacted by different
estimates of international impacts. The URP represented by the orange line shows an
unadjusted URP that assumes both U.S. and international anthropogenic impacts will decrease
uniformly to zero in 2064. The black URP line represents an adjusted URP, assuming constant
international anthropogenic impacts over time. In both URP series, the U.S. anthropogenic
contribution is uniformly decreasing in each period.17
17ln the example figure, the U.S. visibility impairment improvement is calculated as a fixed percentage in each
implementation period. However, consistent with the Regional Haze Rule, there is no regulatory requirement to
achieve "uniform progress." The actual improvement in visibility impairment during each implementation period
may vary.
18
-------
NAT MUSA HINT ^Unadjusted URP ^^Adjusted URP
Figure 6. URP lines based on alternate international projections. For each year except 2004,
there are two bars representing two different possibilities (1st: decreasing international (INT);
2nd: constant international). Note that both realizations have the same decreasing U.S.
anthropogenic (USA) contribution and same constant natural conditions (NAT).
Projecting international emissions to 2064 may be speculative and somewhat uncertain. EPA
therefore believes that recent year estimates of visibility impacts are most appropriate to use in
estimating international adjustments. There are existing recent estimates of global PM and PM
precursor emissions that have been used in various modeling studies (Galmarini, et al., 2016;
Janssens-Maenhout, et al., 2015; Li, et al., 2015; Hoesly, et al., 2018). Therefore, for the second
implementation period, EPA recommends estimating international impacts in a recent year,
with more recent years being best able to reflect current international emissions and trends. In
choosing the analysis year, additional practical considerations, such as the availability of
emission information or modeling results may constrain a state's options. For example, it may
only be practical to use 2011, 2014, or 2016 base year international emissions to calculate the
adjustment because only that information is readily available.
In some cases, additional data on emissions trends may be used to estimate future year
international impacts, if future changes are well known. This is especially relevant for North
American emissions sources (generally within the regional modeling domain), where future
trends in some non-U.S. emissions sectors (e.g., on-road mobile sources and commercial ships)
19
-------
may be relatively well characterized in existing inventories. Since 2028 is the end of the second
implementation period, it is the most likely future modeling analysis year. To the extent that
high-quality international emissions projections are available, it may be appropriate to calculate
international anthropogenic adjustments based on 2028 emissions.
Since international anthropogenic visibility impacts are likely to change in the future, in
subsequent planning periods, a new international adjustment can be made for each
implementation period. In this way, an iterative process over time allows the glidepath to be
periodically adjusted to reflect future trends in international anthropogenic emissions.
Eventually, as we get closer in time to the URP endpoint, the international anthropogenic
visibility impact will become less uncertain.
3.3 Estimating the Anthropogenic International Visibility Impacts
The methods to quantify international visibility impacts are largely independent of the chosen
year. The Regional Haze Rule requires that a state's approach be based on scientifically valid
data and methods. Due to long-range transport and secondary PM components involved in
international transport, photochemical chemical transport modeling (CTM) is the preferred
approach for quantifying international contributions to visibility. Detailed guidance on
performing CTM simulations is available in EPA's photochemical modeling guidance for ozone,
PM2.5, and regional haze (EPA, 2018).
Using CTMs, there are several potential ways to quantify international anthropogenic impacts
in Class I areas:
1)	The simplest approach is to perform brute force "zero-out" model runs, which involves
at least two model runs: one "base case" run with all emissions, and one with
anthropogenic emissions from outside of the U.S. removed from the original base case
simulation. The difference between these simulations provides an estimate of the air
quality impact due to the international anthropogenic emissions.
2)	An alternative approach to isolating international anthropogenic impacts in
photochemical grid models is "photochemical source apportionment." Some
photochemical models have been developed with a photochemical source
apportionment capability, which tracks emissions from specific sources or groups of
sources and/or source regions through chemical transformation, transport, and
deposition processes to estimate the apportionment of predicted PM2.5 species
concentrations (Kwok, et al., 2013; Kwok, et al., 2015; Ramboll, 2018). Source
apportionment can be used to track PM formed from international anthropogenic
emissions sources.
From the CTM runs, whether based on brute-force or source apportionment (or a combination
of both), PM species concentration increments due to international anthropogenic emissions
can be calculated for each of the 20 percent most impaired days (for each Class I area). The PM
concentration increments on the 20 percent most impaired days are then converted to
20
-------
extinction and averaged. The "delta deciview" adjustment factor associated with the
international anthropogenic impact can then be calculated as follows18:
Adv-10 ln(bextnatura| conditions"*"bextjn(erna(jona| anthropogenic)/ bextnatura| conditions	(Eqn. 4)
The Class I area-specific adjustment factor (in deciviews) is added to the natural conditions
value for each Class I area to get the adjusted URP.19 In this process, the following issues should
be given appropriate treatment:
The air quality model (or combination of models) that is used.
The modeling system typically includes a global/hemispheric modeling simulation and a
regional photochemical modeling simulation. The global component is often used to supply
"boundary" conditions to the regional simulation. To the extent practical, the modeling
platforms with the two scales should be consistent using the same (or similar) meteorology,
vertical resolution, emissions, and representation of chemical species. This is particularly true
for the international emissions but is also true for gas-phase and aerosol modeling components.
The consistency is particularly important when performing anthropogenic zero-out and/or
source apportionment simulations that cross horizontal grid scales (going from a lower
resolution global/hemispheric model to a higher resolution nested regional model).20
The validity of the estimates of both international and U.S emissions.
Before estimating source contributions, the "basecase" simulation, both for global and regional
models, should be able to reasonably reproduce historical PM measurements and calculated
visibility values. Thus, model performance evaluation and diagnostic evaluation should both
play a role as described in the photochemical modeling guidance (EPA, 2018) (Simon, et al.,
2012). This will provide confidence that both U.S. and international emissions and visibility
impacts are reasonably well represented.
After the basecase has been evaluated and shown capable of representing historical PM and
visibility, then the models can be used to quantify international impacts. Quantifying the
international sources, as previously stated, may be done using zero-out (sensitivity) or source
apportionment model runs, or a combination of both.21 Unless the international impacts are
primarily from the portions of North America included in the smaller scale (regional) modeling
simulation, the modeling will likely require coordinated efforts between a global/hemispheric
and regional CTM simulation. When this is the case, it is especially important to understand the
18There may be multiple ways this aggregation across days might be done, keeping in mind that the URP line is in
units of deciviews, so the adjustment must also be in units of deciviews.
19Note that the Regional Haze Rule does not allow a state to subtract an estimate of the impacts of international
anthropogenic sources when projecting the RPGs for the end of the implementation period, as an alternative to
adding international anthropogenic impacts to the 2064 endpoint of the URP glidepath.
2tThe boundary between the global and regional models should be sufficiently far removed from U.S. emissions
sources and/or the Class I areas being considered.
21For example, zero-out modeling could be used in the global or hemispheric model to feed boundary conditions to
the regional model. Then source apportionment technology could be used to track international anthropogenic
emissions within the regional model. The available options depend on the source apportionment capability of the
chosen global, hemispheric, and/or regional CTMs.
21
-------
harmony of inventories at the different scales. Regardless of the technology used for
quantification of source contributions (sensitivity or source apportionment), the most
important aspect is appropriate selection of source sectors and domains over which they apply.
Previous studies provide guidance on which inventories need to be considered in international
contribution analysis. The majority of studies have focused on ozone because of its long-range
transport capability (Zhang, et al., 2011; Emery, et al., 2012). Estimating international
contribution continues to evolve and applications should review all emission sectors and
consider the appropriate divisions between natural, international anthropogenic, and domestic
anthropogenic sources.
When the emissions inventory coverages cross scales (global to regional), the emissions should
be consistent between the two scales and the sensitivity (zero-out) or source apportionment
should also be consistent between the two scales. Emissions that cross scales include aircraft
and international shipping. These inventories require special consideration if they are used in
estimating international contributions. The assignment of domestic versus foreign emissions
depends on the jurisdiction of the waters and/or air space. Where the assignment is unclear,
the appropriate EPA Regional office should be consulted. For brute force modeling, the
"boundary conditions" for the regional perturbation simulation would be provided by a
consistent perturbation in the global model.
Model contributions will vary, and a range of estimates should be considered and discussed to
provide context. Particularly for sensitivity modeling, the sequential order of emission
perturbations influences the result (zeroing the international source or the local source give
different answers). Thus, two estimates of international source contribution can be developed
and used to characterize a range of possible results. This is particularly important for haze that
is strongly influenced by secondary organic aerosol and/or nitrate. For example, nitrate
concentrations can increase when removing international sulfate due to chemical
displacement. Estimates of all species should be characterized and the realism of the estimate
considered before simply adding to natural conditions.
Because the adjustment factor will be added to the endpoint of the URP and ultimately used to
calculate an adjusted URP for comparison with the RPG, it is important that the modeling be
consistent across the URP framework (e.g., the same or similar model, domain, meteorology,
and emissions should be used in both the RPG modeling and in the modeling used to adjust the
endpoint).
3.4 EPA Review of an International URP Adjustment
EPA's approval for a URP adjustment will be part of EPA's review of the full SIP submission for
the second implementation period, and not a separate action in advance of SIP submission. In
this way, EPA's decision to approve or not approve the adjustment will be made in the context
of the complete SIP submission, with public notice and an opportunity to comment. States are
encouraged to consult with their EPA Regional office during the development of any proposed
adjustment approach. Any proposed adjustment must be adequately documented to allow
22
-------
public comment and EPA review. An adequate explanation of the adjustment will necessarily
show the unadjusted and adjusted values for the 2064 endpoint and for the URP.
Whether and what adjustment should be made to the URP to account for impacts from
international anthropogenic emissions will be a new issue for the SIP for each implementation
period. EPA's approval of an adjustment approach included in the SIP revision due in 2021 does
not mean that the same adjustment will be automatically approved if included in the SIP
revision due in 2028, for example.
23
-------
References
Baker, K. R., Emery, C., Dolwick, P. & Yarwood, G., 2015. Photochemical grid model estimates of
lateral boundary contributions to ozone and particulate matter across the continental United
States. ATMOSPHERIC ENVIRONMENT, 12, Volume 123, pp. 49-62.
Baker, K. R. & Kelly, J. T., 2014. Single source impacts estimated with photochemical model
source sensitivity and apportionment approaches. Atmospheric Environment, 10, Volume 96,
pp. 266-274.
Copeland, S., Pitchford, M. & Ames, R., 2008. Regional Haze Rule Natural Level Estimates Using
the Revised IMPROVE Aerosol Reconstructed Light Extinction Algorithm, s.l.:
http://vista.cira.colostate.edu/improve/Publications/GrayLit/032_NaturalCondllpaper/Copelan
d_etal_NaturalConditionsll_Description.pdf.
Dolwick, P. et al., 2015. Comparison of background ozone estimates over the western United
States based on two separate model methodologies. ATMOSPHERIC ENVIRONMENT, 5, Volume
109, pp. 282-296.
Emery, C. et al., 2012. Regional and global modeling estimates of policy relevant background
ozone over the United States. ATMOSPHERIC ENVIRONMENT, 2, Volume 47, pp. 206-217.
EPA, 2003a. Guidance for Tracking Progress Under the Regional Haze Rule, s.l.:
https://www3.epa.gov/ttnamtil/files/ambient/visible/tracking.pdf.
EPA, 2003b. Guidance for Estimating Natural Visibility Conditions Under the Regional Haze Rule,
s.l.: https://www3.epa.gov/ttnamtil/files/ambient/visible/natural.pdf.
EPA, 2017. Documentation for the EPA's Preliminary 2028 Regional Haze Modeling, RTP, NC:
U.S. Environmental Protection Agency .
EPA, 2018. Modeling Guidance for Demonstrating Attainment of Air Quality Goals for Ozone,
PM2.5, and Regional Haze, RTP, NC: U.S. Environmental Protection Agency.
Galmarini, S. et al., 2016. Technical note: Harmonization of the multi-scale multi-model
activities HTAP, AQMEII and MICS-Asia: simulations, emission inventories, boundary conditions
and output formats. Atmospheric Chemistry and Physics Discussions, Volume 2016, pp. 1-20.
Gantt, B., Beaver, M., Timin, B. & Lorang, P., 2018. Recommended metric for tracking visibility
progress in the Regional Haze Rule. Journal of Air & Waste Management, 68(5), pp. 438-445.
Henderson, B. H. et al., 2014. A database and tool for boundary conditions for regional air
quality modeling: description and evaluation. GEOSCIENTIFICMODEL DEVELOPMENT, Volume
7, pp. 339-360.
Hoesly, R. M. et al., 2018. Historical (1750-2014) anthropogenic emissions of reactive gases and
aerosols from the Community Emissions Data System (CEDS). Geoscientific Model Development,
1, Volume 11, pp. 369-408.
24
-------
Jaffe, D. et al., 2008. Interannual Variations in PM2.5 due to Wildfires in the Western United
States. Environ. Sci. Technol, Volume 42, p. 2812-2818.
Janssens-Maenhout, G. et al., 2015. HTAPv2.2: a mosaic of regional and global emission grid
maps for 2008 and 2010 to study hemispheric transport of air pollution. Atmospheric Chemistry
and Physics, Volume 15, pp. 11411-11432.
Kwok, R. H. F., Baker, K. R., Napelenok, S. L. &Tonnesen, G. S., 2015. Photochemical grid model
implementation and application of VOC, NOx, and 03 source apportionment. Geosci. Model
Dev., 1, Volume 8, pp. 99-114.
Kwok, R. H. F., Napelenok, S. L. & Baker, K. R., 2013. Implementation and evaluation of PM2.5
source contribution analysis in a photochemical model. Atmospheric Environment, 12, Volume
80, pp. 398-407.
Lam, Y. F. & Fu, J. S., 2009. A novel downscaling technique for the linkage of global and regional
air quality modeling. Atmos. Chem. Phys., 12, Volume 9, pp. 9169-9185.
Li, M. et al., 2015. MIX: a mosaic Asian anthropogenic emission inventory for the MICS-Asia and
the HTAP projects. Atmospheric Chemistry and Physics Discussions, Volume 15, pp. 34813-
34869.
Pitchford, M. et al., 2007. Revised algorithm for estimating light extinction from IMPROVE
particle speciation data. J. Air & Waste Manage. Assoc., 57(11), pp. 1326-1336.
Ramboll, 2018. User's Guide COMPREHENSIVE AIR QUALITY MODEL WITH EXTENSIONS Version
6.50, Novato: s.n.
Simon, H., Baker, K. R. & Phillips, S., 2012. Compilation and interpretation of photochemical
model performance statistics published between 2006 and 2012. Atmospheric Environment,
Volume 61, pp. 124-139.
Spracklen, D. et al., 2007. Wildfires drive interannual variability of organic carbon aerosol in the
western U.S. in summer. Geophys. Res. Lett., Volume 34.
Tombach, I., 2008. Natural Haze Levels Sensitivity, Assessment of Refinements of Estimates of
Natural Conditions. Prepared for the Western Governors Association, s.l.:
http://www.wrapair.org/forums/aamrf/projects/NCS/Haze_Sensitivity_Report-Final.pdf.
Zhang, L. et al., 2011. Improved estimate of the policy-relevant background ozone in the United
States using the GEOS-Chem global model with 1/2° x 2/3° horizontal resolution over North
America. Atmospheric Environment, 12, Volume 45, pp. 6769-6776.
25
-------
APPENDIX A.
Site
First Implementation Period Approach
Baseline Visibility Current Visibility
Condition for Condition for
P90
20% Haziest Days Most Impaired /nnfiAl
(2000-2004) Days (2012-2016) 1 '
Recommended Approach
Baseline Visibility Current Visibility
Condition for 20% Condition for 20%
Most Impaired Most Impaired
Days (2000-2004) Days (2012-2016)
Derived-
NC
(2064)
e3 (M
carbon
m1)
dust
ACAD1
23.21
16.50
12.43
22.01
15.28
10.39
10.44
3.11
AGTI1
23.50
17.65
7.64
21.62
16.73
7.63
10.85
8.86
BADL1
17.14
14.71
8.06
14.98
12.56
6.09
9.17
7.49
BALD1
11.51
10.19
6.24
8.64
7.55
4.04
6.65
5.42
BAND1
12.23
10.71
6.26
9.70
8.88
4.59
5.60
4.36
BIBE1
17.30
15.29
7.16
15.57
14.25
5.33
7.60
8.59
BLIS1
12.63
12.64
6.05
10.06
9.39
4.91
11.14
2.84
B0AP1
13.80
14.06
6.73
11.61
10.98
5.36
9.38
7.83
BOWA1
19.99
16.51
11.61
18.94
14.58
9.11
11.11
3.16
BRCA1
11.65
9.29
6.80
8.42
6.88
4.08
6.13
4.26
BRID1
11.12
10.84
6.45
7.96
6.58
3.90
7.75
2.83
BRIG1
29.01
21.62
12.24
27.43
20.44
10.69
20.15
9.07
BRIS13

21.29
11.93

19.67
9.28
18.10
9.12
CABI1
14.09
14.27
7.52
10.73
9.97
5.65
13.14
4.13
CACR1
26.36
20.68
11.58
23.99
19.22
9.47
16.84
7.80
CANY1
11.25
9.91
6.43
8.79
7.36
4.11
5.53
5.02
CAPI1
9.97
9.97
6.03
8.62
7.27
4.13
5.07
5.14
CHAS1
26.10
19.96
11.03
24.62
18.12
8.97
24.69
6.68
CHIR1
13.43
11.94
7.20
10.50
9.78
4.93
4.81
7.87
COHU1
30.30
20.15
10.78
28.83
18.59
9.52
18.17
4.44
CRLA1
13.74
12.96
7.62
9.36
8.64
5.22
8.67
2.37
CRMOl
14.00
14.08
7.53
11.91
9.28
4.97
7.26
4.72
DENA1
9.86
9.25
7.31
7.06
6.97
4.79
3.58
1.60
DOME1
19.43
17.95
7.46
17.20
15.42
6.18
14.13
11.58
DOSOl
29.05
19.96
10.39
28.29
18.88
8.92
13.57
3.40
EVER1
22.31
18.06
12.15
19.54
15.33
8.34
10.00
7.90
GAMOl
11.29
10.97
6.38
8.95
7.50
4.66
10.17
3.03
GICL1
13.11
11.11
6.66
8.93
8.02
4.22
5.73
4.40
GLAC1
20.47
16.91
9.18
16.19
13.69
6.99
22.24
7.50
GRCA2
11.66
9.56
7.04
7.94
6.95
4.18
5.97
4.74
GRGU1
22.82
15.20
11.99
21.93
13.92
9.78
12.07
3.23
GRSA1
12.78
10.63
6.66
9.66
8.28
4.45
8.01
6.69
GRSM1
30.28
19.66
11.24
29.16
18.42
10.05
16.09
4.48
GUMOl
17.19
14.93
6.65
14.60
12.86
4.83
6.25
12.95
HACRlb
13.33
9.16
7.43
12.67
8.37
4.78
1.24
2.01
HAVOl
18.86
19.01
7.17
18.66
18.82
5.64
1.65
1.93
HECA1
18.55
17.17
8.32
16.51
13.09
6.57
13.88
5.00
HEGL1
26.75
20.73
11.30
25.17
19.32
9.30
20.30
6.84
HOOV1
12.87
11.84
7.71
8.97
7.91
4.91
8.92
4.00
IKBA1
13.35
11.80
6.68
11.19
9.52
5.22
6.78
6.14
ISLE1
20.74
17.46
12.37
19.53
16.03
10.15
12.05
4.22
JARB1
12.07
12.85
7.87
8.73
7.82
5.23
7.45
8.00
JARI1
29.12
20.18
11.13
28.08
18.70
9.48
26.27
3.13
JOSH1
19.62
14.97
7.19
17.74
13.15
6.09
7.82
9.81
KAIS1
14.75
15.45
7.12
12.67
11.45
5.98
11.16
5.19
KALM1
15.51
14.40
9.44
13.35
12.12
7.80
12.46
2.43
KPB01C
14.11
12.94
11.31
10.47
10.48
6.96
3.39
2.32
LABE1
15.05
15.03
7.85
11.29
9.92
6.16
10.38
3.81
LA VOl
14.15
12.69
7.31
11.50
9.97
6.14
12.36
2.59
LIGOl
28.77
19.15
11.22
28.05
17.36
9.70
18.22
2.83
LOST1
19.57
18.37
8.00
18.27
15.89
5.88
10.17
9.28
LYEBl"
24.45
17.11
11.73
23.57
16.07
10.23
11.44
2.75
MACA1
31.37
23.04
11.08
29.83
22.03
9.79
19.44
4.28
MELA1
17.72
17.75
7.89
16.63
15.50
5.95
9.14
9.09
MEVE1
13.03
9.88
6.81
9.22
7.03
4.20
5.05
5.33
MING1
29.54
22.34
11.62
26.65
20.70
9.28
28.55
10.81
MOHOl
14.86
13.14
8.43
12.10
9.74
6.60
7.75
2.74
MONTI
14.48
15.15
7.73
10.84
9.61
5.43
14.92
4.89
MOOS1
21.72
15.64
12.01
20.66
14.07
9.97
11.13
2.54
26
-------
Site
First Implementation Period Approach
Baseline Visibility Current Visibility
Condition for Condition for
P90
20% Haziest Days Most Impaired
(2000-2004) Days (2012-2016) ( '
Recommended Approach
Baseline Visibility Current Visibility
Condition for 20% Condition for 20%
Most Impaired Most Impaired
Days (2000-2004) Days (2012-2016)
Derived-
NC
(2064)
e3 (M
carbon
m1)
dust
MORA1
18.25
14.93
8.54
16.53
13.35
7.66
13.33
2.53
MOZI1
10.52
9.16
6.08
7.29
5.63
3.16
5.70
3.23
NOAB1
11.46
11.30
6.83
8.78
7.17
4.54
10.18
4.23
NOCA1
13.96
12.65
8.39
12.57
10.42
6.79
8.20
1.97
OKEF1
27.13
20.47
11.44
25.34
18.73
9.47
20.65
5.50
OLYM1
16.74
13.38
8.44
14.93
12.24
6.88
8.78
1.76
PASA1
15.23
13.94
8.25
10.41
9.17
5.97
9.42
2.58
PEFOl
13.21
10.73
6.49
9.82
8.49
4.21
6.75
7.84
PINN1
18.46
16.03
7.99
17.02
14.35
6.96
11.33
5.88
PORE1
22.81
19.98
15.77
19.38
15.94
9.75
6.78
8.23
RAFA1
18.86
16.08
7.57
17.00
14.14
6.85
7.65
8.20
REDW1
18.45
17.88
13.91
13.64
12.70
8.54
5.86
4.44
ROMA1
26.48
20.21
12.12
25.25
18.32
9.79
23.38
5.35
ROMOl
13.83
11.54
7.15
11.12
8.66
4.93
8.54
5.32
SACR1
18.03
17.33
6.81
16.54
15.36
5.50
9.01
14.44
SAGA1
19.94
14.83
6.99
17.89
13.63
6.12
8.49
7.11
SAGOl
22.17
15.98
7.30
20.43
14.80
6.19
11.94
7.77
SAGU1
14.83
12.69
6.46
12.64
10.96
5.16
6.15
9.62
SAMA1
26.03
20.64
11.67
24.30
18.17
9.19
21.26
5.22
SAPE1
10.17
9.13
5.72
7.66
6.66
3.36
5.66
4.53
SAWT1
13.78
17.12
6.42
9.62
8.52
4.67
12.35
2.57
SENE1
24.16
19.20
12.65
23.62
18.41
11.11
13.67
2.52
SEQU1
24.62
21.10
7.70
23.23
19.20
6.29
23.11
11.47
SHEN1
29.31
19.71
11.35
28.32
18.40
9.52
15.06
3.92
SHROl
27.89
18.65
11.47
27.32
16.87
10.01
13.99
3.09
SIAN1
13.67

6.59
10.76

5.14
6.77
5.91
SIME1
18.56
16.97
15.60
13.67
13.69
8.49
3.42
4.63
SIPS1
29.03
20.95
10.99
27.71
19.77
9.55
21.66
4.79
SNPA1
17.84
15.54
8.43
15.37
13.07
7.25
12.33
1.79
STAR1
18.57
14.54
8.92
14.53
11.53
6.59
13.10
5.66
SULA1
13.41
15.55
7.43
10.06
8.53
5.48
11.78
3.22
SWAN1
25.49
19.06
11.55
24.40
17.44
9.79
16.47
5.01
SYCA28
15.26
14.59
6.65
12.16
11.45
4.68
13.12
15.93
THROl
17.74
15.97
7.80
16.35
13.70
5.96
9.87
8.71
THSI1
15.34
15.28
8.79
12.80
11.48
7.30
12.62
4.01
TONT1
13.94
12.47
6.54
11.34
10.63
5.06
7.14
8.76
TRCR1
11.61
10.03
8.40
9.16
8.85
6.38
5.11
2.38
TRIN1
16.32
16.02
7.90
11.97
10.45
6.24
10.36
3.61
ULBE1
15.14
14.31
8.16
12.76
10.79
5.87
9.82
6.17
UPBU1
26.27
20.57
11.57
24.25
18.85
9.43
17.22
7.72
VIIS1
17.02
18.49
10.68
14.29
15.60
8.53
2.60
21.54
VOYA2
19.27
17.08
12.06
17.75
15.04
9.38
11.48
4.14
WEMI1
10.33
9.17
6.21
7.81
6.74
3.98
6.51
3.93
WHIT1
13.70
13.21
6.80
11.31
10.41
4.89
7.16
7.13
WHPA1
12.76
12.02
8.35
10.48
8.51
6.15
6.89
2.41
WHPE1
10.41
9.11
6.08
7.35
7.02
3.53
5.13
3.50
WHRI1
9.61
7.93
6.06
6.30
5.18
3.02
4.92
3.56
WICA1
15.84
13.55
7.71
13.09
10.63
5.64
8.02
4.62
WIMOl
23.81
19.53
7.53
22.15
18.79
6.92
13.95
9.94
YELL2
11.76
12.26
6.44
8.30
7.65
3.98
10.08
3.06
YOSE1
17.63
15.84
7.64
13.52
11.89
6.29
13.14
5.19
ZICAlf
12.97
10.32
6.70
10.72
8.66
5.08
5.54
6.90
aSite data combined with BRET1 starting 01-01-08
bSite data combined with HALE1 starting 01-01-08
cSite data combined with TUXE1 starting 01-01-15
dSite data combined with LYBR1 starting 01-01-12
eSite data combined with SYCA1 starting 01-01-16
fSite data combined with ZI0N1 starting 01-01-04
27
-------
APPENDIX B
For each of the following sites, the left column gives the total extinction budget for days classified as the 20 percent most impaired
in 2015 (or 2013 when noted), middle column gives the time series from 2000-2016 of the annual average total extinction budget for
days classified as the 20 percent most impaired , and right column shows the visibility conditions on the 20 percent most impaired
days from 2000 to 2016, For all extinction budget figures, the following color scale applies: sulfate (yellow), nitrate (red), OMC (teal),
LAC (black), FS (tan), CM (brown), and sea salt (blue). For all visibility conditions figures, the blue points are annual average values;
red points are 5-year averages and the orange line is the glidepath between 2000-2004 and 2064.22
Acadia National Park, ME
80
60
C
.2 40 H
O
c
X
111
20 H
sulfate
¦	nitrate
OMC
¦	lac
FS
¦	sea salt
120
100
I |l
J11
C
o
o
c
X
LU
"I—I-
100
200
Julian Day
300
sulfate
nitrate
CM
sea salt
:
24.0
22.0 -
20.0 -
> 18.0
16.0 -
14.0 -
Annual Avg.
5-yr Avg.
Glidepath
12.0
2016
2000
2004
Year
~r
2008
Year
2012
2016
•"Updated site-specific graphics summarizing visibility status and trends following the Regional Haze Rule metrics can be found at
http://views.cira.colostate.edu/fed/SiteBrowser/Default.aspx?appkey=SBCF_VisSum.
28
-------
Agua Tibia, CA
o	100 200
Julian Day
Badlands National Park, SD
30 -
c 20 H
o
o
c
X
LLI
10 -
I I I
I ,
300
100	200
Julian Day
300
100 -
c
o
¦5 40 -
x
LLI
c
O
X
LU
24.0
22.0
(/)
5
a>
>
o
CD
Q
20.0
18.0
16.0
14.0
2016
> 14.0
29
-------
Mount Baldy, AZ
LU 8
o	100	200
Julian Day
Bandelier National Monument, NM
300
30 -
25 -
20 -
c
o
15 -
x
LU 10
I J
100	200
Julian Day
300
18 -
_ 15
E
H 12 H
c:
o
'o 9 H
c:
LU
6 -
3 -
0
2000
ii:l!
	T
2004

2008
Year
2012
9.50
2016
E 16
O 12
> 8.00
6.50 -H-
11.0
10.0
C/)
£

o
CD
Q
2016
30
-------
Big Bend National Park, TX
§ 30
Q 15.0
* 20
LU
100	200
Julian Day
Bliss State Park, CA
2000 2004 2008 2012
Year
30 -
c
o
o
c
X
LU
20
10
100	200
Julian Day
300

i—1—1—1—i—'—'—'—r
2004 2008 2012
Year
31
-------
Bosque del Apache, NM
5 20 -
0	100	200
Julian Day
Boundary Waters Canoe Area, MN
300
60
50 -
40 -
I 30
o
_c
X
LU 20
10 -
'I l;
100	200
Julian Day
300
30 -
o
x
LU
13.0 -
5 12.0 -
I
Q) 11.5
10.5 -
80 -
60 -
c
o
o 40 -
x
LU
20 -
22.0
20.0
03
5
CD
>
o
a>
Q
18.0
16.0
14.0
12.0
2016
32
-------
Breton Island, LA
100
¦!
>< 40
0	100	200	300
Julian Day
Bryce Canyon National Park, UT
30 -
25 -
20
I-
_c
X
LU 10 -\
5 -
i|il|!
I
i i
i ,
i IJ
i—i—i—r
100
200
Julian Day
300
¦!
I
LU 6 -
	1—I	
2000 2004 2008

24.0 -I

23.0 -

22.0 -
co


21.0 -
0
>

n

CB
20.0 -
Q


19.0 -

18.0 -

17.0 -
2016
2000

9.50 -I

9.00 -

8.50 -
03

£
8.00 -
CD
>

CI

a)
7.50 -
O


7.00 -

6.50 -

6.00 -
2016
2016
Year
~i—1—1—1—i—'—'—'—r
2000 2004 2008 2012
Year
2016
33
-------
Bridger Wilderness, WY
20 -
16 -
' 12 -
C
o
0
1	•
UJ
4 -
0
I I
0	100	200	300
Julian Day
Brigantine National Wildlife Refuge, NJ
150 -
120
— 90 -
C
o
o
¦¦§ 60 h
x
LU
30
n—l—i—r~
100	200
Julian Day
300
15 -
^ 12
E
c 9
O
H—'
o
x 6
Lll
iSIfllflLi!
,i»n
!!l!
¦ ii
8.40
i	i	i	i	i	i	i	n
2000 2004 2008 2012 2016
Year
> 7.20


> 24.0
34
-------
Cabinet Mountains, MT
40
O 20
24 -
20 -
16
In
O
X
LU 8 -
100	200
Julian Day
300
2000
11.4
2016
S 10.5 -
0) 10.2
9.6 -
Caney Creek, AR
100
30.0
120 -
> 24.0
ii
5 90
ii;;
m 60 -
100	200
Julian Day
2008 2012
Year
35
-------
Cartyonlands National Park, UT
¦!!!i
LiJ 6 _

9.50 -1

9.00 -

8.50 -
03


8.00 -
0
>

n

03
7.50 -
Q


7.00 -

6.50 -

6.00 -
0	100	200
Julian Day
Capitol Reef National Park, UT
300
2000
	1	i	i	1	1	n
2004 2008
Year
2012
2016
2000
2016
20 -
16 -
12
c
o
o
•I 8
X
LU
l!
II
! i
I

100	200
Julian Day
300
'I!
a) 8.0
i—1—1—¦—i—'—'—'—r
2004 2008 2012
Year
36
-------
Chassahowitzka National Wildlife Refuge, FL
100
C
O
o
c
X
LU
o 60
Q 20.0
:ii;
16.0 —,
0	100	200	300
Julian Day
Chiricahua National Monument, AZ
25 -
^ 20 H
E
o
c
X
LU
15 -
10 -
5 -
I '
I I
!
i
100	200
Julian Day
300
10.8 -
10.4
ill
I!
> 10.0 -
9.6
37
-------
Cohutta, GA
9 40
0	100	200
Julian Day
Crater Lake National Park, OR
60 -
50 -
40 -
~ 30
o
c
•*->
x	
LLI 20 -
10 -
I  24.0

9.60 -

9.30 -
03

£
9.00 -
CD
>

a

a)
8.70 -
O


8.40 -

8.10 -

7.80 -
2016
"i—1—1—1—i—'—'—'—r
2000 2004 2008 2012
Year
2016
38
-------
Craters of the Moon National Monument, ID
30 -
25 -
20 -
C
o
~ 15
c
X
UU 10 ^
5 -
0
100
"i—i—r-
200
Julian Day
Denali National Park, AK
20 -
16
-—• 12 H
C
o
o
M 8
X
LU
II
.i i
11
i!
100	200
Julian Day
-i—i—r
300
300
14.0
c
o
o
X
LU
13.0 -
> 11.0
10.0 -
9.0
2000 2004 2008 2012 2016
Year
2000
2004
2012
2016
2016
h—1—1—1—\—'—'—'—r
2000 2004 2008 2012
Year
2016
39
-------
Dome Lands Wilderness, CA
80 -
60
¦J§ 40
o
c
'¦*—I
X
LU
20 -
i
I
'III
0	100	200
Julian Day
Dolly Sods Wilderness, WV
i—i—i—r
300
Q 16.0
LU 20
120 -
240
200 -
E 90
E 160 -
> 24.0
O 120 -
O 60
100	200
Julian Day
i—1—1—¦—i—'—'—'—r
2004 2008 2012
Year
40
-------
Everglades National Park, FL
O 40
0	100	200
Julian Day
Gates of the Mountains, MT
300
25 -
*T" 20
E
I 15
o
c
X 10
LU
5 -
I
100	200	300
Julian Day
C
o
o
X
LU

> 18.0
¦Jl!|3!!
2000 2004 2008 2012 2016
Year
c
o
X
LU
> 8.00
i—1—1—1—\—'—'—'—r
2004 2008 2012
Year
41
-------
Gila Wilderness, NM
10.0
9.5
9.0
03
s
CD
">
o 8.5
a)
Q
8.0
7.5
\i
\A

\ I
V
\ \
\ \
v\\
\ \


L
T
100	200
Julian Day
Glacier National Park, MT
i r
2000 2004 2008 2012
2016
2000 2004
Year
T
2008
Year
80 -
60 -
.2 40
o
c
X
LU
20 -\
c
o
x
LU
100	200
Julian Day
300
2000 2004
2008 2012 2016
Year
2012 2016
ii
> 15.0
i—1—1—1—i—'—'—'—r
2000 2004 2008 2012
Year
2016
42
-------
Grand Canyon National Park, AZ
> 7.50
6.00
0	100	200
Julian Day
Great Gulf Wilderness, NH
Year
22.0 -
120 -
? 18.0
O 40
03 16.0
y 60
100	200
Julian Day
43
-------
Great Sarid Dunes National Monument, CO
20 -
16 -
te
c
.0
o
••§ 8 -
x
LU
4 -
0 -
0	100	200	300
Julian Day
Great Smoky Mountains National Park, TN
100 -
80 -
VE
2 60 -
c
o
o
.E 40 -
-~—»
x
LU
20 -
0 -
0	100	200	300
Julian Day
44
-------
Guadalupe Mountains National Park, TX
I !
LU 20
100
200
300
2000 2004 2008 2012
Julian Day	Year
Haleakala National Park, HI (combined HALE1 and HACR1 starting 01/01/2007)
30 -
25 -
20 -
c
¦B 15
o
_c
X
LU 10
M I
! !
|'|
100	200
Julian Day
300
c 20
17.0
16.0
15.0
03
s
cu
'> 14.0
o
Q)
Q
13.0
12.0
11.0 -H
2016
2000

14.0

13.0

12.0
03


11.0
CD
>

O

CI)
10 0
Q


9.0

8.0

7.0
2016
2000
45
-------
46
-------
Hercules-Glades, MO
26.0 -
24.0
i[
Q 22.0
£ 40
x
W 60
Hoover, CA
30
25 -
20 -
.2	15
o
"x	10
LLI	u
5 -
100	200
Julian Day
i—1—1—1—i—¦—¦—¦—i—1—1—1—r
2004 2008 2012 2016
Year
'i i !•
100	200
Julian Day
300
2? 8.50
03 8.00
LU 6
7.00 -
2016
47
-------
ike's Backbone, AZ
o 12
0	100	200
Julian Day
Isle Royale National Park, Ml
300
2000
ll'l
12.0
2016
? 10.5 -
a) 10.0
9.0 -
100 -f
> 18.0
200
Julian Day
i—1—1—¦—i—'—'—'—r
2004 2008 2012
Year
48
-------
Jarbidge Wilderness, NV
 24.0
49
-------
Joshua Tree National Park, CA
5 20 -
Kaiser, CA
40 -
30 -
C
o
a 20
x
LU
10 -
100	200
Julian Day
100	200
Julian Day
300
300
o
LU
I

19.0 -|

18.0 -

17.0 -
If)

s
16.0 -
0)
>

CI

(1)
15.0 -
Q


14.0 -

13.0 -

12.0 -
2000 2004 2008 2012
Year
2016
2000
Year
2016
..
O 12.0
50
-------
Kalmiopsis, OR
40
100	200
Julian Day
Lava Beds National Monument, CA
40
£ 30 -
C
o
"o 20 H
X
LL1
10
c
o
o
X
LU
::
S 13.2
 10.5
51
-------
Lassen Volcanic National Park, CA
c
o
o
c
X
UJ
12.0 -
% 11.0 -
a) 10.5
>< 10
LLI
0	100	200
Julian Day
Linville Gorge, NC
250
200 -
J- 150
o
0
1	100
LLI
50

u
33.0
100	200
Julian Day
300
2000 2004 2008
Year
2012 2016
> 24.0
52
-------
Lostwood, ND
C
o
o
c
X
LU
iii
.2 40
21.0
20.0
19.0
03
5
CD
'> 18.0
o
CD
Q
17.0
16.0
15.0


	1	1	1	
\\
100
200
300
2000
2004
2008
2012
2016
2000
2004
Julian Day	Year
Lye Brook Wilderness, VT (combined LYBR1 and LYEB1 starting 01/01/2012)
r
2008
Year
2012
2016
150 -
120 -
C
o
o
c
X
LU
100	200
Julian Day
300

£ 20.0
0) 18.0

53
-------
Mammoth Cave National Park, KY
180 -
300 -
250 -
150 -
120 -
5 28.0
= 150
CD 26.0
X
LU 60
i'siii-
LU 100
100	200
Julian Day
2008
Year
Medicine Lake, MT
100 -
> 16.0
o 30 -
100	200	300
Julian Day
2000
2004
2012
2016
~i—1—1—1—i—'—'—'—r
2004 2008 2012
Year
54
-------
Mesa Verde National Park, CO
18
15 -
E 12
C
.2 9
-4—1
o
c
LU
6 -
3 -
Mingo, MO

120 -

100 -
te


80 -
c

o
-
o
60 -


X
-
LU
40 -

20 -
I

100	200
Julian Day
¦—i—i—i—
300
100	200	300
Julian Day

j:
2000
n	I	i	I	1
2004 2008
Year
2016
C
o
o
c
X
LU
i
28.0
26.0
03
5
CD
>
o
a>
Q
24.0
22.0
20.0
18.0
2016
"i—1—1—1—i—'—'—'—r
2000 2004 2008 2012
Year
2016
55
-------
Mount Hood, OR
i;
C 20 -
13.0
12.0 -
g 11.0
cu
>
o
a)
a io.o
9.0
8.0
100	200
Julian Day
300
2000 2004 2008
Year
2012
2016
2000
2004
—l—~
2008
Year
Monture, MT
30
20
c
o
o
c
x
LU
10
100	200
Julian Day
300
2000 2004 2008
Year
2012 2016
2012
2016
"? 10.5
Q> 10.0
9.0 -
2016
56
-------
Moosehorn National Wildlife Refuge, ME
::
c 40
> 18.0

>
100	200
Julian Day
Mount Rainier National Park, WA
2008 2012
Year
40
*T" 30 -
E
1-H
_c
X
LL1
10 H
I !
i—i—i—i—i—i—i—i—
100	200
Julian Day
300

> 15.0

i—1—1—1—i—'—'—'—r
2004 2008 2012
Year
57
-------
Mount Zirkel Wilderness, CO
18 -
15 -
E 12
c
o
o
c
X
LU
9 -
6 -
0
0	100	200
Julian Day
North Absaroka, WY
24 H
20
TE 16
C
.2 12
-f—'
o
_c
X Q
LJJ o
100	200
Julian Day
300
i—i—r-
300
. :
Q 6.0
LU 6
Year
18 -
15
12 -
I M
X
LU 6
!|!
¦ I
!¦ I
2000
—i—i	
2004
:i.!
i1"
2008
Year
~i—1—r
2012
2016
cn
I 8 00
>
o
03 7.50
Q
58
-------
North Cascades National Park, WA
c
o
o
c
X
UJ
Q 11.0
X
LU 10 -
0	100	200
Julian Day
Okefenokee National Wildlife Refuge, GA
2000 2004 2008 2012
Year
E 120
> 22.0
100	200
Julian Day
59
-------
Olympic National Park, WA
		
•*= 20
16.0
15.0
03
g
CD
>
O
Q)
Q
14.0
13.0
12.0
11.0 -4
100	200
Julian Day
2016
Pasayten, WA
30 -
c 20
o
o
c
X
LU
10 -
' I 1
100	200	300
Julian Day
S!
3 10.0
i—1—1—1—\—'—'—'—r
2004 2008 2012
Year
2016
60
-------
Petrified Forest National Park, AZ
o	100 200
Julian Day
Pinnacles National Monument, CA
300
60
40 -
c
o
o
c
X
LU
20
I !
100	200
Julian Day
300
24 -
20 -
E 16
C
.2 12
•*—>
o
X Q
LU o
4 -

11.0
2000
2016
O 15.0
x 20
2000 2004 2008
Year
2012
2016
i—1—1—1—i—'—'—'—r
2000 2004 2008 2012
Year
2016
61
-------
Point Reyes National Seashore, CA
120
22.0
80 -
60 -
c
o
c 40 H
-t—>
X
LLI
20 ^
i! Hi
100	200
Julian Day
San Rafael, CA
40 -
£ 30 -
c
o
'o 20 H
c
X
LU
10

I I
I
100	200
Julian Day
300
-i—i—r
300
5 19.0 -
CD 18.0
>< 40
16.0 -
2000 2004
I
o 15.0
"
X 20
62
-------
Redwood National Park, CA
15.0 -
5 14.0 -
a) 13.5
O 20 -
100	200
Julian Day
Cape Romain National Wildlife Refuge, SC
c
o
o
c
X
LU
180 -
28.0 -
150
5 24.0
03 22.0
O 90
18.0 -
100	200
Julian Day
63
-------
Rocky Mountain National Park, CO
¦!
> 10.0
-= 20
100	200
Julian Day
Salt Creek, NM
60 -
c 40
o
o
c
X
LU
20 -
i l' in ni
I ' I l| I | II
¦n

m 20
19.0
18.0
17.0
03
g
CD
'> 16.0
o
CD
O
15.0
14.0
13.0



\
100	200
Julian Day
300
2000 2004 2008 2012
Year
2016
2000
2004
2008
Year
2012
2016
64
-------
San Gabriel, CA
*; 20
o	100	200
Julian Day
San Gorgonio Wilderness, CA
60 -
c 40
o
o
c
X
LU
20
h
100	200
Julian Day
300
300
C
O
o
X
LU

c
o
" 40 -
x
LU
19.0
18.0 -
17.0
03
| 16.0 I
>
o
a) 15.0
Q
14.0
13.0 -
12.0
2016
2000
2004
T
2008
Year
2012
2016
22.0
20.0
03
5
03
>
o
a>
Q
18.0
16.0
14.0
2016
65
-------
Saguaro National Monument, AZ
13.0 -
II
5 12.0 -
5 40
II
Q) 11.5
O 15 -
10.5 -
iHi,
2000 2004
100	200
Julian Day
St. Marks, FL
100 -
C
O
o
c
X
LU
> 22.0
I

100	200
Julian Day
66
-------
San Pedro Parks, NM

03
g 7.50
>
a> 7.00
Q
c 20
0	100	200
Julian Day
Sawtooth National Forest, ID
Year
9.60 -
i
? 9.00
03 8.70
x 10
8.10 -
100	200
Julian Day
67
-------
Seney, Mi
150 -
120 -
te
1	90 -
c
o
o
C 60 -
x
LU
30 -
0 -
0	100	200	300
Julian Day
Sequoia National Park, CA
180 -
150 -
120 -
2
.1 90 -
'+-»	J
o
is60 -
30 -
0 -
0	100	200	300
Julian Day
l
I 1	I
I I i.i !| i
-i
2000 2004 2008 2012 2016
Year
2000 2004 2008 2012 2016
Year
—i—
2004
1—l—7
2008
Year
2012
—1—I—I—i—I—I—I—.—I—¦—¦—¦—I—1—.—.—f—
2000 2004 2008 2012 2016
Year
—i—
2000
2016
68
-------
Shenandoah National Park, VA
120 1
33.0
240 -
> 24.0
LU 80
Julian Day
Shining Rock Wilderness, NC
200 -
160 -
c 120 -
c 40
x 80
100	200	300
Julian Day
2000 2004 2008 2012 2016
Year
~i—1—1—1—i—'—'—'—r
2000 2004 2008 2012
Year
2016
69
-------
Sierra Ancha, AZ (2013 data shown on figures in left column)
y!
x 10
100	200
Julian Day
Simeonof, AK
100 -
cz
o
o
c
X
LU
Julian Day
300
24 -
20 -
16 -
~ 12
o
LU 8
4 -

¦in
!i !l
0 —t—r
2000
I
2004
n i r~
2008
Year
¦ ¦ I
i i r
2012

11.4-1

11.1 -

10.8 -


s
10.5 -
cu
>

n

a)
10.2 -
O


9.9 -

9.6 -

9.3 -
2016
2000
2016
40 -
E 30 -
c
o
o 20 -
x
LU
10 -
03
| 145
>
o
a> 14.0
Q
2000 2004 2008 2012
Year
2016
i—1—1—1—i—'—'—'—r
2000 2004 2008 2012
Year
2016
70
-------
Sipsey Wilderness, AL
240 -
200 -
E 90
w
160 -
> 24.0
-= 120 -
O 60
x
W 80 -
rpSs
100	200
Julian Day
300
2008 2012
Year
Snoqualmie Pass, WA
60 -
40 -
c
o
o
C
X
LU
20 -
J1 I
' I i'i'1
i 11
I'll
E 30 -
c
o
o 20 -
x
LU
100	200
Julian Day
300
? 14.5
Q> 14.0
13.0 -
71
-------
Starkey, OR
60 -
§ 30
O 30
x 20
LU
2000 2004
100	200
Julian Day
Suia Peak, MT (2013 data shown for figures in right column)
20 -
-P 16
£
c 12
O
"¦+—1
o
c
x 8 H
LU
I i
c
o
X
LU
100	200	300
Julian Day
17.0
16.0 -
15.0
if)
£ 14.0 -1

o
a) 13.0
Q
12.0
11.0 -
10.0
2016
2000
10.0 -
9.5 -
03
s
CD
'> 9.0
o
CD
Q
8.5
8.0 -
2016
2000
72
-------
Swanquarter, NC
o 30 -
9 60
Q 20.0
'¦!oD
16.0 —,
2012 2016
100	200
Julian Day
Sycamore Canyon, AZ (2013 data shown for figures in right column)
30 -
25
E 20
! 15 H
o
UJ 10
5 -
|l I
' I
w 10
13.0 -
12.5
in
I 120
>
o
a)
Q 11.5 H
11.0 -
10.5
100	200	300
Julian Day
2000 2004
2008 2012
Year
2016
-i—i—i—|—i—i—.—f—¦—¦—¦—r
2000 2004 2008 2012
Year
2016
73
-------
Theodore Roosevelt, ND
E 40
> 15.0
.2 30 -
14.0
LU 20
13.0
0	100	200
Julian Day
Three Sisters Wilderness, OR
300
2000
2004
2012
2016
i—1—1—1—i—¦—¦—¦—i—1—1—1—r
2000 2004 2008 2012 2016
Year
40 -
£ 30
c
o
o 20 H
X
LU
10 -
II
100	200
Julian Day
300
> 12.0
LU 10
2000 2004 2008 2012
Year
74
-------
Tonto National Monument, AZ
CD 11.0
>< 10 -
100	200
Julian Day
Trinity, CA (2013 data shown for figures in right column)
c
o
30
25
20 :
15 :
x
w 10 4
I'll
i—i—:—\—i—i—i—r
100	200
300
Julian Day
CD 11.0
LU 10
2000 2004 2008 2012 2016
Year
75
-------
Trapper Creek, AK
30 -
c
o
o
c
X
UJ
20 -
24 -
20
E 16
c
O 12
o
10 -
X
LU
8 -
0
300
0	100	200
Julian Day
Tuxedni, AK (2013 data shown for figures in right column)
12.0
2000 2004
2016
30
c 20
"¦+—1
o
c
X
LU
10 -
!
;l ! "¦
!!
Q 10.0
LU g
100	200
Julian Day
300
2000
2004
2008 2012
Year
2016
i—1—1—1—i—'—'—'—r
2000 2004 2008 2012
Year
2016
76
-------
UL Berid, MT
c 20
> 12.0
10.0
100	200
Julian Day
Upper Buffalo Wilderness, AR
5 24.0
£ 120
fi1'!
O 40 -
03 22.0
100	200
Julian Day
i—1—1—¦—i—'—'—'—r
2004 2008 2012
Year
77
-------
Virgin Islands National Park, VI
> 15.0
uu 20
-
2000 2004
100	200
Julian Day
Voyageurs National Park, MN
80 T
60 -
50 -
40
B 30 H
o
LU 20 -
10 I
I !
I
~i—i—r
100
200
Julian Day
300
l!
n—r
2000 2004
I—I—I—^^—[—
2008 2012
2016
Year
78
-------
Weminuche Wilderness, CO

!!!•
!!
8.50
8.00
.3? 7.50
7.00 -
6.50 -
6.00
0	100	200	300
Julian Day
White Mountain, NM
2000
2004 2008
Year
2012 2016
40 -
30 -
.2 20
o
X
LU
10 H
11
> 11.0
LU 10 -
100	200
Julian Day
300
2000 2004 2008 2012
Year
2016
i—1—1—1—i—'—'—'—r
2000 2004 2008 2012
Year
2016
79
-------
White Pass, WA
40 -f
10.5 -
II1
Q) 9.0
8.0 -
Julian Day
Wheeler Peak, NM (2013 data shown for figures in right column)
2000 2004 2008 2012 2016
Year
25 -
^ 20
E
c 15
o
c
x 10
LU
5 -
I II
' I
I I
I I

!!s

> 7.20
100	200
Julian Day
300
	1—i	1—i—i—i—i—i—h—r
2000 2004 2008 2012 2016
Year
i—1—1—¦—i—'—'—'—r
2000 2004 2008 2012
Year
2016
80
-------
White River National Forest, CO

18

15


E

2
12
c

o

—'
o
9
c

'•*—>
X

LU
6

3

0
I I
I
!"!!i
i'h:
100	200
Julian Day
300
Wind Cave, SD
30 -
25
20 -
c
o
15 -
x
LU 10
5 -
l I
1
I
100	200
Julian Day
300
15 i
12 -
C
o
o
c
X
LU
7.00 -
l:

S 6.00 -
CD 5.50
4.50 -
2004
2012
~i	r
2000
2008
Year
> 12.0
O 20
81
-------
Wichita Mountains, OK
120 -
100 -
£ 40
Q 20.0
x
UJ 40 -
100	200
Julian Day
Yellowstone National Park, WY
24
20 -
TE 16 -
C
.2 12 H
-f—*
o
c
UJ
8 -
4 -
I
100	200
Julian Day
300
SBeIb|iFi
2!
!•:
> 7.60
82
-------
Yosemite National Park, CA
40 -
J 30
c
o
" 20
x
LU
100	200
Julian Day
> 13.0
2000 2004 2008 2012
Year
Zion National Park, UT (combined ZION1 and ZICA1 starting 1/1/04)
o
c
X
LU
20 -
16 -
12
8 -
I '
100	200
Julian Day
300

I
> 10.0 -
! !i
::::
¦
x 10
83
-------
84
-------
United States	Office of Air Quality Planning and Standards	Publication No. EPA-454/R-18-010
Environmental Protection	Air Quality Assessment Division	December 2018
Agency	Research Triangle Park, NC
85
-------