1
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
VI.
Air
Quality
Modeling
Approach
and
Results
Overview
In
this
section
we
summarize
the
air
quality
modeling
approach
used
for
the
proposed
rule,
we
address
major
comments
on
the
fundamental
aspects
of
EPA's
proposed
approach,
and
we
describe
the
updated
and
improved
approach,

based
on
those
comments,
that
we
are
finalizing
today.
This
section
also
contains
the
results
of
EPA's
final
air
quality
modeling,
including:
(
1)
identifying
the
future
baseline
PM2.5
and
8­
hour
ozone
nonattainment
counties
in
the
East;

(
2)
quantifying
the
contribution
from
emissions
in
upwind
States
to
nonattainment
in
these
counties;
(
3)
quantifying
the
air
quality
impacts
of
the
CAIR
reductions
on
PM2.5
and
8­
hour
ozone;
and
(
4)
describing
the
impacts
on
visibility
in
Class
I
areas
of
implementing
CAIR
compared
to
implementing
the
regional
haze
requirement
for
best
available
retrofit
technology
(
BART).

We
present
the
air
quality
models,
model
configuration,

and
evaluation;
and
then
the
emissions
inventories
and
meteorological
data
used
as
inputs
to
the
air
quality
models.
Next,
we
provide
the
updated
interstate
contributions
for
PM2.5
and
8­
hour
ozone
and
those
States
2
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
that
make
a
significant
contribution
to
downwind
nonattainment,
before
considering
cost.
Finally,
we
present
the
estimated
impacts
of
the
CAIR
emissions
reductions
on
air
quality
and
visibility.
As
described
below,
our
air
quality
modeling
for
today's
rule
utilizes
the
Community
Multiscale
Air
Quality
(
CMAQ)
model
in
conjunction
with
2001
meteorological
data
for
simulating
PM2.5
concentrations
and
associated
visibility
effects
and
the
Comprehensive
Air
Quality
Model
with
Extensions
(
CAMx)
with
meteorological
data
for
three
episodes
in
1995
for
simulating
8­
hour
ozone
concentrations.
Our
approach
to
modeling
both
PM2.5
and
8­

hour
ozone
involves
applying
these
tools
(
i.
e.,
CMAQ
for
PM2.5
and
CAMx
for
8­
hour
ozone)
using
updated
emissions
inventory
data
for
2001,
2010,
and
2015
to
project
future
baseline
concentrations,
interstate
transport,
and
the
impacts
of
CAIR
on
projected
nonattainment
of
PM2.5
and
8­

hour
ozone.
We
provide
additional
information
on
the
development
of
our
updated
CAIR
air
quality
modeling
platform,
the
modeling
analysis
techniques,
model
evaluation,
and
results
for
PM2.5
and
8­
hour
ozone
modeling
in
the
CAIR
Notice
of
Final
Rulemaking
Emissions
Inventory
Technical
Support
Document
(
NFR
EITSD)
and
the
Air
Quality
Modeling
Technical
Support
Document
(
NFR
AQMTSD).
3
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
A.
What
Air
Quality
Modeling
Platform
Did
EPA
Use?

1.
Air
Quality
Models
a.
The
PM2.5
Air
Quality
Model
and
Evaluation
Overview
In
the
NPR,
we
used
the
Regional
Model
for
Simulating
Aerosols
and
Deposition
(
REMSAD)
as
the
tool
for
simulating
base
year
and
future
concentrations
of
PM2.5.
Like
most
photochemical
grid
models,
the
predictions
of
REMSAD
are
based
on
a
set
of
atmospheric
specie
mass
continuity
equations.
This
set
of
equations
represents
a
mass
balance
in
which
all
of
the
relevant
emissions,
transport,

diffusion,
chemical
reactions,
and
removal
processes
are
expressed
in
mathematical
terms.
The
modeling
domain
used
for
this
analysis
covers
the
entire
continental
United
States
and
adjacent
portions
of
Canada
and
Mexico.

The
EPA
applied
REMSAD
for
an
annual
simulation
using
meteorology
and
emissions
for
1996.
We
used
the
results
of
this
1996
Base
Year
model
run
to
evaluate
how
well
the
modeling
system
(
i.
e.,
the
air
quality
model
and
input
data
sets)
replicated
measured
data
over
the
time
period
and
domain
simulated.
We
performed
a
model
evaluation
for
PM2.5
and
speciated
components
(
e.
g.,
sulfate,
nitrate,
elemental
carbon,
organic
carbon,
etc.)
as
well
as
nitrate,
sulfate
4
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
and
ammonium
wet
deposition,
and
visibility.
The
evaluation
used
available
1996
ambient
measurements
paired
with
REMSAD
predictions
corresponding
to
the
location
and
time
periods
of
the
measured
data.
We
quantified
model
performance
using
various
statistical
and
graphical
techniques.
Additional
information
on
the
model
evaluation
procedures
and
results
are
included
in
the
Notice
of
Proposed
Rulemaking
Air
Quality
Modeling
Technical
Support
Document
(
NPR
AQMTSD).

The
EPA
received
numerous
comments
on
various
elements
of
the
proposed
PM2.5
air
quality
modeling
approach.
The
major
comments
are
responded
to
below.
Other
comments
are
addressed
the
Response
to
Comment
(
RTC)
document.
Regarding
REMSAD,
commenters
argued
that:
(
1)
the
REMSAD
model
is
an
inappropriate
tool
for
modeling
PM2.5;
(
2)
the
scientific
formulation
of
the
model
is
simplistic
and
outdated
and
that
other
models
with
better
science
are
available
and
should
be
used;
and
(
3)
results
from
REMSAD
are
directionally
correct
but
better
tools
should
be
used
as
the
basis
for
the
final
determinations
on
transport
and
projected
nonattainment.

We
agree
that
models
with
more
refined
science
are
available
for
PM2.5
modeling
and
we
have
selected
one
of
these
models,
the
CMAQ
as
the
tool
for
PM2.5
modeling
for
the
final
CAIR.
The
CMAQ
model
is
a
publicly
available,
5
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
peer­
reviewed,
state­
of­
the­
science
model
with
a
number
of
science
attributes
that
are
critical
for
accurately
simulating
the
oxidant
precursors
and
non­
linear
organic
and
inorganic
chemical
relationships
associated
with
the
formation
of
sulfate,
nitrate,
and
organic
aerosols.

Several
of
the
important
science
aspects
of
CMAQ
that
are
superior
to
REMSAD
include:
(
1)
updated
gaseous/
heterogeneous
chemistry
that
provides
the
basis
for
the
formation
of
nitrates
and
includes
a
current
inorganic
nitrate
partitioning
module;
(
2)
in­
cloud
sulfate
chemistry,

which
accounts
for
the
non­
linear
sensitivity
of
sulfate
formation
to
varying
pH;
(
3)
a
state­
of­
the­
science
secondary
organic
aerosol
module
that
includes
a
more
comprehensive
gas­
particle
partitioning
algorithm
from
both
anthropogenic
and
biogenic
secondary
organic
aerosol;
and
(
4)
the
full
CB­
IV
chemistry
mechanism,
which
provides
a
complete
simulation
of
aerosol
precursor
oxidants.

However,
even
though
REMSAD
does
not
have
all
the
scientific
refinements
of
CMAQ,
we
believe
that
REMSAD
treats
the
key
physical
and
chemical
processes
associated
with
secondary
aerosol
formation
and
transport.
Thus,
we
believe
that
the
conclusions
based
on
the
proposal
modeling
using
REMSAD
are
valid
and
therefore
support
today's
6
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
findings
based
only
on
CMAQ
that:
(
1)
there
will
be
widespread
PM2.5
nonattainment
in
the
eastern
U.
S.
in
2010
and
2015
absent
the
reductions
from
CAIR;
(
2)
upwind
States
in
the
eastern
part
of
the
United
States
contribute
to
the
PM2.5
nonattainment
problems
in
other
downwind
States;
(
3)

States
with
high
emissions
tend
to
contribute
more
than
States
with
low
emissions;
(
4)
States
close
to
nonattainment
areas
tend
to
contribute
more
than
other
States
farther
upwind;
and
(
5)
the
CAIR
controls
will
produce
major
benefits
in
terms
of
bringing
areas
into
or
closer
to
attainment.

Comments
and
Responses
(
i)
REMSAD
Science
and
Evaluation
Comment:
Some
commenters
stated
that
REMSAD
is
an
inappropriate
model
for
use
in
simulating
PM2.5.
Other
commenters
said,
more
specifically,
that
the
chemical
mechanism
in
REMSAD
(
i.
e.,
micro
CB­
IV)
is
simplified
and
not
validated,
and
that
the
model
has
not
been
scientifically
peer­
reviewed.

Response:
The
EPA
disagrees
with
comments
claiming
that
REMSAD
is
an
inappropriate
tool
for
modeling
PM2.5.
The
EPA
believes
that
REMSAD
is
appropriate
for
regional
and
national
modeling
applications
because
the
model
does
7
1
Even
so,
EPA
acknowledges
that
REMSAD
has
certain
limitations
not
found
in
CMAQ.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
include
the
key
physical
and
chemical
processes
associated
with
secondary
aerosol
formation
and
transport.
1
Specifically,
REMSAD
simulates
both
gas
phase
and
aerosol
chemistry.
The
gas
phase
chemistry
uses
a
reducedform
version
of
Carbon
Bond
chemical
mechanism
(
micro­

CBIV
Formation
of
inorganic
secondary
particulate
species,

such
as
sulfate
and
nitrate,
are
simulated
through
chemical
reactions
within
the
model.
Aerosol
sulfate
is
formed
in
both
the
gas
phase
and
the
aqueous
phase.
The
REMSAD
model
also
accounts
for
the
production
of
secondary
organic
aerosols
through
chemistry
processes
involving
volatile
organic
compounds
(
VOC)
and
directly
emitted
organic
particles.
Emissions
of
non­
reactive
particles
(
e.
g.,

elemental
carbon)
are
treated
as
inert
species
which
are
advected
and
deposited
during
the
simulation.

With
regard
to
comments
on
the
micro
CB­
IV
chemical
mechanism,
although
this
mechanism
treats
fewer
organic
carbon
species
compared
to
the
full
CB­
IV,
the
inorganic
portion
of
the
reduced
mechanism
is
identical
to
the
full
chemical
mechanism.
The
intent
of
the
CB­
IV
mechanism
is
to:
(
a)
provide
a
faithful
representation
of
the
linkages
between
emissions
of
ozone
precursor
species
and
secondary
8
2
Whitten,
G.
memorandum:
Comparison
of
REMSAD
Reduced
Chemistry
to
Full
CB­
4.
February
19,
2001.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
aerosol
precursor
species;
(
b)
treat
the
oxidizing
capacity
of
the
troposphere,
represented
primarily
by
the
concentrations
of
radicals
and
hydrogen
peroxide;
and
(
c)

simulate
the
rate
of
oxidation
of
the
nitrogen
oxide
(
NOx)

and
sulfur
dioxide
(
SO2),
which
are
precursors
to
secondary
aerosols.
The
EPA
agrees
that
micro
CB­
IV
is
simplified
compared
to
the
full
CB­
IV
mechanism.
However,
performance
testing
of
micro
CB­
IV
indicates
that
this
simplified
mechanism
is
similar
to
the
full
CB­
IV
chemical
mechanism
in
simulating
ozone
formation
and
approximates
other
species
reasonably
well
(
e.
g.,
hydroxyl
radical,
hydroperoxy
radical,
the
operator
radical,
hydrogen
peroxide,
nitric
acid,
and
peroxyacetyl
nitrate).
2
The
REMSAD
model
was
subjected
to
a
scientific
peerreview
(
Seigneur
et
al.,
1999)
and
EPA
has
incorporated
the
major
science
improvements
that
were
recommended
by
the
peer­
review
panel.
These
improvements
were
included
in
the
version
of
REMSAD
used
for
the
NPR
modeling.
Specifically,

the
following
updates
have
been
implemented
into
REMSAD
Version
7.06,
which
was
used
for
the
proposed
CAIR
control
strategy
simulations:
(
1)
the
nighttime
chemistry
treatment
9
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
was
updated
to
improve
the
treatment
of
the
gas
phase
species
NO3
and
N2O5;
(
2)
the
effects
of
temperature
and
pressure
dependence
on
chemical
rates
were
added;
(
3)
the
MARS­
A
aerosol
partitioning
module
was
added
for
calculating
particle
and
gas
phase
fractions
of
nitrate;
(
4)
aqueous
phase
formation
of
sulfate
was
updated
by
including
reactions
for
oxidation
of
SO2
by
ozone
and
oxygen,
(
5)

peroxynitric
acid
(
PNA)
chemistry
was
added;
and
(
6)
a
module
for
calculating
biogenic
and
anthropogenic
secondary
organic
aerosols
was
developed
and
integrated
into
REMSAD.

We
believe
that
these
changes
adequately
respond
to
the
peer
review
comments
and
have
bolstered
the
scientific
credibility
of
this
model.

(
ii)
Use
of
CMAQ
Instead
of
REMSAD
for
PM2.5
Modeling
Comment:
Some
commenters
claimed
that
REMSAD
is
outdated
and
that
other
models
with
more
sophisticated
science
are
available.
Commenters
said
that
EPA
should
utilize
the
best
available
science
through
use
of
the
most
comprehensive
photochemical
model
for
simulating
aerosols.
Commenters
specifically
stated
that
EPA
should
use
more
recently
developed
models
such
as
the
CMAQ
model
or
the
aerosol
version
of
the
Comprehensive
Air
Quality
Model
with
Extensions
(
CAMx­
PM).
10
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Response:
The
EPA
agrees
that
photochemical
models
are
now
available
that
are
more
scientifically
sophisticated
than
REMSAD.
In
this
regard,
and
in
response
to
commenters'

recommendations
on
specific
models,
EPA
has
selected
CMAQ
as
the
modeling
tool
for
the
final
CAIR
modeling
analysis.
As
stated
above,
the
CMAQ
model
is
a
publicaly
available,

peerreviewed
state­
of­
the­
science
model
with
a
number
of
science
attributes
that
are
critical
for
accurately
simulating
the
oxidant
precursors
and
non­
linear
organic
and
inorganic
chemical
relationships
associated
with
the
formation
of
sulfate,
nitrate,
and
organic
aerosols.
As
listed
above,
the
important
science
aspects
of
CMAQ
that
are
superior
to
REMSAD
include:
(
1)
updated
gaseous/
heterogeneous
chemistry
that
provides
the
basis
for
the
formation
of
nitrates
and
includes
a
current
inorganic
nitrate
partitioning
module;
(
2)
in­
cloud
sulfate
chemistry,

which
accounts
for
the
non­
linear
sensitivity
of
sulfate
formation
to
varying
pH;
(
3)
a
state­
of­
the­
science
secondary
organic
aerosol
module
that
includes
a
more
comprehensive
gas­
particle
partitioning
algorithm
from
both
anthropogenic
and
biogenic
secondary
organic
aerosol;
and
(
4)
the
full
CB­
IV
chemistry
mechanism,
which
provides
a
complete
simulation
of
aerosol
precursor
oxidants.
11
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
(
iii)
Model
Evaluation
Comment:
A
number
of
commenters
claimed
that
EPA's
air
quality
model
evaluation
for
1996
was
deficient
because
it
lacked
sufficient
ambient
measurements,
especially
in
urban
areas,
to
judge
model
performance.
Commenters
said
that
EPA
should:
(
1)
update
the
evaluation
to
a
more
recent
time
period
in
order
to
take
advantage
of
greatly
expanded
ambient
PM2.5
species
measurements,
especially
in
urban
areas;
and
(
2)
calculate
model
performance
statistics
over
monthly
and/
or
seasonal
time
periods
using
daily/
weekly
observed/
model­
predicted
data
pairs.

Some
commenters
said
that
the
1996
data
were
so
limited
that
it
is
not
possible
to
determine
whether
REMSAD
could
be
used
with
confidence
to
assess
the
effects
of
emissions
changes.
Still,
other
commenters
said
that
the
performance
of
REMSAD
for
the
1996
modeling
platform
was
poor.

Commenters
acknowledged
that
there
are
no
universally
accepted
or
EPA­
recommended
quantitative
criteria
for
judging
the
acceptability
of
PM2.5
model
performance.
In
the
absence
of
such
model
performance
acceptance
criteria,

some
commenters
said
that
performance
should
be
judged
by
comparing
EPA's
model
performance
results
to
the
range
of
results
obtained
by
other
groups
in
the
air
quality
modeling
12
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
community
who
conducted
other
recent
regional
PM2.5
model
applications.
A
few
commenters
also
identified
specific
model
performance
ranges
and
criteria
that
they
said
should
be
achievable
for
sulfate
and
PM2.5,
given
the
current
state­
of­
science
for
aerosol
modeling
and
measurement
uncertainty.
The
specific
values
cited
by
these
commenters
are
±
30
percent
to
±
50
percent
for
fractional
bias,
50
percent
to
75
percent
for
fractional
error,
and
50
percent
for
normalized
error.

Response:
The
EPA
agrees
that
the
limited
amount
of
ambient
PM2.5
species
data
available
in
1996
affected
our
ability
to
evaluate
model
performance,
especially
in
urban
areas,
and
there
were
deficiencies
in
the
performance
of
REMSAD
using
the
1996
model
inputs.
Also,
EPA
agrees
that
a
model
evaluation
should
be
performed
for
a
more
recent
time
period
in
order
to
address
these
concerns.
Thus,
we
conclude
that
the
1996
modeling
platform
which
includes
1996
emissions,

1996
meteorology,
and
1996
ambient
data
should
be
updated
and
improved,
as
recommended
by
commenters.

The
EPA
has
developed
a
new
modeling
platform
which
includes
emissions,
meteorological
data,
and
other
model
inputs
for
2001.
This
platform
was
used
to
confirm
the
ability
of
our
modeling
system
to
replicate
ambient
PM2.5
13
3
The
2001
modeling
platform
is
described
in
full
in
the
NFR
EITSD
and
NFR
AQMTSD.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
and
component
species
in
both
urban
and
rural
areas
and,

thus,
establish
the
credibility
of
this
platform
for
PM2.5
modeling
as
part
of
CAIR.
3
In
2001,
there
was
an
extensive
set
of
ambient
PM2.5
measurements
including
133
urban
Speciation
Trends
Network
(
STN)
monitoring
sites
across
the
nation,
with
105
of
these
in
the
East.
This
network
did
not
exist
in
1996.
Also,
the
number
of
mainly
suburban
and
rural
monitoring
sites
in
the
Clean
Air
Status
and
Trends
Network
(
CASTNET)
and
Interagency
Monitoring
of
Protected
Visual
Environments
(
IMPROVE)
network
has
increased
to
over
200
in
2001,
compared
to
approximately
120
operating
in
1996.

The
EPA
evaluated
CMAQ
for
the
2001
modeling
platform
using
the
extensive
set
of
2001
monitoring
data
for
PM2.5
species.
The
evaluation
included
a
statistical
analysis
in
which
the
model
predictions
and
measurements
were
paired
in
space
and
in
time
(
i.
e.,
daily
or
weekly
to
be
consistent
with
the
sampling
protocol
of
the
monitoring
network).

Model
performance
statistics
were
calculated
for
each
network
with
separate
statistics
for
sites
in
the
West
and
14
4
For
the
purposes
of
this
analysis,
we
have
defined
"
East"
as
the
area
to
the
east
of
100
degrees
longitude,
which
runs
from
approximately
the
eastern
half
of
Texas
through
the
eastern
half
of
North
Dakota.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
the
East.
4
In
response
to
comments
that
performance
statistics
should
be
calculated
over
monthly
and/
or
seasonal
time
periods,
we
elected
to
use
seasonal
time
periods
in
order
to
be
consistent
with
our
use
of
quarterly
average
PM2.5
species
as
part
of
the
procedure
for
projecting
future
concentrations,
as
described
below
in
section
VI.
B.
1.
In
addition,
the
sampling
frequency
at
the
CASTNET,
IMPROVE,

and
STN
sites
may
not
provide
sufficient
samples
in
a
1­

month
period
to
provide
a
robust
calculation
of
model
performance
statistics.
Details
of
EPA's
model
evaluation
for
CMAQ
using
the
2001
modeling
platform
are
in
the
report
"
Updated
CMAQ
Model
Performance
Evaluation
for
2001"
which
can
be
found
in
the
docket
for
today's
rule.

The
EPA
agrees
that
there
are
no
universally
accepted
performance
criteria
for
PM2.5
modeling
and
that
performance
should
be
judged
by
comparison
to
the
performance
found
by
other
groups
in
the
air
quality
modeling
community.
In
this
respect,
we
have
compared
our
CMAQ
2001
model
performance
results
to
the
range
of
performance
found
in
other
recent
15
5
These
other
modeling
studies
represent
a
wide
range
of
modeling
analyses
which
cover
various
models,
model
configurations,
domains,
years
and/
or
episodes,
chemical
mechanisms,
and
aerosol
modules.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
regional
PM2.5
model
applications
by
other
groups.
5
Details
of
this
comparison
can
be
found
in
the
CMAQ
evaluation
report.
Below
is
a
summary
of
performance
results
from
other,
non­
EPA
modeling
studies,
for
summer
sulfate
and
winter
nitrate.
It
should
be
noted
that
nitrate
and
sulfate
are
the
two
species
most
relevant
for
CAIR.

Overall,
the
general
range
of
fractional
bias
(
FB)
and
fractional
error
(
FE)
statistics
for
the
better
performing
model
applications
are
as
follows:

­
summer
sulfate
is
in
the
range
of
­
10
percent
to
+
30
percent
for
FB
and
35
percent
to
50
percent
for
FE;
and
­
winter
nitrate
is
in
the
range
of
+
50
percent
to
+
70
percent
for
FB
and
85
percent
to
105
percent
for
FE.

The
corresponding
performance
statistics
for
EPA's
2001
CMAQ
application
as
well
as
the
1996
REMSAD
application
used
for
the
proposal
modeling
are
provided
in
Table
VI­
1.

Table
VI­
1.
Selected
Performance
Evaluation
Statistics
from
the
CMAQ
2001
Simulation
and
the
REMSAD
1996
Simulation.

Eastern
U.
S.
CMAQ
2001
REMSAD
1996
FB
(%)
FE
(%)
FB
(%)
FE
(%)
16
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Sulfate
(
Summer)
STN
14
44
­
­

IMPROVE
10
42
­
20
51
CASTNet
3
22
­
21
59
Nitrate
(
Winter)
STN
15
73
­
­

IMPROVE
21
92
67
103
The
results
indicate
that
the
performance
for
CMAQ
in
2001
is
within
the
range
or
better
than
that
found
by
other
groups
in
recent
applications.
The
performance
also
meets
the
benchmark
goals
suggested
by
several
commenters.
In
addition,
the
CMAQ
performance
is
considerably
improved
over
that
of
the
REMSAD
1996
performance
for
summer
sulfate
and
winter
nitrate,
which
were
near
the
bounds
or
outside
the
range
of
other
recent
applications.

The
CMAQ
model
performance
results
give
us
confidence
that
our
applications
of
CMAQ
using
the
new
modeling
platform
provide
a
scientifically
credible
approach
for
assessing
PM2.5
concentrations
for
the
purposes
of
CAIR.

b.
Ozone
Air
Quality
Modeling
Platform
and
Model
Evaluation
Overview
The
EPA
used
the
CAMx,
version
3.10
in
the
NPR
to
assess
8­
hour
ozone
concentrations
and
the
impacts
of
ozone
and
ozone
precursor
transport
on
elevated
levels
of
ozone
across
the
eastern
U.
S.
The
CAMx
is
a
publicly
available
Eulerian
model
that
accounts
for
the
processes
that
are
17
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
involved
in
the
production,
transport,
and
destruction
of
ozone
over
a
specified
three­
dimensional
domain
and
time
period.
The
CAMx
model
was
run
with
1995/
96
base
year
emissions
to
evaluate
the
performance
of
the
modeling
platform
to
replicate
observed
concentrations
during
the
three
1995
episodes.
This
evaluation
was
comprised
principally
of
statistical
assessments
of
hourly,
1­
hour
daily
maximum,
and
8­
hour
daily
maximum
ozone
predictions.

As
described
in
the
NPR
AQMTSD,
model
performance
of
CAMx
for
ozone
was
judged
against
the
results
from
previous
regional
ozone
model
applications.
This
analysis
indicates
that
model
performance
was
comparable
to
or
better
than
that
found
in
previous
applications
and
is,
therefore,
acceptable
for
the
purposes
of
CAIR
ozone
modeling.

The
EPA
did
not
receive
comments
on
the
CAMx
model
or
the
model
performance
for
ozone.
The
EPA
did
receive
comments
on
the
choice
of
episodes
for
ozone
modeling,
the
meteorological
data
for
these
episodes,
the
spatial
resolution
of
our
modeling,
and
consistency
between
ozone
and
PM2.5
modeling
in
terms
of
methods
for
projecting
future
air
quality
concentrations.
As
described
below
and
in
the
RTC
document
and
NFR
AQMTSD,
we
continue
to
believe
that:

(
1)
the
three
1995
episodes
are
representative
episodes
for
18
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
regional
modeling
of
8­
hour
ozone;
and
(
2)
the
meteorological
data
for
these
episodes
and
spatial
resolution
are
adequate
for
use
in
our
modeling
for
CAIR.

Thus,
the
ozone
air
quality
assessments
in
today's
rule
rely
on
CAMx
modeling
of
meteorological
data
for
the
three
1995
episodes
for
the
domain
and
spatial
resolution
used
for
the
NPR.
As
discussed
below,
we
ran
CAMx
for
the
updated
2001
emissions
inventory
and
the
updated
2010
and
2015
Base
Case
inventories
as
part
of
the
process
to
project
8­
hour
ozone
for
these
future
year
scenarios.
We
revised
our
method
of
projecting
future
ozone
concentrations
to
be
consistent
with
the
method
we
are
using
for
PM2.5.

c.
Model
Grid
Cell
Configuration
As
described
in
the
NPR
AQMTSD,
the
PM2.5
modeling
for
the
proposal
was
performed
for
a
domain
(
i.
e.,
area)

covering
the
48
States
and
adjacent
portions
of
Canada
and
Mexico.
Within
this
domain,
the
model
predictions
were
calculated
for
a
grid
network
with
a
spatial
resolution
of
approximately
36
km.
Our
8­
hour
ozone
modeling
for
proposal
was
performed
using
a
nested
grid
network.
The
outer
portion
of
this
grid
has
a
spatial
resolution
of
approximately
36
km.
The
inner
"
nested"
area,
which
covers
a
large
portion
of
the
eastern
U.
S.,
has
a
resolution
of
19
6
U.
S.
EPA,
2000:
Draft
Guidance
for
Demonstrating
Attainment
of
the
Air
Quality
Goals
for
PM2.5
and
Regional
Haze;
Draft
1.1,
Office
of
Air
Quality
Planning
and
Standards,
Research
Triangle
Park,
NC.

7
U.
S.
EPA,
1999:
Draft
Guidance
on
the
Use
of
Models
and
Other
Analyses
in
Attainment
Demonstrations
for
the
8­
Hour
Ozone
NAAQS,
Office
of
Air
Quality
Planning
and
Standards,
Research
Triangle
Park,
NC.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
approximately
12
km.

Comment:
Some
commenters
said
that
the
36
km
grid
cell
size
used
by
EPA
in
modeling
PM2.5
and
the
36
km/
12
km
grid
resolution
used
for
ozone
modeling
are
too
coarse
and
are
inconsistent
with
EPA's
draft
modeling
guidance.

Response:
We
disagree
with
these
comments
and
continue
to
believe
that
the
grid
dimensions
for
our
PM2.5
modeling
and
our
8­
hour
ozone
modeling
are
not
too
coarse
nor
are
they
inconsistent
with
our
draft
guidance
documents
for
PM2.5
modeling6
and
ozone
modeling.
7
The
draft
guidance
for
PM2.5
modeling
states
that
36
km
resolution
is
acceptable
for
regional
scale
applications
in
portions
of
the
domain
outside
of
nonattainment
areas.
For
portions
of
the
domain
which
cover
nonattainment
areas,
12
km
resolution
or
less
is
recommended
by
the
guidance.
However,
as
stated
in
the
guidance
document,
these
recommendations
were
based
on
guidance
for
8­
hour
ozone
modeling
because
there
was
a
lack
of
PM2.5
modeling
at
different
grid
resolutions
at
the
time
20
8
VISTAS
Emissions
and
Air
Quality
Modeling­
Phase
I
Task
4cd
Report:
Model
Performance
Evaluation
and
Model
Sensitivity
Tests
for
Three
Phase
I
Episodes.
ENVIRON
International
Corporation,
Alpine
Geophysics,
and
University
of
California
at
Riverside,
September
7,
2004.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
the
guidance
was
drafted.
In
addition,
the
PM2.5
guidance
states
that
exceptions
to
these
recommendations
can
be
made
on
a
case­
by­
case
basis.

For
several
reasons,
we
believe
that
36
km
resolution
is
sufficient
for
PM2.5
modeling
for
the
purposes
of
CAIR.

First,
recent
analyses
that
compare
36
km
to
12
km
modeling
of
PM2.58
indicate
that
spatial
mean
concentrations
of
gas
phase
and
aerosol
species
at
36
km
and
12
km
are
quite
similar.
A
comparison
of
model
predictions
versus
observations
indicates
that
the
model
performance
is
similar
at
12
km
and
36
km
in
both
rural
and
urban
areas.
Thus,

using
12
km
resolution
does
not
necessarily
provide
any
additional
confidence
in
the
results.
Second,
ambient
measurements
of
sulfate
and
to
a
significant
extent
nitrate,

which
are
the
pollutants
of
most
importance
for
CAIR,
do
not
exhibit
large
spatial
differences
between
rural
and
urban
areas,
as
described
elsewhere
in
today's
rule.
This
implies
that
it
is
not
necessary
to
use
fine
resolution
modeling
in
order
to
properly
capture
the
regional
concentration
patterns
of
these
pollutants.
21
9
Irwin,
J.
et.
al.
"
Examination
of
model
predictions
at
different
horizontal
grid
resolutions."
Submitted
for
Publication
to
Environmental
Fluid
Mechanics.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Our
draft
8­
hour
ozone
modeling
guidance
recommends
using
36
km
resolution
for
regional
modeling
with
nested
grid
cells
not
exceeding
12
km
over
urban
portions
of
the
modeling
domain.
The
guidance
states
that
4
to
5
km
resolution
for
urban
areas
is
preferred,
if
feasible.
In
addition,
if
12
km
modeling
is
used
then
plume­
in­
grid
treatment
for
large
point
sources
of
NOx
should
be
considered.

Our
modeling
for
CAIR
is
consistent
with
this
guidance
in
that
we
use
36
km
resolution
for
the
outer
portions
of
the
region;
12
km
resolution
covering
nearly
all
urban
areas
in
the
domain;
and
a
plume­
in­
grid
algorithm
for
major
NOx
point
sources
in
the
region.
In
addition,
analyses
that
compare
model
12
km
resolution
to
4
km
resolution
for
portions
of
our
1995
episodes
indicate
that
the
spatial
fields
predicted
at
both
12
km
and
4
km
have
many
common
features
in
terms
of
the
areas
of
high
and
low
ozone.
9
In
a
comparison
of
model
predictions
to
observation,
the
12
km
modeling
was
found
to
be
somewhat
more
accurate
than
the
finer
4
km
modeling.

2.
Emissions
Inventory
Data
22
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
For
the
proposed
rule,
emissions
inventories
were
created
for
the
48
contiguous
States
and
the
District
of
Columbia.
These
inventories
were
estimated
for
a
2001
base
year
to
reflect
current
emissions
and
for
2010
and
2015
future
baseline
scenarios.
The
inventories
were
prepared
for
electric
generating
units
(
EGUs),
industrial
and
commercial
sources
(
non­
EGUs),
stationary
area
sources,

on­
road
vehicles,
and
non­
road
engines.
The
inventories
contained
both
annual
and
typical
summer
season
day
emissions
for
the
following
pollutants:
oxides
of
nitrogen
(
NOx);
volatile
organic
compounds
(
VOC);
carbon
monoxide
(
CO);
sulfur
dioxide
(
SO2);
direct
particulate
matter
with
an
aerodynamic
diameter
less
than
10
micrometers
(
PM10)
and
less
than
2.5
micrometers
(
PM2.5);
and
ammonia
(
NH3).
A
summary
of
the
development
of
these
inventories
is
provided
below.
Additional
information
on
the
emissions
inventory
used
for
proposal
can
be
found
in
the
NPR
AQMTSD.

Because
the
complete
2001
National
Emission
Inventory
(
NEI)
and
future­
year
projections
consistent
with
that
NEI
were
not
available
in
a
form
suitable
for
air
quality
modeling
when
needed
for
the
proposal,
we
developed
a
reasonably
representative
"
proxy"
inventory
for
2001.
For
the
EGU,
mobile,
and
non­
road
emissions
sectors,
1996­
to­
23
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
2001
adjustment
ratios
were
created
by
dividing
State­
level
total
emissions
for
each
pollutant
for
2001
by
the
corresponding
consistent
1996
emissions.
These
adjustment
ratios
were
then
multiplied
by
the
REMSAD­
ready
1996
emissions
for
these
two
sectors
to
produce
REMSAD­
ready
files
for
the
2001
proxy.
For
non­
EGUs
and
stationary
area
sources,
linear
interpolations
were
performed
between
the
REMSAD­
ready
1996
emissions
and
the
REMSAD­
ready
2010
Base
Case
emissions
to
produce
2001
proxy
emissions
for
these
two
sectors.
Details
on
the
creation
of
the
2001
proxy
inventory
used
for
proposal
are
provided
in
the
NPR
AQMTSD.

The
NPR
future
2010
and
2015
base
case
emissions
reflect
projected
economic
growth
and
control
programs
that
are
to
be
implemented
by
2010
and
2015,
respectively.

Control
programs
included
in
these
future
base
cases
include
those
State,
local,
and
Federal
measures
already
promulgated
and
other
significant
measures
expected
to
be
promulgated
before
the
final
rule
is
implemented.
Future
year
2010
and
2015
Base
Case
EGU
emissions
were
obtained
from
versions
2.1
and
2.1.6
of
the
Integrated
Planning
Model
(
IPM).

Comment:
Several
commenters
stated
that
the
emission
inventory
used
for
the
"
proxy"
2001
base
year
was
not
sufficient
for
the
rulemaking,
primarily
because
it
was
24
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
developed
from
a
1996
modeling
inventory
by
applying
various
adjustment
factors.
Commenters
suggested
that:
(
1)
more
upto
date
inventories
were
now
available
and
should
be
used;

(
2)
the
most
recent
Continuous
Emissions
Monitoring
(
CEM)

data
or
throughput
information
should
be
used
to
derive
a
2001
EGU
inventory;
and
(
3)
EPA
should
use
the
2001
MOBILE6
and
NONROAD2002
models
for
estimating
on­
road
mobile
and
non­
road
engine
emissions,
respectively.

Response:
The
EPA
believes
that
the
base
year
for
modeling
should
be
as
recent
as
possible,
given
the
availability
of
nationally
complete
emissions
estimates
and
ambient
monitoring
data.
For
the
analyses
of
the
final
rule,
EPA
has
used
a
base
year
inventory
developed
specifically
for
2001.
The
base
year
inventory
for
the
electric
utility
sector
now
uses
measured
CEM
emissions
data
for
2001.
The
non­
EGU
point
source
and
stationary­
area
source
sectors
are
based
on
the
final
1999
NEI
data
submittals
from
State,

local,
and
Tribal
air
agencies.
This
inventory
is
the
latest
available
quality­
assured
and
reviewed
national
emission
data
set
for
these
sectors.
The
1999
data
for
non­

EGU
point
and
stationary­
area
sources
were
projected
to
represent
a
2001
inventory
using
State/
county­
specific
and
sector­
specific
growth
rates.
The
on­
road
mobile
inventory
25
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
uses
MOBILE
version
6.2
and
the
non­
road
engines
inventory
uses
the
NONROAD2004
model,
both
with
updated
input
parameters
to
calculate
emissions
for
2001.
More
detailed
information
on
the
development
of
the
emissions
inventories
can
be
found
in
the
NFR
EITSD.

Comment:
Commenters
stated
that
EPA
failed
to
develop
an
accurate
and
comprehensive
ammonia
emission
inventory
from
soil,
fertilizer,
and
animal
husbandry
sources.

Response:
The
2001
inventory
used
for
the
analyses
for
the
final
rule
includes
a
new
national
county­
level
ammonia
inventory
developed
by
EPA
using
the
latest
emission
rates
selected
based
on
a
comprehensive
literature
review,
and
activity
levels
as
provided
by
the
U.
S.
Census
of
Agriculture
for
animal
husbandry.
The
2001
inventory
from
fertilizer
application
sources
was
compiled
from
State
and
local
submissions
to
EPA
for
1999,
augmented
as
necessary
with
EPA
estimates,
and
grown
to
2001
using
State/

countyspecific
and
category­
specific
growth
rates.
With
regard
to
background
soil
emissions
of
NH3,
EPA
believes
that
the
current
state
of
understanding
of
background
soil
ammonia
releases
and
sinks
is
insufficient
to
warrant
including
these
emission
sources
in
modeling
inventories
at
this
time.

Comment:
Two
commenters
indicated
that
EPA
should
revise
26
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
2010
and
2015
base
case
emissions
by
improving
the
methods
for
estimating
economic
growth
and
not
rely
on
the
Bureau
of
Economic
Analysis
(
BEA)
data
used
for
proposal.

Response:
In
response
to
these
comments,
EPA
has
refined
its
economic
growth
projections.
In
addition
to
updated
versions
of
the
MOBILE6,
NONROAD,
and
IPM
models,
EPA
developed
new
economic
growth
rates
for
stationary,
area,

and
non­
EGU
point
sources.
For
these
two
sectors,
the
final
approach
uses
a
combination
of:
(
1)
regional
or
national
fuel­
use
forecast
data
from
the
U.
S.
Department
of
Energy
for
source
types
that
map
to
fuel
use
sectors
(
e.
g.,

commercial
coal,
industrial
natural
gas);
(
2)
State­
specific
growth
rates
from
the
Regional
Economic
Model,
Inc.
(
REMI)

Policy
Insight
®
model,
version
5.5;
and
(
3)
forecasts
by
specific
industry
organizations
and
Federal
agencies.
For
more
detail
on
the
growth
methodologies,
please
refer
to
the
NFR
EITSD.

3.
Meteorological
Data
In
order
to
solve
for
the
change
in
pollutant
concentrations
over
time
and
space,
the
air
quality
model
requires
certain
meteorological
inputs
that,
in
part,
govern
the
formation,
transport,
and
destruction
of
pollutant
material.
Two
separate
sets
of
meteorological
inputs
were
27
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
used
in
the
air
quality
modeling
completed
as
part
of
the
NPR.
The
meteorological
input
files
for
the
proposal
PM2.5
modeling
were
developed
from
a
Fifth­
Generation
NCAR
/

Pennsylvania
State
Mesoscale
Model
(
MM5)
model
simulation
for
the
entire
year
of
1996.
The
gridded
meteorological
data
for
the
three
1995
ozone
episodes
were
developed
using
the
Regional
Atmospheric
Modeling
System
(
RAMS).
Both
of
these
models
are
publicly­
available,
widely­
used,
prognostic
meteorological
models
that
solve
the
full
set
of
physical
and
thermodynamic
equations
which
govern
atmospheric
motions.
Further,
each
of
these
specific
meteorological
data
sets
has
been
utilized
in
past
EPA
rulemaking
modeling
analyses
(
e.
g.,
the
Nonroad
Land­
based
Diesel
Engines
Standards).

Comment:
Several
commenters
claimed
that
the
1996
meteorological
modeling
data
used
to
support
the
fine
particulate
modeling
were
outdated
and
non­
representative.

We
also
received
recommendations
from
commenters
on
benchmarks
to
be
used
as
goals
for
judging
the
adequacy
of
meteorological
modeling.

Response:
The
EPA
draft
PM2.5
modeling
guidance
which
provides
general
recommendations
on
meteorological
periods
to
model
for
PM2.5
purposes
lists
three
primary
general
28
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
criteria
for
consideration:
a)
variety
of
meteorological
conditions;
b)
existence
of
an
extensive
air
quality/
meteorological
data
bases;
and
c)
sufficient
number
of
days.
The
approach
recommended
in
the
guidance
for
modeling
annual
PM2.5
is
to
use
a
single,
representative
year.
Based
on
the
comments
received
and
the
criteria
outlined
in
the
guidance,
EPA
developed
meteorological
data
for
the
entire
calendar
year
of
2001.
This
year
was
chosen
for
the
PM2.5
modeling
platform
based
on
several
factors,

specifically:
(
1)
it
corresponds
to
the
most
recent
set
of
emissions
data;
(
2)
there
are
considerable
ambient
PM2.5
species
data
for
use
in
model
evaluation
(
as
described
in
section
VI.
A.
1.,
above);
and
(
3)
Federal
Reference
Method
(
FRM)
PM2.5
data
for
this
year
are
included
in
the
calculation
of
the
most
recent
PM2.5
design
values
used
for
designating
PM2.5
nonattainment
areas.
In
view
of
these
factors,
EPA
believes
that
2001
meteorology
are
representative
for
PM2.5
modeling
for
the
purposes
of
this
rule.

The
new
2001
meteorological
data
used
for
PM2.5
modeling
were
derived
from
an
updated
version
of
the
MM5
model
used
for
the
1996
meteorology
used
for
proposal.
The
version
of
MM5
used
for
the
2001
simulation
contains
more
29
10
Environ,
Enhanced
Meteorological
Modeling
and
Performance
Evaluation
for
Two
Texas
Ozone
Episodes.
August
2001.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
sophisticated
physics
options
with
respect
to
features
like
cloud
microphysics
and
land­
surface
interactions,
and
more
refined
vertical
resolution
of
the
atmosphere
compared
to
the
version
used
for
modeling
1996
meteorology.
While
there
are
currently
no
universally
accepted
criteria
for
judging
the
adequacy
of
meteorological
model
performance,
EPA
compared
the
2001
MM5
model
performance
against
the
benchmark
goals10
recommended
by
some
commenters.
The
benchmark
goals
suggest
that
temperature
bias
should
be
within
the
range
of
approximately
+
0.5
degrees
C
and
errors
less
than
or
equal
to
2.0
degrees
C
are
typical.

In
general,
the
model
performance
statistics
for
our
2001
meteorological
modeling
are
in
line
with
the
above
benchmark
goals.
Specfically,
the
mean
temperature
bias
of
our
2001
meteorological
modeling
was
approximately
0.6
degrees
C
and
the
mean
error
was
approximately
2.0
degrees
C.
The
evaluation
of
the
2001
MM5
for
humidity
(
water
vapor
mixing
ratio)
shows
biases
of
less
than
0.5
g/
kg
and
errors
of
approximately
1
g/
kg,
which
compare
favorably
to
the
goals
of
+
1
g/
kg
for
bias
and
2
g/
kg
or
less
error.
Model
performance
for
winds
in
our
2001
simulation
was
also
30
11
Hogrefe,
C.
et.
al.
"
Evaluating
the
performance
of
regional­
scale
photochemical
modeling
systems:
Part
1­
meteorological
predictions."
Atmospherics
Environment,
vol.
35
(
2001),
pp.
4159­
4174.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
improved
compared
to
what
has
historically
been
found
in
MM5
modeling
studies.
The
index
of
agreement
for
surface
winds
in
the
2001
case
equaled
0.86,
which
is
far
better
than
the
benchmark
goal
of
0.60.
The
precipitation
evaluation
results
show
that
the
model
generally
replicates
the
observed
data,
but
is
overestimating
precipitation
in
the
summer
months.
More
information
about
the
model
performance
evaluation
and
the
MM5
configuration
is
provided
in
the
NFR
AQMTSD.

Comment:
Several
groups
criticized
the
lack
of
quantitative
meteorological
model
evaluation
data
for
the
1995
RAMS
meteorological
modeling
used
for
episodic
ozone
modeling.

Response:
A
peer­
reviewed,
quantitative
evaluation
of
the
RAMS
model
performance
for
this
meteorological
period
is
provided
by
Hogrefe,
et.
al.
11
This
analysis
was
performed
using
RAMS
predictions
for
June
through
August
of
1995.
The
results
show
that
the
RAMS
biases
and
errors
are
generally
in
line
with
past
meteorological
model
simulations
by
other
groups
outside
EPA.
The
EPA
remains
satisfied
that
the
1995
RAMS
meteorological
inputs
for
the
three
CAMx
ozone
modeling
31
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
episodes
are
of
sufficient
quality
and
we
have
continued
to
use
these
inputs
for
the
ozone
analyses
for
the
final
rule.

Comment:
The
EPA
received
several
comments
on
the
episodes
selected
for
ozone
modeling.
There
was
general
criticism
that
the
ozone
modeling
did
not
follow
EPA's
own
guidance
for
the
selection
of
episodes.
Additionally,
there
was
specific
criticism
that
the
episodes
did
not
provide
for
a
reasonable
test
of
the
8­
hour
ozone
NAAQS
in
some
areas.

Response:
The
draft
8­
hour
ozone
guidance
recommends,
at
a
minimum,
that
four
criteria
be
used
to
select
episodes
which
are
appropriate
to
model.
This
guidance
is
generally
intended
for
local
attainment
demonstrations,
as
opposed
to
regional
transport
analyses,
but
it
does
recommend
that
in
applying
a
regional
model
one
should
choose
episodes
meeting
as
many
of
the
criteria
as
possible,
though
it
acknowledges
there
may
be
tradeoffs.
Given
the
large
number
of
nonattainment
areas
within
the
ozone
domain,
it
would
be
extremely
difficult
to
assess
the
criteria
on
a
area­
by­
area
basis.
However,
from
a
general
perspective,
the
1995
episodes
address
all
of
the
primary
criteria,
which
include:

1)
a
variety
of
meteorological
conditions;
2)
measured
ozone
values
that
are
close
to
current
air
quality;
3)
extensive
meteorological
and
air
quality
data;
and
4)
a
sufficient
32
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
number
of
days.
More
detail
is
provided
in
the
NFR
AQMTSD,

but
here
is
a
brief
description
of
how
each
of
the
four
primary
criteria
are
met
by
the
1995
cases.

With
regard
to
the
criteria
of
meteorological
variations,
we
have
completed
inert
tracer
simulations
for
each
of
the
three
1995
episodes
that
show
different
transport
patterns
in
all
three
cases.
For
example
the
June
case
involves
east­
to­
west
transport;
the
July
case
involves
west­
to­
east
transport;
and
the
August
case
involves
southto
north
transport.
In
a
separate
analysis
to
determine
whether
the
1995
modeling
days
correspond
to
commonly
occurring
and
ozone­
conducive
meteorology,
EPA
has
applied
a
multi­
variate
statistical
approach
for
characterizing
daily
meteorological
patterns
and
investigating
their
relationship
to
8­
hour
ozone
concentrations
in
the
eastern
U.
S.
Across
the
16
sites
for
which
the
analysis
was
completed,
there
were
five
to
six
distinct
sets
of
meteorological
conditions,

called
regimes,
that
occurred
during
the
ozone
seasons
studied.
An
analysis
of
the
8­
hour
daily
maximum
ozone
concentrations
for
each
of
the
meteorological
regimes
was
undertaken
to
determine
the
distribution
of
ozone
concentrations
and
the
frequency
of
occurrence
of
each
regimes.
The
EPA
determined
that
between
60
and
70
percent
33
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
of
the
episode
days
we
modeled
are
associated
with
the
most
frequently
occurring,
high
ozone
potential,
meteorological
regimes.
These
results
also
provide
support
that
the
episodes
being
modeled
are
representative
of
conditions
present
when
high
ozone
concentrations
are
measured
throughout
the
modeling
domain.
For
the
second
criteria,

EPA
has
completed
an
analysis
which
shows
that
the
1995
episodes
contain
observed
8­
hour
daily
maximum
ozone
values
that
approximate
recent
ambient
concentrations
over
the
eastern
U.
S.
Additional
analyses
performed
by
EPA
and
others
have
concluded
that
each
of
the
three
episodes
involves
widespread
areas
of
elevated
ozone
concentrations.

The
synoptic
meteorological
pattern
of
the
July
1995
episode
has
been
identified
by
one
of
the
commenters
as
representing
a
classic
set
of
conditions
necessary
for
high
ozone
over
the
eastern
U.
S.
While
the
ozone
was
not
quite
as
widespread
in
the
June
and
August
1995
episodes,
these
periods
also
contained
exceedances
of
the
8­
hour
ozone
NAAQS
in
most
portions
of
the
region.

We
believe
that
there
is
ample
meteorological
and
air
quality
data
available
to
support
an
evaluation
of
the
modeling
for
these
episodes.
Specifically,
there
were
over
700
ozone
monitors
reporting
across
the
domain
for
use
in
34
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
model
evaluation.
As
noted
above,
the
model
performance
for
these
episodes
compares
favorably
to
the
recommendations
in
EPA's
urban
modeling
guidance.
In
addition,
the
modeling
period
is
comprised
of
30
days,
not
including
model
ramp­
up
periods
which
is
considerably
more
than
is
typically
used
in
an
attainment
demonstration
modeling
submitted
to
EPA
by
a
State.
Finally,
EPA's
draft
ozone
guidance
also
indicates
as
one
of
four
secondary
criteria
that
extra
weight
can
be
assigned
to
modeling
episodes
for
which
there
is
prior
experience
in
modeling.
The
1995
CAIR
ozone
episodes
have
been
successfully
used
to
drive
the
air
quality
modeling
completed
for
several
recent
notice­
and­
comment
rulemakings
(
Tier­
2,
Heavy
Duty
Engine,
and
NonRoad).
Based
on
the
analyses
discussed
above
and
the
adherence
to
the
modeling
guidance,
EPA
is
satisfied
that
the
1995
CAMx
episodes
are
appropriate
for
continued
use.

B.
How
did
EPA
Project
Future
Nonattainment
for
PM2.5
and
8­
Hour
Ozone?

1.
Projection
of
Future
PM2.5
Nonattainment
a.
Methodology
for
Projecting
Future
PM2.5
Nonattainment
In
the
NPR,
we
assessed
the
prospects
for
future
attainment
and
nonattainment
in
2010
and
2015
of
the
PM2.5
annual
NAAQS.
The
approach
for
identifying
areas
expected
35
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
to
be
nonattainment
for
PM2.5
in
the
future
involved
using
the
model
predictions
in
a
relative
way
to
forecast
current
PM2.5
design
values
to
2010
and
2015.
The
modeling
portion
of
this
approach
included
annual
simulations
for
2001
proxy
emissions
and
for
2010
and
2015
Base
Case
emissions
scenarios.
As
described
below,
the
predictions
from
these
runs
were
used
to
calculate
relative
reduction
factors
(
RRFs)
which
were
then
applied
to
current
PM2.5
design
values
from
FRM
sites
in
the
East.
This
approach
is
consistent
with
the
procedures
in
the
draft
of
EPA's
PM2.5
modeling
guidance.

To
determine
the
current
PM2.5
air
quality
for
use
in
projecting
design
values
to
the
future,
we
selected
the
higher
of
the
1999­
2001
or
2000­
2002
design
value
(
the
most
recent
ambient
data
at
the
time
of
the
proposal)
for
each
monitor
that
measured
nonattainment
in
2000­
2002.
For
those
sites
that
were
attaining
the
PM2.5
standard
based
on
their
2000­
2002
design
value,
we
used
the
value
from
this
period
as
the
starting
point
for
projecting
2010
and
2015
air
quality
at
these
sites.

The
procedure
for
calculating
future
year
PM2.5
design
values
is
called
the
Speciated
Modeled
Attainment
Test
(
SMAT).
The
test
uses
model
predictions
in
a
relative
sense
36
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
to
estimate
changes
expected
to
occur
in
each
major
PM2.5
species.
These
species
are
sulfate,
nitrate,
organic
carbon,
elemental
carbon,
crustal,
and
un­
attributed
mass.

The
relative
change
in
model­
predicted
species
concentrations
were
applied
to
ambient
species
measurements
in
order
to
project
each
species
for
the
future
year
scenarios.
We
applied
a
spatial
interpolation
to
the
IMPROVE
and
STN
speciation
data
as
a
means
for
estimating
species
composition
fractions
for
the
FRM
monitoring
sites.

Future
year
PM2.5
was
calculated
by
summing
the
projected
concentrations
of
each
species.
The
SMAT
technical
procedures,
as
applied
for
the
NPR,
are
contained
in
the
NPR
and
NPR
AQMTSD.

As
noted
above,
the
procedures
for
determining
future
year
PM2.5
concentrations
were
applied
for
each
FRM
site.

For
counties
with
only
one
FRM
site,
the
forecast
design
value
for
that
site
was
used
to
determine
whether
or
not
the
county
was
predicted
to
be
nonattainment
in
the
future.
For
counties
with
multiple
monitoring
sites,
the
site
with
the
highest
future
concentration
was
selected
for
that
county.

Those
counties
with
future
year
concentrations
of
15.1

g/
m3
(
as
rounded
up
from
15.05

g/
m3)
or
more
were
predicted
to
be
nonattainment.
Based
on
the
modeling
performed
for
the
37
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
NPR,
61
counties
in
the
East
were
forecast
to
be
nonattainment
for
the
2010
Base
Case.
Of
these,
41
were
forecast
to
remain
nonattainment
for
the
2015
Base
Case.

Comment:
Some
commenters
said
that
EPA
has
not
established
the
credibility
of
using
models
in
a
relative
sense
to
estimate
future
PM2.5
concentrations
and
that
poor
performance
of
REMSAD
for
1996
calls
into
question
the
use
of
models
to
adequately
determine
the
effects
of
changes
in
emissions.
One
commenter
said
that
a
mechanistic
model
evaluation,
in
which
model
predictions
of
PM2.5
precursor
photochemical
oxidants
are
compared
to
corresponding
measurements,
is
an
approach
for
gaining
confidence
in
the
ability
of
a
model
to
provide
a
credible
response
to
emission
changes.

Response:
The
EPA
believes
the
future
year
nonattainment
projections
should
be
based
on
using
model
predictions
in
a
relative
sense.
By
applying
the
model
in
a
relative
way,

each
measured
component
of
PM2.5
is
adjusted
upward
or
downward
based
on
the
percent
change
in
that
component,
as
determined
by
the
ratio
of
future
year
to
base
year
model
predictions.
The
EPA
feels
that
by
using
this
approach,
we
are
able
to
reduce
the
risk
that
overprediction
or
underprediction
of
PM2.5
component
species
may
unduly
affect
38
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
our
projection
of
future
year
nonattainment.

The
EPA
agrees
with
commenters
that
one
way
to
establish
confidence
in
the
credibility
of
this
approach
is
to
determine
whether
model
predictions
of
PM2.5
precursors
are
generally
comparable
to
corresponding
measured
data.
In
this
regard,
we
compared
the
CMAQ
predictions
to
observations
for
several
precursor
gases
for
which
measurements
were
available
in
2001.
These
gases
include
sulfur
dioxide,
nitric
acid,
and
ozone.

The
results
for
the
East
are
summarized
in
Table
VI­
2.

Additional
details
on
this
analysis
can
be
found
in
the
CMAQ
evaluation
report.
The
results
indicate
that
for
both
summer
and
winter
ozone,
the
fractional
bias
and
error
is
within
the
recommended
range
for
urban
scale
ozone
modeling
included
in
EPA's
draft
guidance
for
8­
hour
ozone
modeling.

For
the
other
species
examined,
there
are
limited
ambient
data
and
few
other
studies
against
which
to
compare
our
findings.
Still,
our
performance
results
for
these
species
are
within
the
range
suggested
as
acceptable
by
commenters
for
sulfate
(
i.
e.,
±
30
percent
to
±
60
percent
for
fractional
bias
and
50
percent
to
75
percent
for
fractional
error).

Thus,
EPA
believes
that
the
applications
of
CMAQ
for
CAIR
using
the
new
modeling
platform
provides
a
scientifically
39
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
credible
method
of
estimating
the
projected
impact
on
PM2.5
concentrations
expected
to
result
from
emissions
changes.

Table
VI­
2.
CMAQ
Model
Performance
Statistics
for
Ozone,
Total
Nitrate,
and
Nitric
Acid
in
the
East.

Eastern
U.
S.
CMAQ
2001
FB
(%)
FE
(%)

Ozone
AIRS
(
Summer)
13
21
AIRS
(
Winter)
­
9
31
Sulfur
Dioxide
CASTNet
(
Summer)
31
48
CASTNet
(
Winter)
39
43
Nitric
Acid
CASTNet
(
Summer)
29
39
CASTNet
(
Winter)
­
21
55
Comment:
Several
commenters
said
that
EPA's
SMAT
approach
is
flawed
and
suggested
alternative
methods
for
attributing
individual
species
mass
to
the
FRM
measured
PM2.5
mass.
One
commenter
detailed
several
different
methods
to
apportion
the
FRM
mass
to
individual
PM2.5
species.
They
refer
to
two
different
estimation
methods
as
the
"
FRM
equivalent"

approach
and
the
"
best
estimate"
approach.

Response:
The
EPA
agrees
that
alternative
methodologies
can
be
used
to
apportion
PM2.5
species
fractions
to
the
FRM
data.
We
believe
that
revising
SMAT
to
use
a
methodology
similar
to
an
"
FRM
equivalent"
methodology,
as
described
in
the
Notice
of
Data
Availability
(
69
FR
47828;
August
6,
40
12
Procedures
for
Estimating
Future
PM2.5
Values
for
the
CAIR
Final
Rule
by
Application
of
the
(
Revised)
Speciated
Modeled
Attainment
Test
(
SMAT),
docket
number
OAR­
2003­
0053­
1907.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
2004),
is
warranted.
Since
nonattainment
designation
determinations
and
future
year
nonattainment
projections
are
based
on
measured
FRM
data,
we
believe
that
the
PM2.5
species
data
should
be
adjusted
to
best
conform
to
what
is
measured
on
the
FRM
filters.
Based
on
comments,
EPA
has
revised
our
technique
for
projecting
current
PM2.5
data
to
incorporate
some
aspects
of
the
commenter's
"
FRM
equivalent"

methodology.
As
described
in
more
detail
in
the
NFR
AQMTSD,

we
believe
our
revised
methodology
to
be
the
most
scientifically
credible
estimation
of
what
is
measured
on
the
FRM
filters.

Full
documentation
of
the
revised
EPA
SMAT
methodology
is
contained
in
the
updated
SMAT
report12.
In
brief,
we
revised
the
SMAT
methodology
to
take
into
account
several
known
differences
between
what
is
measured
by
speciation
monitors
and
what
is
measured
on
FRM
filters.
Among
the
revisions
were
calculations
to
account
for
nitrate,

ammonium,
and
organic
carbon
volatilization,
blank
PM2.5
mass,
particle
bound
water,
the
degree
of
neutralization
of
sulfate,
and
the
uncertainty
in
estimating
organic
carbon
mass.
41
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Comment:
Several
commenters
noted
that
the
future
year
design
values
were
based
on
projections
of
the
1999
 
2001
and/
or
2000­
2002
FRM
monitoring
data
and
that
there
are
more
recent
design
value
data
available
for
the
2001­
2003
design
value
period.
Commenters
also
noted
that
the
2001­
2003
data
shows
lower
PM2.5
concentrations
at
the
majority
of
sites
and
therefore,
by
projecting
the
highest
design
value,
we
are
overestimating
the
future
year
PM2.5
values.

Response:
As
stated
above,
the
PM2.5
projection
methodology
in
the
NPR
used
the
higher
of
the
1999­
2001
or
2000­
2002
PM2.5
design
value
data.
The
draft
modeling
guidance
for
PM2.5
specifies
the
use
of
the
higher
of
the
three
design
value
periods
which
straddle
the
emissions
year.
The
emissions
year
is
2001
and
therefore
the
three
periods
would
be
1999­
2001,
2000­
2002,
and
2001­
2003.
Since
the
2001­
2003
data
is
now
available,
we
are
using
it
as
part
of
the
current
year
PM2.5
calculations
for
the
final
rule.

The
observation
by
a
commenter
that
the
2001­
2003
data
are
generally
lower
than
in
the
previous
two
design
value
periods
(
i.
e.,
1999­
2001
and
2000­
2002)
leads
to
the
issue
of
how
to
reduce
the
influence
of
year­
to­
year
variability
in
meteorology
and
emissions
on
our
estimate
of
current
air
quality.
As
a
consequence
of
this
year­
to­
year
variability
42
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
in
concentrations,
relying
on
design
values
from
any
single
period,
as
in
the
approach
used
for
proposal,
may
not
provide
a
robust
representation
of
current
air
quality
for
use
in
forecasting
the
future.
Specifically,
the
lower
PM2.5
values
in
2001­
2003
may
not
be
representative
of
the
current
modeling
period.
To
address
the
issue
of
year­

toyear
variability
in
the
ambient
data
we
have
modified
our
methodology
to
use
an
average
of
the
three
design
value
periods
that
straddle
the
base
year
emissions
year
(
i.
e.,

2001).
In
this
case
it
is
the
average
of
the
1999­
2001,

2000­
2002,
and
2001­
2003
design
values.
The
average
of
the
three
design
values
is
not
a
straight
5­
year
average.

Rather,
it
is
a
weighted
average
of
the
1999­
2003
period.

That
is,
by
averaging
1999­
2001,
2000­
2002,
and
2001­
2003,

the
value
from
2001
is
weighted
three
times;
2000
and
2002
are
each
weighted
twice
and
1999
and
2003
are
each
weighted
once.
This
approach
has
the
desired
benefits
of:
(
1)

weighting
the
PM2.5
values
towards
the
middle
year
of
the
5­

year
period,
which
is
the
2001
base
year
for
our
emissions
projections;
and
(
2)
smoothing
out
the
effects
of
year­

toyear
variability
in
emissions
and
meteorology
that
occurs
over
the
full
5­
year
period.
We
have
adopted
this
method
for
use
in
projecting
future
PM2.5
nonattainment
for
the
final
rule
analysis.
We
plan
to
incorporate
this
new
43
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
methodology
into
the
next
draft
version
of
our
PM2.5
modeling
guidance.

b.
Projected
2010
and
2015
Base
Case
PM2.5
Nonattainment
Counties
For
the
final
rule,
we
have
revised
the
projected
PM2.5
nonattainment
counties
for
2010
and
2015
by
applying
CMAQ
for
the
entire
year
(
i.
e.,
January
through
December)
of
2001
using
2001
Base
Year
and
2010
and
2015
future
Base
Case
emissions
from
the
new
modeling
platform,
as
described
in
section
VI.
A.
2.
The
2010
and
2015
Base
Case
PM2.5
nonattainment
counties
were
determined
applying
the
updated
SMAT
method
using
current
1999­
2003
PM2.5
air
quality
coupled
with
the
PM2.5
species
from
the
2001
Base
Year
and
2010
and
2015
Base
Case
CMAQ
model
runs.
For
counties
with
multiple
monitoring
sites,
the
site
with
the
highest
future
concentration
was
selected
for
that
county.
Those
counties
with
future
year
design
values
of
15.05

g/
m3
or
higher
were
predicted
to
be
nonattainment.
The
result
is
that,
without
controls
beyond
those
included
in
the
Base
Case,
79
counties
in
the
East
are
projected
to
be
nonattainment
for
the
2010
Base
Case.
For
the
2015
Base
Case,
74
counties
in
the
East
are
projected
to
be
nonattainment
for
PM2.5.

In
light
of
the
uncertainties
inherent
in
regionwide
44
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
modeling
many
years
into
the
future,
of
the
79
nonattainment
counties
projected
for
the
2010
Base
Case,
we
have
the
most
confidence
in
our
projection
of
nonattainment
for
those
counties
that
are
not
only
forecast
to
be
nonattainment
in
2010,
based
on
the
SMAT
method,
but
that
also
measure
nonattainment
for
the
most
recent
period
of
available
ambient
data
(
i.
e.,
2001­
2003).
In
our
analysis
for
the
2010
Base
Case,
there
are
62
such
counties
in
the
East
that
are
both
"
modeled"
nonattainment
and
currently
have
"
monitored"
nonattainment.
We
refer
to
these
counties
as
having
"
modeled
plus
monitored"
nonattainment.
Out
of
an
abundance
of
caution,
we
are
using
only
these
62
"
modeled
plus
monitored"
counties
as
the
downwind
receptors
in
determining
which
upwind
States
make
a
significant
contribution
to
PM2.5
in
downwind
States.

The
79
counties
in
the
East
that
we
project
will
be
nonattainment
for
PM2.5
in
2010
and
the
subset
of
62
counties
that
are
also
"
monitored"
nonattainment
in
2001­

2003,
are
identified
in
Table
VI­
3.
The
2015
Base
Case
PM2.5
nonattainment
counties
are
provided
in
Table
VI­
4.

Table
VI­
3.
Projected
PM2.5
Concentrations
(

g/
m3)
for
Nonattainment
Counties
in
the
2010
Base
Case.

State
County
2010
Base
"
Modeled
+
Monitored"

Alabama
DeKalb
Co
15.23
No
45
State
County
2010
Base
"
Modeled
+
Monitored"

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Alabama
Jefferson
Co
18.57
Yes
Alabama
Montgomery
Co
15.12
No
Alabama
Morgan
Co
15.29
No
Alabama
Russell
Co
16.17
Yes
Alabama
Talladega
Co
15.34
No
Delaware
New
Castle
Co
16.56
Yes
District
of
Columbia
15.84
Yes
Georgia
Bibb
Co
16.27
Yes
Georgia
Clarke
Co
16.39
Yes
Georgia
Clayton
Co
17.39
Yes
Georgia
Cobb
Co
16.57
Yes
Georgia
DeKalb
Co
16.75
Yes
Georgia
Floyd
Co
16.87
Yes
Georgia
Fulton
Co
18.02
Yes
Georgia
Hall
Co
15.60
No
Georgia
Muscogee
Co
15.65
No
Georgia
Richmond
Co
15.68
No
Georgia
Walker
Co
15.43
Yes
Georgia
Washington
Co
15.31
No
Georgia
Wilkinson
Co
16.27
No
Illinois
Cook
Co
17.52
Yes
Illinois
Madison
Co
16.66
Yes
Illinois
St.
Clair
Co
16.24
Yes
Indiana
Clark
Co
16.51
Yes
Indiana
Dubois
Co
15.73
Yes
Indiana
Lake
Co
17.26
Yes
Indiana
Marion
Co
16.83
Yes
Indiana
Vanderburgh
Co
15.54
Yes
Kentucky
Boyd
Co
15.23
No
Kentucky
Bullitt
Co
15.10
No
Kentucky
Fayette
Co
15.95
Yes
Kentucky
Jefferson
Co
16.71
Yes
Kentucky
Kenton
Co
15.30
No
Maryland
Anne
Arundel
Co
15.26
Yes
Maryland
Baltimore
City
16.96
Yes
Michigan
Wayne
Co
19.41
Yes
Missouri
St.
Louis
City
15.10
No
New
Jersey
Union
Co
15.05
Yes
New
York
New
York
Co
16.19
Yes
North
Carolina
Catawba
Co
15.48
Yes
North
Carolina
Davidson
Co
15.76
Yes
North
Carolina
Mecklenburg
Co
15.22
No
46
State
County
2010
Base
"
Modeled
+
Monitored"

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Ohio
Butler
Co
16.45
Yes
Ohio
Cuyahoga
Co
18.84
Yes
Ohio
Franklin
Co
16.98
Yes
Ohio
Hamilton
Co
18.23
Yes
Ohio
Jefferson
Co
17.94
Yes
Ohio
Lawrence
Co
16.10
Yes
Ohio
Mahoning
Co
15.39
Yes
Ohio
Montgomery
Co
15.41
Yes
Ohio
Scioto
Co
18.13
Yes
Ohio
Stark
Co
17.14
Yes
Ohio
Summit
Co
16.47
Yes
Ohio
Trumbull
Co
15.28
No
Pennsylvania
Allegheny
Co
20.55
Yes
Pennsylvania
Beaver
Co
15.78
Yes
Pennsylvania
Berks
Co
15.89
Yes
Pennsylvania
Cambria
Co
15.14
Yes
Pennsylvania
Dauphin
Co
15.17
Yes
Pennsylvania
Delaware
Co
15.61
Yes
Pennsylvania
Lancaster
Co
16.55
Yes
Pennsylvania
Philadelphia
Co
16.65
Yes
Pennsylvania
Washington
Co
15.23
Yes
Pennsylvania
Westmoreland
Co
15.16
Yes
Pennsylvania
York
Co
16.49
Yes
Tennessee
Davidson
Co
15.36
No
Tennessee
Hamilton
Co
16.89
Yes
Tennessee
Knox
Co
17.44
Yes
Tennessee
Sullivan
Co
15.32
No
West
Virginia
Berkeley
Co
15.69
Yes
West
Virginia
Brooke
Co
16.63
Yes
West
Virginia
Cabell
Co
17.03
Yes
West
Virginia
Hancock
Co
17.06
Yes
West
Virginia
Kanawha
Co
17.56
Yes
West
Virginia
Marion
Co
15.32
Yes
West
Virginia
Marshall
Co
15.81
Yes
West
Virginia
Ohio
Co
15.14
Yes
West
Virginia
Wood
Co
16.66
Yes
Table
VI­
4.
Projected
PM2.5
Concentrations
(

g/
m3)
for
Nonattainment
Counties
in
the
2015
Base
Case.
47
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
State
County
2015
Base
Alabama
DeKalb
Co
15.24
Alabama
Jefferson
Co
18.85
Alabama
Montgomery
Co
15.24
Alabama
Morgan
Co
15.26
Alabama
Russell
Co
16.10
Alabama
Talladega
Co
15.22
Delaware
New
Castle
Co
16.47
District
of
Columbia
15.57
Georgia
Bibb
Co
16.41
Georgia
Chatham
Co
15.06
Georgia
Clarke
Co
16.15
Georgia
Clayton
Co
17.46
Georgia
Cobb
Co
16.51
Georgia
DeKalb
Co
16.82
Georgia
Floyd
Co
17.33
Georgia
Fulton
Co
18.00
Georgia
Hall
Co
15.36
Georgia
Muscogee
Co
15.58
Georgia
Richmond
Co
15.76
Georgia
Walker
Co
15.37
Georgia
Washington
Co
15.34
Georgia
Wilkinson
Co
16.54
Illinois
Cook
Co
17.71
Illinois
Madison
Co
16.90
Illinois
St.
Clair
Co
16.49
Illinois
Will
Co
15.12
Indiana
Clark
Co
16.37
Indiana
Dubois
Co
15.66
Indiana
Lake
Co
17.27
Indiana
Marion
Co
16.77
Indiana
Vanderburgh
Co
15.56
Kentucky
Boyd
Co
15.06
Kentucky
Fayette
Co
15.62
Kentucky
Jefferson
Co
16.61
Kentucky
Kenton
Co
15.09
Maryland
Baltimore
City
17.04
Maryland
Baltimore
Co
15.08
Michigan
Wayne
Co
19.28
Mississippi
Jones
Co
15.18
Missouri
St.
Louis
City
15.34
New
York
New
York
Co
15.76
North
Carolina
Catawba
Co
15.19
48
State
County
2015
Base
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
North
Carolina
Davidson
Co
15.34
Ohio
Butler
Co
16.32
Ohio
Cuyahoga
Co
18.60
Ohio
Franklin
Co
16.64
Ohio
Hamilton
Co
18.03
Ohio
Jefferson
Co
17.83
Ohio
Lawrence
Co
15.92
Ohio
Mahoning
Co
15.13
Ohio
Montgomery
Co
15.16
Ohio
Scioto
Co
17.92
Ohio
Stark
Co
16.86
Ohio
Summit
Co
16.14
Ohio
Trumbull
Co
15.05
Pennsylvania
Allegheny
Co
20.33
Pennsylvania
Beaver
Co
15.54
Pennsylvania
Berks
Co
15.66
Pennsylvania
Delaware
Co
15.52
Pennsylvania
Lancaster
Co
16.28
Pennsylvania
Philadelphia
Co
16.53
Pennsylvania
York
Co
16.22
Tennessee
Davidson
Co
15.36
Tennessee
Hamilton
Co
16.82
Tennessee
Knox
Co
17.34
Tennessee
Shelby
Co
15.17
Tennessee
Sullivan
Co
15.37
West
Virginia
Berkeley
Co
15.32
West
Virginia
Brooke
Co
16.51
West
Virginia
Cabell
Co
16.86
West
Virginia
Hancock
Co
16.97
West
Virginia
Kanawha
Co
17.17
West
Virginia
Marshall
Co
15.52
West
Virginia
Wood
Co
16.69
2.
Projection
of
Future
8­
Hour
Ozone
Nonattainment
a.
Methodology
for
Projecting
Future
8­
Hour
Ozone
Nonattainment
The
approach
for
projecting
future
8­
hour
ozone
concentrations
used
by
EPA
in
the
NPR
was
based
on
applying
49
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
the
model
in
a
relative
sense
to
estimate
the
change
in
ozone
between
the
base
year
(
2001)
and
each
future
scenario.

Projected
8­
hour
ozone
design
values
in
2010
and
2015
were
estimated
by
combining
the
relative
change
in
model
predicted
ozone
from
2001
to
the
future
scenario
with
an
estimate
of
the
base
year
ambient
8­
hour
ozone
design
value.

These
procedures
for
calculating
future
case
ozone
design
values
are
consistent
with
EPA's
draft
modeling
guidance
for
8­
hour
ozone
attainment
demonstrations.
The
draft
guidance
specifies
the
use
of
the
higher
of
the
design
values
from
(
a)
the
period
that
straddles
the
emissions
inventory
base
year
or
(
b)
the
design
value
period
which
was
used
to
designate
the
area
under
the
ozone
NAAQS.
At
the
time
of
the
proposal,
2000­
2002
was
the
design
value
period
which
both
straddled
the
2001
base
year
inventory
and
was
also
the
latest
period
available.

Comment:
Commenters
noted
that
the
procedures
used
by
EPA
for
projecting
future
8­
hour
ozone
concentrations
differ
from
the
procedures
used
for
projecting
PM2.5.
These
commenters
said
that
EPA
should
harmonize
the
two
approaches.

Response:
In
response
to
comments,
we
have
made
several
changes
in
the
approach
to
projecting
future
8­
hour
ozone
50
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
nonattainment
in
order
to
follow
an
approach
that
is
consistent
with
the
manner
in
which
PM2.5
projections
are
determined.
The
approach
we
are
using
to
project
PM2.5
for
the
final
rule
analysis
is
described
in
section
VI.
B.
1,

above.
In
order
to
harmonize
the
ozone
approach
with
the
approach
used
for
PM2.5,
we
are
using
the
weighted
average
of
the
design
values
for
the
periods
that
straddle
the
emission
base
year
(
i.
e.,
2001).
These
periods
are
1999­

2001,
2000­
2002,
and
2001­
2003.
In
this
approach,
the
fourth­
high
ozone
value
from
2001
is
weighted
3
times,
2000
and
2002
are
weighted
twice,
and
1999
and
2003
are
weighted
once.
This
has
the
desired
effect
of
weighting
the
projected
ozone
values
towards
the
middle
year
of
the
5­
year
period,
which
is
the
emissions
year
(
2001),
while
accounting
for
the
emissions
and
meteorological
variability
that
occurs
over
the
full
5­
year
period.
The
average
weighted
concentration
is
expected
to
be
more
representative
as
a
starting
point
for
future
year
projections
than
choosing
(
a)

the
single
design
value
period
that
straddles
the
base
year
or
(
b)
the
design
value
used
for
designations.
We
plan
to
incorporate
this
new
methodology
into
the
next
draft
version
of
our
ozone
modeling
guidance.

Comment:
One
commenter
claimed
that
the
2010
and
2015
ozone
51
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
projections
in
the
proposal
base
cases
were
too
optimistic,

that
is,
that
the
modeling
was
underestimating
the
number
of
areas
that
may
be
in
nonattainment
in
the
future.
The
commenter
urged
a
more
conservative
approach
to
assessing
the
future
attainment
status
of
areas.

Response:
The
technical
basis
for
the
comment
stemmed
from
the
assertion
that
the
regional
ozone
modeling
that
EPA
performed
for
the
proposal
was
not
of
"
SIP­
quality."
The
EPA
response
to
the
specific
technical
issues
with
regard
to
episode
selection
and
grid
resolution
can
be
found
in
section
VI.
A
as
well
as
in
the
response
to
comments
document.
The
EPA
remains
confident
that
the
CAIR
8­
hour
ozone
modeling
platform
is
appropriate
for
assessing
potential
levels
of
future
nonattainment.

b.
Projected
2010
and
2015
Base
Case
8­
Hour
Ozone
Nonattainment
Counties
For
the
final
rule,
we
have
revised
our
projections
of
ozone
nonattainment
for
the
2010
and
2015
Base
Cases
by
applying
CAMx
for
the
three
1995
ozone
episodes
using
2001
Base
Year
and
2010
and
2015
future
Base
Case
emissions
from
the
new
modeling
platform,
as
described
in
section
VI.
A.
2.

The
revised
2010
and
2015
Base
Case
8­
hour
ozone
nonattainment
counties
were
determined
by
applying
the
52
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
relative
change
in
8­
hour
ozone
predicted
by
these
CAMx
model
runs
to
the
weighted
average
1999­
2003
8­
hour
ozone
concentrations
as
described
above
and,
in
more
detail,
in
the
NFR
AQMTSD.
For
counties
with
multiple
monitoring
sites,
the
site
with
the
highest
future
concentration
was
selected
for
that
county.
Those
counties
with
future
year
design
values
of
85
parts
per
billion
(
ppb)
or
higher
were
predicted
to
be
nonattainment.

As
a
result
of
our
updated
modeling
we
project
that,

without
controls
beyond
those
in
the
Base
Case,
there
will
be
40
8­
hour
ozone
nonattainnment
counties
in
2010
and
22
nonattainment
counties
in
2015.
All
of
the
40
counties
that
we
are
projecting
to
be
nonattainment
for
the
2010
Base
Case
are
also
measuring
nonattainment
based
on
the
most
recent
design
value
period
(
i.
e.,
2001­
2003).
We
refer
to
these
counties
as
"
modeled
plus
monitored"
nonattainment,
as
described
above
in
section
IV.
B.
1
for
PM2.5.
We
are
using
these
40
counties
as
the
downwind
receptors
to
determine
which
States
make
a
significant
contribution
to
8­
hour
ozone
nonattainment
in
downwind
States.

The
counties
we
are
projecting
to
be
nonattainment
for
8­
hour
ozone
in
the
2010
Base
Case
and
2015
Base
Case
are
listed
in
Table
VI­
5
and
Table
VI­
6,
respectively.

Table
VI­
5.
Projected
2010
Base
Case
8­
hour
Ozone
53
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Nonattainment
Counties
and
Concentrations
(
ppb).

State
County
2010
Base
Connecticut
Fairfield
Co
92.6
Connecticut
Middlesex
Co
90.9
Connecticut
New
Haven
Co
91.6
Delaware
New
Castle
Co
85.2
District
of
Columbia
85.0
Georgia
Fulton
Co
86.5
Maryland
Anne
Arundel
Co
88.8
Maryland
Cecil
Co
89.7
Maryland
Harford
Co
93.0
Maryland
Kent
Co
86.2
Michigan
Macomb
Co
85.5
New
Jersey
Bergen
Co
86.9
New
Jersey
Camden
Co
91.9
New
Jersey
Gloucester
Co
91.8
New
Jersey
Hunterdon
Co
89.0
New
Jersey
Mercer
Co
95.6
New
Jersey
Middlesex
Co
92.4
New
Jersey
Monmouth
Co
86.6
New
Jersey
Morris
Co
86.5
New
Jersey
Ocean
Co
100.5
New
York
Erie
Co
87.3
New
York
Richmond
Co
87.3
New
York
Suffolk
Co
91.1
New
York
Westchester
Co
85.3
Ohio
Geauga
Co
87.1
Pennsylvania
Bucks
Co
94.7
Pennsylvania
Chester
Co
85.7
Pennsylvania
Montgomery
Co
88.0
Pennsylvania
Philadelphia
Co
90.3
Rhode
Island
Kent
Co
86.4
Texas
Denton
Co
87.4
Texas
Galveston
Co
85.1
Texas
Harris
Co
97.9
Texas
Jefferson
Co
85.6
Texas
Tarrant
Co
87.8
Virginia
Arlington
Co
86.2
Virginia
Fairfax
Co
85.7
Wisconsin
Kenosha
Co
91.3
Wisconsin
Ozaukee
Co
86.2
Wisconsin
Sheboygan
Co
88.3
54
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Table
VI­
6.
Projected
2015
Base
Case
8­
hour
Ozone
Nonattainment
Counties
and
Concentrations
(
ppb).

State
County
2015
Base
Connecticut
Fairfield
Co
91.4
Connecticut
Middlesex
Co
89.1
Connecticut
New
Haven
Co
89.8
Maryland
Anne
Arundel
Co
86.0
Maryland
Cecil
Co
86.9
Maryland
Harford
Co
90.6
Michigan
Macomb
Co
85.1
New
Jersey
Bergen
Co
85.7
New
Jersey
Camden
Co
89.5
New
Jersey
Gloucester
Co
89.6
New
Jersey
Hunterdon
Co
86.5
New
Jersey
Mercer
Co
93.5
New
Jersey
Middlesex
Co
89.8
New
Jersey
Ocean
Co
98.0
New
York
Erie
Co
85.2
New
York
Suffolk
Co
89.9
Pennsylvania
Bucks
Co
93.0
Pennsylvania
Montgomery
Co
86.5
Pennsylvania
Philadelphia
Co
88.9
Texas
Harris
Co
97.3
Texas
Jefferson
Co
85.0
Wisconsin
Kenosha
Co
89.4
C.
How
did
EPA
Assess
Interstate
Contributions
to
Nonattainment?

1.
PM2.5
Contribution
Modeling
Approach
For
the
proposed
rule,
EPA
performed
State­
by­
State
zero­
out
modeling
to
quantify
the
contribution
from
emissions
in
each
State
to
future
PM2.5
nonattainment
in
other
States
and
to
determine
whether
that
contribution
55
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
meets
the
air
quality
prong
(
i.
e.,
before
considering
cost)

of
the
"
contribute
significantly"
test.
The
zero­
out
modeling
technique
provides
an
estimate
of
downwind
impacts
by
comparing
the
model
predictions
from
the
2010
Base
Case
to
the
predictions
from
a
run
in
which
all
anthropogenic
SO2
and
NOx
emissions
are
removed
from
specific
States.

Counties
forecast
to
be
nonattainment
for
PM2.5
in
the
proposal
2010
Base
Case
were
used
as
receptors
for
quantifying
interstate
contributions
of
PM2.5.
For
each
State­
by­
State
zero­
out
run
we
projected
the
annual
average
PM2.5
concentration
at
each
receptor
using
the
proposed
SMAT
technique,
as
described
in
the
NPR
AQMTSD.
The
contribution
from
an
upwind
State
to
nonattainment
at
a
given
downwind
receptor
was
determined
by
calculating
the
difference
in
PM2.5
concentration
between
the
2010
Base
Case
and
the
zeroout
run
at
that
receptor.
We
followed
this
process
for
each
State­
by­
State
zero­
out
run
and
each
receptor.
For
each
upwind
State,
we
identified
the
largest
contribution
from
that
State
to
a
downwind
nonattainment
receptor
in
order
to
determine
the
magnitude
of
the
maximum
downwind
contribution
from
each
State.
The
maximum
downwind
contribution
was
proposed
as
the
metric
for
determining
whether
or
not
the
contribution
was
significant.
As
described
in
section
III,

EPA
proposed,
in
the
alternative,
a
criterion
of
0.10

g/
m3
56
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
and
0.15

g/
m3
for
determining
whether
emissions
in
a
State
make
a
significant
contribution
(
before
considering
cost)
to
PM2.5
nonattainment
in
another
State.
Details
on
these
procedures
can
be
found
in
the
NPR
AQMTSD.

Comments:
Commenters
questioned
the
use
of
zero­
out
modeling
and
said
that
EPA
should
support
the
development
of
a
source
apportionment
model
for
PM2.5
contributions.
The
commenter
recommended
that
EPA
delay
the
final
rule
until
such
a
technique
can
be
used.
Another
commenter
provided
results
of
a
sulfate
source
apportionment
technique
currently
under
development
along
with
modeling
results
which
showed
that
the
zero­
out
technique
and
source
apportionment
for
sulfate
provide
similar
results
in
terms
of
the
magnitude
and
extent
of
downwind
impacts.
The
commenter
noted
that
the
results
suggest
that
zero­
out
modeling
may
somewhat
underestimate
the
transport
of
sulfate.

Response:
The
EPA
continues
to
believe
that
the
zero­
out
technique
is
a
credible
method
for
quantifying
interstate
PM2.5
contributions.
This
is
supported
by
a
commenter's
results
showing
that
the
zero­
out
technique
and
source
apportionment
appear
to
give
similar
results.
We
accept
the
commenter's
modeling
for
sulfate
source
apportionment
results
which
indicate
that
the
zero­
out
technique
does
not
57
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
overestimate
interstate
transport.
Moreover,
EPA
rejects
the
notion
that
we
should
delay
needed
reductions
while
we
await
alternative
assessment
techniques.

2.
8­
Hour
Ozone
Contribution
Modeling
Approach
In
the
proposal,
EPA
quantified
the
impact
of
emissions
from
specific
upwind
States
on
8­
hour
ozone
concentrations
in
projected
downwind
nonattainment
areas.
The
procedures
we
followed
to
assess
interstate
ozone
contribution
for
the
proposal
analysis
are
summarized
below.
We
are
using
these
same
procedures
along
with
the
updated
CAMx
modeling
platform,
as
described
in
section
VI.
A.,
to
assess
ozone
contributions
for
today's
rule.
Details
on
these
procedures
can
be
found
in
the
NFR
AQMTSD.

We
applied
two
different
modeling
techniques,
zero­
out
and
source
apportionment,
to
assess
the
contributions
of
emissions
in
upwind
States
on
8­
hour
ozone
nonattainment
in
downwind
States.
The
outputs
of
the
two
modeling
techniques
were
evaluated
in
terms
of
three
key
contribution
factors
to
determine
which
States
make
a
significant
contribution
to
downwind
ozone
nonattainment
as
described
in
section
VI.
B.
2.

The
zero­
out
and
source
apportionment
modeling
techniques
provide
different,
but
equally
valid,
technical
approaches
to
quantifying
the
downwind
impact
of
emissions
from
upwind
States.
The
zero­
out
modeling
analysis
provides
an
estimate
58
13
The
six
States
are
Kansas,
Nebraska,
North
Dakota,
Oklahoma,
South
Dakota,
and
Texas.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
of
downwind
impacts
by
comparing
the
model
predictions
from
the
2010
Base
Case
and
the
predictions
from
a
model
run
in
which
all
anthropogenic
NOx
and
VOC
emissions
are
removed
from
specific
States.
The
source
apportionment
modeling
quantifies
downwind
impacts
by
tracking
and
allocating
the
amounts
of
ozone
formed
from
man­
made
NOx
and
VOC
emissions
in
upwind
States.
Because
large
portions
of
the
six
States
along
the
western
border
of
the
modeling
domain13
are
outside
the
area
covered
by
our
modeling,
EPA
did
not
analyze
the
contributions
to
downwind
ozone
nonattainment
for
these
States.

In
the
analysis
done
at
proposal,
EPA
considered
three
fundamental
factors
for
evaluating
whether
emissions
in
an
upwind
State
make
large
and/
or
frequent
contributions
to
downwind
nonattainment:
(
1)
the
magnitude
of
the
contribution;
(
2)
the
frequency
of
the
contribution;
and
(
3)

the
relative
amount
of
the
contribution
when
compared
against
contributions
from
other
areas.
The
factors
are
the
basis
for
several
metrics
that
can
be
used
to
assess
a
particular
impact.
The
metrics
used
in
this
analysis
were
the
same
as
those
used
in
the
NOx
SIP
Call.

Within
these
three
factors,
eight
specific
metrics
were
59
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
calculated
to
assess
the
contribution
of
each
of
the
31
States
to
the
residual
nonattainment
counties.
For
the
zero­
out
modeling,
EPA
considered:
(
1)
the
maximum
contribution
(
magnitude);
(
2)
the
number
and
percentage
of
exceedances
with
contributions
in
certain
concentration
ranges
(
frequency);
(
3)
the
total
contribution
relative
to
the
total
exceedance
level
ozone
in
the
receptor
area
(
relative
amount);
and
(
4)
the
population­
weighted
total
contribution
relative
to
the
total
population­
weighted
exceedance
level
ozone
in
the
receptor
area
(
relative
amount).
For
the
source
apportionment
modeling
EPA
considered:
(
5)
the
maximum
contribution
(
magnitude);
(
6)

the
highest
daily
average
contribution
(
magnitude);
(
7)
the
number
and
percentages
of
exceedances
with
contributions
in
certain
concentration
ranges
(
frequency);
and
(
8)
the
total
average
contribution
to
exceedance
ozone
in
the
downwind
area
(
relative
amount).
The
values
for
these
metrics
were
calculated
using
only
those
periods
during
which
the
model
predicted
8­
hour
average
ozone
concentrations
greater
than
or
equal
to
85
ppb
in
at
least
one
of
the
model
grid
cells
associated
with
the
receptor
county
in
the
2010
base
case.

Grid
cells
were
linked
to
a
specific
nonattainment
county
if
any
part
of
the
grid
cell
covered
any
portion
of
the
projected
2010
nonattainment
county.
60
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
The
first
step
in
evaluating
the
contribution
factors
was
to
screen
out
linkages
for
which
the
contributions
were
clearly
small.
This
initial
screening
was
based
on
two
criteria:
(
1)
the
maximum
contribution
had
to
be
greater
than
or
equal
to
2
ppb
from
either
of
the
two
modeling
techniques;
and
(
2)
the
total
average
contribution
to
exceedance
of
ozone
in
the
downwind
area
had
to
be
greater
than
1
percent.
If
either
screening
test
was
not
met,
then
the
linkage
was
not
considered
significant.
Those
linkages
that
had
contributions
which
exceeded
the
screening
criteria
were
evaluated
further
in
steps
2
through
4.

In
step
2,
we
evaluated
the
contributions
in
each
linkage
based
on
the
zero­
out
modeling
and
in
step
3
we
evaluated
the
contributions
in
each
linkage
based
on
the
source
apportionment
modeling.
In
step
4,
we
considered
the
results
of
both
step
2
and
step
3
to
determine
which
of
the
linkages
were
significant.
For
both
techniques,
EPA
determined
whether
the
linkage
is
significant
by
evaluating
the
magnitude,
frequency,
and
relative
amount
of
the
contributions.
Each
upwind
State
that
made
relatively
large
and/
or
frequent
contributions
to
nonattainment
in
the
downwind
area,
based
on
these
factors,
was
considered
to
contribute
significantly
to
nonattainment
in
the
downwind
area.
61
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
The
EPA
believes
that
each
of
the
factors
provides
an
independent
measure
of
contribution,
however,
there
had
to
be
at
least
two
different
factors
that
indicated
large
and/
or
frequent
contributions
in
order
for
the
linkage
to
be
found
significant.
In
this
regard,
the
finding
of
a
significant
contribution
for
an
individual
linkage
was
not
based
on
any
single
factor.
Further,
each
of
the
modeling
approaches
had
to
show
at
least
one
indicator
of
a
large
and/
or
frequent
contribution
in
order
for
the
linkage
to
be
found
significant.
The
EPA
received
several
general
comments
on
the
procedures
for
assessing
interstate
contributions
of
ozone
to
projected
residual
nonattainment
areas,
as
discussed
below.

Comment:
A
commenter
opposed
the
use
of
population­
weighted
metrics
to
determine
whether
an
upwind
State's
impact
on
a
location
in
another
State
is
significant.

Response:
The
commenter's
concern
was
that
transport
contributions
to
rural
areas
with
low
populations
were
not
being
weighted
appropriately.
This
is
not
a
valid
concern
because
the
relative
contribution
factor
from
the
zero­
out
modeling
is
presumed
to
be
met
if
either
of
the
two
criteria
(
population­
weighted,
or
non­
population­
weighted)
show
large
contributions.

Comment:
Also,
EPA
received
a
specific
comment
on
a
certain
62
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
linkage
that
was
deemed
to
be
significant
in
the
analysis
done
to
support
the
NPR.
The
commenter
objected
to
the
conclusion
that
Mississippi
significantly
contributes
to
residual
ozone
exceedances
near
Memphis.
The
objection
resulted
from
issues
with
grid
resolution,
episode
selection,
and
the
fact
that
the
zero­
out
and
source
apportionment
modeling
for
Mississippi
included
some
emissions
from
Tennessee
and
Arkansas
due
to
the
irregular
State
boundaries.

Response:
As
noted
in
section
VI.
B.
2,
Crittenden
County,
AR
is
no
longer
projected
to
be
a
nonattainment
area
in
the
2010
Base
Case.
As
a
result,
the
issue
of
Mississippi's
contribution
to
ozone
in
the
Memphis
area
is
moot.

D.
What
Are
the
Estimated
Interstate
Contributions
to
PM2.5
and
8­
Hour
Ozone
Nonattainment?

1.
Results
of
PM2.5
Contribution
Modeling
In
this
section,
we
present
the
interstate
contributions
from
emissions
in
upwind
States
to
PM2.5
nonattainment
in
downwind
nonattainment
counties.
States
which
contribute
0.2

g/
m3
or
more
to
PM2.5
nonattainment
in
another
State
are
determined
to
contribute
significantly
(
before
considering
cost).
We
calculated
the
interstate
PM2.5
contributions
using
the
State­
by­
State
zero­
out
63
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
modeling
technique,
as
indicated
above
in
section
VI.
C.
1.

This
technique
is
described
in
the
NFR
AQMTSD.
We
performed
zero­
out
modeling
using
CMAQ
for
each
of
37
States
individually
(
i.
e.,
Alabama,
Arkansas,
Connecticut,

Delaware,
Florida,
Georgia,
Illinois,
Indiana,
Iowa,
Kansas,

Kentucky,
Louisiana,
Maine,
Maryland
combined
with
the
District
of
Columbia,
Massachusetts,
Michigan,
Minnesota,

Mississippi,
Missouri,
Nebraska,
New
Hampshire,
New
Jersey,

New
York,
North
Carolina,
North
Dakota,
Ohio,
Oklahoma,

Pennsylvania,
Rhode
Island,
South
Carolina,
South
Dakota,

Tennessee,
Texas,
Vermont,
Virginia,
West
Virginia,
and
Wisconsin).

We
calculated
each
State's
contribution
to
PM2.5
in
each
of
the
62
counties
that
are
projected
to
be
nonattainment
in
the
2010
Base
Case
(
i.
e.,
"
modeled"

nonattainment)
and
are
also
"
monitored"
nonattainment
in
2001­
2003,
as
described
in
section
VI.
B.
1.
b.
The
maximum
contribution
from
each
upwind
State
to
downwind
PM2.5
nonattainment
is
provided
in
Table
VI­
7.
The
contributions
from
each
State
to
nonattainment
in
each
nonattainment
county
are
provided
in
the
NFR
AQMTSD.
Based
on
the
Stateby
State
modeling,
there
are
23
States
and
the
District
of
64
14
As
noted
above,
we
combined
Maryland
and
the
District
of
Columbia
as
a
single
entity
in
our
contribution
modeling.
This
is
a
logical
approach
because
of
the
small
size
of
the
District
of
Columbia
and,
hence,
it's
emissions,
and
it's
close
proximity
to
Maryland
(
see
the
proposal
eleswhere
in
today's
Federal
Register
concerning
control
of
emissions
of
PM2.5
precursors
from
New
Jersey
and
from
Delaware).
Under
our
analysis,
Maryland
and
the
District
of
Columbia
are
linked
as
significant
contributors
to
the
same
downwind
nonattainment
counties.
EPA
received
no
adverse
comment
on
this
approach.
We
also
considered
these
entities
separately,
and
in
view
of
the
close
proximity
of
these
two
areas
we
believe
that
Maryland
is
linked
as
a
significant
contributor
to
nonattainment
in
the
District
of
Columbia
and
that
the
District
of
Columbia
is
linked
as
a
significant
contributor
to
nonattainment
in
Maryland.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Columbia14
which
contribute
0.2

g/
m3
or
more
to
downwind
PM2.5
nonattainment
(
Alabama,
the
District
of
Columbia,

Florida,
Georgia,
Illinois,
Indiana,
Iowa,
Kentucky,

Louisiana,
Maryland,
Michigan,
Minnesota,
Mississippi,

Missouri,
New
York,
North
Carolina,
Ohio,
Pennsylvania,

South
Carolina,
Tennessee,
Texas,
Virginia,
West
Virginia,

and
Wisconsin).
In
Table
VI­
8,
we
provide
a
list
of
the
downwind
nonattainment
counties
to
which
each
upwind
State
contributes
0.2

g/
m3
or
more
(
i.
e.,
the
upwind
State­

todownwind
nonattainment
"
linkages").

Table
VI­
7.
Maximum
Downwind
PM2.5
Contribution
(

g/
m3)
for
each
of
37
States.

Upwind
State
Maximum
Downwind
Contribution
Upwind
State
Maximum
Downwind
Contribution
Alabama
0.98
Nebraska
0.07
Arkansas
0.19
New
Hampshire
<
0.05
65
Upwind
State
Maximum
Downwind
Contribution
Upwind
State
Maximum
Downwind
Contribution
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Connecticut
<
0.05
New
Jersey
0.13
Delaware
0.14
New
York
0.34
Florida
0.45
North
Carolina
0.31
Georgia
1.27
North
Dakota
0.11
Illinois
1.02
Ohio
1.67
Indiana
0.91
Oklahoma
0.12
Iowa
0.28
Pennsylvania
0.89
Kansas
0.11
Rhode
Island
<
0.05
Kentucky
0.90
South
Carolina
0.40
Louisiana
0.25
South
Dakota
<
0.05
Maine
<
0.05
Tennessee
0.65
Maryland/
DC
0.69
Texas
0.29
Massachusetts
0.07
Vermont
<
0.05
Michigan
0.62
Virginia
0.44
Minnesota
0.21
West
Virginia
0.84
Mississippi
0.23
Wisconsin
0.56
Missouri
1.07
Table
VI­
8.
Upwind
State­
to­
Downwind
Nonattainment
County
Significant
"
Linkages"
for
PM2.5.

Upwind
States
Total
Linkage
s
Downwind
Counties
AL
21
Bibb
GA
Cabell
WV
Catawba
NC
Clark
IN
Clarke
GA
Clayton
GA
Cobb
GA
Davidson
NC
DeKalb
GA
Dubois
IN
Fayette
KY
Floyd
GA
Fulton
GA
Hamilton
OH
Hamilton
TN
Jefferson
KY
Knox
TN
Lawrence
OH
Scioto
OH
Vanderburgh
IN
Walker
GA
FL
7
Bibb
GA
Clarke
GA
Clayton
GA
Cobb
GA
DeKalb
GA
Jefferson
AL
Russell
AL
GA
17
Butler
OH
Cabell
WV
Catawba
NC
Clark
IN
Davidson
NC
Fayette
KY
Hamilton
OH
Hamilton
TN
Jefferson
AL
Jefferson
KY
Kanawha
WV
Knox
TN
Lawrence
OH
Montgomery
OH
Russell
AL
Scioto
OH
Vanderburgh
IN
IL
23
Allegheny
PA
Butler
OH
Cabell
WV
Clark
IN
66
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Cuyahoga
OH
Dubois
IN
Fayette
KY
Franklin
OH
Hamilton
OH
Hamilton
TN
Jefferson
AL
Jefferson
KY
Kanawha
WV
Lake
IN
Lawrence
OH
Mahoning
OH
Marion
IN
Montgomery
OH
Scioto
OH
Stark
OH
Summit
OH
Vanderburgh
IN
Wayne
MI
IN
46
Allegheny
PA
Beaver
PA
Berkeley
WV
Bibb
GA
Brooke
WV
Butler
OH
Cabell
WV
Cambria
PA
Catawba
NC
Clarke
GA
Clayton
GA
Cobb
GA
Cook
IL
Cuyahoga
OH
Davidson
NC
DeKalb
GA
Fayette
KY
Floyd
GA
Franklin
OH
Fulton
GA
Hamilton
OH
Hamilton
TN
Hancock
WV
Jefferson
AL
Jefferson
KY
Jefferson
OH
Kanawha
WV
Knox
TN
Lancaster
PA
Lawrence
OH
Madison
IL
Mahoning
OH
Marion
WV
Marshall
WV
Montgomery
OH
Ohio
WV
Russell
AL
St.
Clair
IL
Scioto
OH
Stark
OH
Summit
OH
Walker
GA
Wayne
MI
Washington
PA
Westmoreland
PA
Wood
WV
IA
5
Cook
IL
Lake
IN
Madison
IL
Marion
IN
St.
Clair
IL
KY
35
Allegheny
PA
Butler
OH
Cabell
WV
Catawba
NC
Clark
IN
Clarke
GA
Cobb
GA
Cuyahoga
OH
Davidson
NC
Dubois
IN
Floyd
GA
Franklin
OH
Hamilton
OH
Hamilton
TN
Jefferson
AL
Jefferson
OH
Kanawha
WV
Knox
TN
Lawrence
OH
Madison
IL
Mahoning
OH
Marion
IN
Marion
WV
Marshall
WV
Montgomery
OH
Ohio
WV
St.
Clair
IL
Scioto
OH
Stark
OH
Summit
OH
Vanderburgh
IN
Walker
GA
Washington
PA
Westmoreland
PA
Wood
WV
LA
2
Jefferson
AL
Russell
AL
MD/
DC
13
Berkeley
WV
Berks
PA
Cambria
PA
Dauphin
PA
Delaware
PA
District
of
Columbia
Lancaster
PA
New
Castle
DE
New
York
NY
Philadelphia
PA
Union
NJ
Westmoreland
PA
York
PA
MI
36
Allegheny
PA
Beaver
PA
Berks
PA
Brooke
WV
Butler
OH
Cabell
WV
Cambria
PA
Clark
IN
Cook
IL
Cuyahoga
OH
Dauphin
PA
Delaware
PA
Fayette
KY
Franklin
OH
Hamilton
OH
Hancock
WV
Jefferson
OH
Lake
IN
Lancaster
PA
Lawrence
OH
Mahoning
OH
Marion
IN
Marion
WV
Marshall
WV
Montgomery
OH
New
Castle
DE
Ohio
WV
Philadelphia
PA
Scioto
OH
Stark
OH
Summit
OH
Union
NJ
67
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Washington
PA
Westmoreland
PA
Wood
WV
York
PA
MN
2
Cook
IL
Lake
IN
MO
9
Clark
IN
Cook
IL
Dubois
IN
Jefferson
KY
Lake
IN
Madison
IL
Marion
IN
St.
Clair
IL
Vanderburgh
IN
MS
1
Jefferson
AL
NY
5
Berks
PA
Lancaster
PA
New
Castle
DE
New
Haven
CT
Union
NJ
NC
7
Anne
Arundel
MD
Baltimore
City
Bibb
GA
Clarke
GA
District
of
Columbia
Kanawha
WV
Knox
TN
OH
51
Anne
Arundel
MD
Allegheny
PA
Baltimore
City
Beaver
PA
Berkeley
WV
Berks
PA
Bibb
GA
Brooke
WV
Cabell
WV
Cambria
PA
Catawba
NC
Clark
IN
Clarke
GA
Clayton
GA
Cobb
GA
Cook
IL
Dauphin
PA
Davidson
NC
DeKalb
GA
Delaware
PA
District
of
Columbia
Dubois
IN
Fayette
KY
Floyd
GA
Fulton
GA
Hamilton
TN
Hancock
WV
Jefferson
AL
Jefferson
KY
Kanawha
WV
Knox
TN
Lake
IN
Lancaster
PA
Madison
IL
Marion
IN
Marion
WV
Marshall
WV
New
Castle
DE
New
York
NY
Ohio
WV
Philadelphia
PA
Russell
AL
St.
Clair
IL
Union
NJ
Vanderburgh
IN
Walker
GA
Washington
PA
Wayne
MI
Westmoreland
PA
Wood
WV
York
PA
PA
25
Anne
Arundel
MD
Baltimore
City
Berkeley
WV
Brooke
WV
Cabell
WV
Catawba
NC
Clarke
GA
Cuyahoga
OH
Davidson
NC
District
of
Columbia
Hancock
WV
Jefferson
OH
Kanawha
WV
Lawrence
OH
Mahoning
OH
Marion
WV
Marshall
WV
New
Castle
DE
New
York
NY
Ohio
WV
Stark
OH
Summit
OH
Union
NJ
Wayne
MI
Wood
WV
SC
9
Bibb
GA
Catawba
NC
Clarke
GA
Clayton
GA
Cobb
GA
Davidson
NC
DeKalb
GA
Fulton
GA
Russell
AL
TN
23
Bibb
GA
Butler
OH
Cabell
WV
Catawba
NC
Clark
IN
Clarke
GA
Clayton
GA
Cobb
GA
Davidson
NC
DeKalb
GA
Dubois
IN
Fayette
KY
Floyd
GA
Fulton
GA
Hamilton
OH
Jefferson
AL
Jefferson
KY
Kanawha
WV
Lawrence
OH
Russell
AL
Scioto
OH
Vanderburgh
TN
Walker
GA
68
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
TX
2
Madison
IL
St
Clair
IL
VA
13
Anne
Arundel
MD
Baltimore
City
Berkeley
WV
Berks
PA
Catawba
NC
Dauphin
PA
Davidson
NC
Delaware
PA
District
of
Lancaster
PA
New
Castle
DE
Philadelphia
PA
York
PA
WV
33
Anne
Arundel
MD
Allegheny
PA
Baltimore
City
Beaver
PA
Berks
PA
Butler
OH
Cambria
PA
Catawba
NC
Clarke
GA
Cuyahoga
OH
Dauphin
PA
Davidson
NC
Delaware
PA
District
of
Columbia
Fayette
KY
Franklin
OH
Hamilton
OH
Jefferson
OH
Knox
TN
Lancaster
PA
Lawrence
OH
Mahoning
OH
Montgomery
OH
New
Castle
DE
New
York
NY
Philadelphia
PA
Scioto
OH
Stark
OH
Summit
OH
Union
NJ
Washington
PA
Westmoreland
PA
York
PA
WI
4
Cook
IL
Lake
IN
Marion
IN
Wayne
MI
2.
Results
of
8­
Hour
Ozone
Contribution
Modeling
In
this
section,
we
present
the
results
of
air
quality
modeling
to
determine
which
upwind
States
contribute
significantly
(
before
considering
cost)
to
8­
hour
ozone
nonattainment
in
downwind
States.
The
analytical
procedures
to
determine
which
States
make
a
significant
contribution
are
based
on
the
zero­
out
and
source
apportionment
modeling
techniques
using
CAMx,
as
described
in
section
VI.
C.
2
and
in
the
NFR
AQMTSD.

We
performed
ozone
contribution
modeling
using
both
of
these
techniques
for
31
States
in
the
East
and
the
District
of
Columbia
(
i.
e.,
Alabama,
Arkansas,
Connecticut,
Delaware,
Georgia,

Florida,
Iowa,
Illinois,
Indiana,
Kentucky,
Louisiana,

Massachusetts,
Maine,
Maryland
combined
with
the
District
of
69
15
As
noted
above,
we
combined
Maryland
and
the
District
of
Columbia
as
a
single
entity
in
our
contribution
modeling.
This
is
a
logical
approach
because
of
the
small
size
of
the
District
of
Columbia
and,
hence,
it's
emissions,
and
it's
close
proximity
to
Maryland
(
see
the
proposal
eleswhere
in
today's
Federal
Register
concerning
control
of
emissions
of
PM2.5
precursors
from
New
Jersey
and
from
Delaware).
Under
our
analysis,
Maryland
and
the
District
of
Columbia
are
linked
as
significant
contributors
to
the
same
downwind
nonattainment
counties.
EPA
received
no
adverse
comment
on
this
approach.
We
also
considered
these
entities
separately,
and
in
view
of
the
close
proximity
of
these
two
areas
we
believe
that
Maryland
is
linked
as
a
significant
contributor
to
nonattainment
in
the
District
of
Columbia
and
that
the
District
of
Columbia
is
linked
as
a
significant
contributor
to
nonattainment
in
Maryland.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Columbia,
Michigan,
Minnesota,
Mississippi,
Missouri,
New
Hampshire,
New
Jersey,
New
York,
North
Carolina,
Ohio,

Pennsylvania,
Rhode
Island,
South
Carolina,
Tennessee,
Vermont,

Virginia,
West
Virginia,
and
Wisconsin).

We
evaluated
the
interstate
ozone
contributions
from
each
of
the
31
upwind
States
and
the
District
of
Columbia
to
each
of
the
40
counties
that
are
projected
to
be
nonattainment
in
the
2010
Base
Case
(
i.
e.,
"
modeled"
nonattainment)
and
are
also
"
monitored"
nonattainment
in
2001­
2003,
as
described
in
section
VI.
B.
2.
b.
We
analyzed
the
contributions
from
upwind
States
to
these
counties
in
terms
of
various
metrics,
described
above
and
in
more
detail
in
the
NFR
AQMTSD.

Based
on
the
State­
by­
State
modeling,
there
are
25
States
and
the
District
of
Columbia15
which
make
a
significant
contribution
(
before
considering
cost)
to
8­
hour
ozone
70
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
nonattainment
in
downwind
States
(
i.
e.,
Alabama,
Arkansas,

Connecticut,
Delaware,
the
District
of
Columbia,
Florida,
Iowa,

Illinois,
Indiana,
Kentucky,
Louisiana,
Massachusetts,
Maryland,

Michigan,
Mississippi,
Missouri,
New
Jersey,
New
York,
North
Carolina,
Ohio,
Pennsylvania,
South
Carolina,
Tennessee,

Virginia,
West
Virginia,
and
Wisconsin).
In
Table
VI­
9,
we
provide
a
list
of
the
downwind
nonattainment
counties
to
which
each
upwind
State
makes
a
significant
contribution
(
i.
e.,
the
upwind
State­
to­
downwind
nonattainment
"
linkages").

Table
VI­
9.
Upwind
State­
to­
Downwind
Nonattainment
County
Significant
"
Linkages"
for
8­
hour
Ozone.

Upwind
States
Total
Linkages
Downwind
Counties
AL
3
Fulton
GA
Harris
TX
Jefferson
TX
AR
3
Galveston
TX
Harris
TX
Jefferson
TX
CT
2
Kent
RI
Suffolk
NY
DE
13
Bucks
PA
Camden
NJ
Chester
PA
Gloucester
NJ
Hunterdon
NJ
Mercer
NJ
Middlesex
NJ
Monmouth
NJ
Montgomery
PA
Morris
NJ
Ocean
NJ
Philadelphia
PA
Suffolk
NY
FL
1
Fulton
GA
IA
3
Kenosha
WI
Macomb
MI
Sheboygan
WI
IL
5
Geauga
OH
Kenosha
WI
Macomb
MI
Ozaukee
WI
Sheboygan
WI
IN
5
Geauga
OH
Kenosha
WI
Macomb
MI
Ozaukee
WI
Sheboygan
WI
KY
3
Fulton
GA
Geauga
OH
Macomb
MI
LA
3
Galveston
TX
Harris
TX
Jefferson
TX
MA
2
Kent
RI
Middlesex
NJ
MD/
DC
23
Arlington
VA
Bergen
NJ
Bucks
PA
Camden
NJ
Chester
PA
District
of
Columbia
Erie
NY
Fairfax
VA
Fairfield
CT
Gloucester
NJ
Hunterton
NJ
Mercer
NJ
Middlesex
NJ
Monmouth
NJ
Montgomery
PA
Morris
NJ
71
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
New
Castle
DE
New
Haven
CT
Ocean
NJ
Philadelphia
PA
Richmond
NY
Suffolk
NY
Westchester
NY
MI
19
Anne
Arundel
MD
Bergen
NJ
Bucks
PA
Camden
NJ
Cecil
MD
Chester
PA
Erie
NY
Geauga
OH
Gloucester
NJ
Kent
MD
Mercer
NJ
Middlesex
NJ
Monmouth
NJ
Morris
NJ
New
Castle
DE
Ocean
NJ
Philadelphia
PA
Richmond
NY
Suffolk
NY
MO
4
Geauga
OH
Kenosha
WI
Ozaukee
WI
Sheboygan
WI
MS
2
Harris
TX
Jefferson
TX
NC
8
Anne
Arundel
MD
Fulton
GA
Harford
MD
Kent
MD
Newcastle
DE
Suffolk
NY
Bucks
PA
Chester
PA
NJ
10
Erie
NY
Fairfield
CT
Kent
RI
Middlesex
CT
Montgomery
PA
New
Haven
CT
Philadelphia
Richmond
NY
Suffolk
NY
Westchester
NY
NY
9
Fairfield
CT
Kent
RI
Mercer
NJ
Middlesex
CT
Middlesex
NJ
Monmouth
NJ
Morris
NJ
New
Haven
CT
Ocean
NJ
OH
28
Anne
Arundel
MD
Arlington
VA
Bergen
NJ
Bucks
PA
Camden
NJ
Cecil
MD
Chester
PA
District
of
Columbia
Fairfax
VA
Fairfield
CT
Gloucester
NJ
Harford
MD
Hunterton
NJ
Kent
MD
Kent
RI
Macomb
MI
Mercer
NJ
Middlesex
CT
Middlesex
NJ
Monmouth
NJ
Montgomery
PA
Morris
NJ
New
Castle
DE
New
Haven
CT
Ocean
NJ
Philadelphia
Suffolk
NY
Westchester
NY
PA
25
Anne
Arundel
MD
Arlington
VA
Bergen
NJ
Camden
NJ
Cecil
MD
District
of
Columbia
Erie
NY
Fairfax
VA
Fairfield
CT
Gloucester
NJ
Harford
MD
Hunterton
NJ
Kent
MD
Kent
RI
Mercer
NJ
Middlesex
CT
Middlesex
NJ
Monmouth
NJ
Morris
NJ
New
Castle
DE
New
Haven
CT
Ocean
NJ
Richmond
NY
Suffolk
NY
Westchester
NY
SC
1
Fulton
GA
TN
1
Fulton
GA
VA
26
Anne
Arundel
MD
Bergen
NJ
Bucks
PA
Camden
NJ
Cecil
MD
Chester
PA
District
of
Columbia
Erie
NY
Fairfield
CT
Gloucester
NJ
Harford
MD
Hunterton
NJ
Kent
MD
Kent
RI
Mercer
NJ
Middlesex
CT
Middlesex
NJ
Monmouth
NJ
Morris
NJ
New
Castle
DE
New
Haven
CT
Ocean
NJ
Philadelphia
Richmond
NY
72
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Suffolk
NY
Westchester
NY
WI
2
Erie
NY
Macomb
MI
WV
25
Anne
Arundel
MD
Bergen
NJ
Bucks
PA
Camden
NJ
Cecil
MD
Chester
PA
Fairfax
VA
Fairfield
CT
Fulton
GA
Gloucester
NJ
Harford
MD
Hunterton
NJ
Kent
MD
Mercer
NJ
Middlesex
NJ
Monmouth
NJ
Montgomery
PA
Morris
NJ
New
Castle
DE
New
Haven
CT
Ocean
NJ
Philadelphia
Richmond
NY
Suffolk
NY
Westchester
NY
E.
What
are
the
Estimated
Air
Quality
Impacts
of
the
Final
Rule?

In
this
section,
we
describe
the
air
quality
modeling
performed
to
determine
the
projected
impacts
on
PM2.5
and
8­
hour
ozone
of
the
SO2
and
NOx
emissions
reductions
in
the
control
region
modeled.
The
modeling
used
to
estimate
the
air
quality
impact
of
these
reductions
assumes
annual
SO2
and
NOx
controls
for
Arkansas,
Delaware,
and
New
Jersey
in
addition
to
the
23­

States
plus
the
District
of
Columbia.
Since
Arkansas,
Delaware,

and
New
Jersey
are
not
included
in
the
final
CAIR
region
for
PM2.5,
the
modeled
estimated
impacts
on
PM2.5
are
overstated
for
today's
final
rule.
However,
EPA
plans
to
propose
to
include
these
three
States
in
the
CAIR
region
for
PM2.5
through
a
separate
regulatory
process.
Thus,
the
estimates
are
reflective
of
the
total
impacts
expected
for
CAIR
assuming
Arkansas,

Delaware,
and
New
Jersey
will
become
part
of
the
annual
SO2
and
NOx
trading
program.

As
discussed
in
section
IV,
EPA
analyzed
the
impacts
of
the
regional
emissions
reductions
in
both
2010
and
2015.
These
73
16
In
addition
to
the
SO2
and
NOx
reductions
in
these
States,
we
also
modeled
summer­
season
only
EGU
NOx
controls
for
Connecticut
and
Massachusetts,
which
significantly
contribute
to
ozone,
but
not
to
PM2.5
nonattainment
in
downwind
areas.

17
For
the
purposes
of
this
discussion,
we
have
calculated
the
percent
reduction
in
total
EGU
emissions
which
includes
units
greater
than
and
less
than
25
MW.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
impacts
are
quantified
by
comparing
air
quality
modeling
results
for
the
regional
control
scenario
to
the
modeling
results
for
the
corresponding
2010
and
2015
Base
Case
scenarios.
The
2010
and
2015
emissions
reductions
from
the
power
generation
sector
include
a
two­
phase
cap
and
trade
program
covering
the
control
region
modeled
(
i.
e.,
the
23
States
plus
the
District
of
Columbia
included
in
today's
rule
and
Arkansas,
Delaware,
and
New
Jersey).
16
Phase
1
of
the
regional
strategy
(
the
2010
reductions)
is
forecast
to
reduce
total
EGU
SO2
emissions17
in
the
control
region
modeled
by
40
percent
in
2010.
Phase
2
(
the
2015
reductions)
is
forecast
to
provide
a
48
percent
reduction
in
EGU
SO2
emissions
compared
to
the
Base
Case
in
2015.
When
fully
implemented
post­
2015,
we
expect
this
rule
to
result
in
more
than
a
70
percent
reduction
in
EGU
SO2
emissions
compared
to
current
emissions
levels.
The
reductions
at
full
implementation
occur
post­
2015
due
to
the
existing
title
IV
bank
of
SO2
allowances,

which
can
be
used
under
the
CAIR
program.
The
net
effect
of
the
strategy
on
total
SO2
emissions
in
the
control
region
modeled
74
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
considering
all
sources
of
emissions,
is
a
28
percent
reduction
in
2010
and
a
32
percent
reduction
in
2015.

For
NOx,
Phase
1
of
the
strategy
is
forecast
to
reduce
total
EGU
emissions
by
44
percent
in
2009.
Total
NOx
emissions
across
the
control
region
(
i.
e.,
includes
all
sources)
are
11
percent
lower
in
the
2010
CAIR
scenario
compared
to
the
emissions
in
the
2010
Base
Case.
In
Phase
2,
EGU
NOx
emissions
are
projected
to
decline
by
54
percent
in
2015
in
this
region.
Total
NOx
emissions
from
all
anthropogenic
sources
are
projected
to
be
reduced
by
14
percent
in
2015.
The
percent
change
in
emissions
by
State
for
SO2
and
NOx
in
2010
and
2015
for
the
regional
control
strategy
modeled
are
provided
in
the
NFR
EITSD.

1.
Estimated
Impacts
on
PM2.5
Concentrations
and
Attainment
We
determined
the
impacts
on
PM2.5
of
the
CAIR
regional
strategy
by
running
the
CMAQ
model
for
this
strategy
and
comparing
the
results
to
the
PM2.5
concentrations
predicted
for
the
2010
and
2015
Base
Cases.
In
brief,
we
ran
the
CMAQ
model
for
the
regional
strategy
in
both
2010
and
2015.
The
model
predictions
were
used
to
project
future
PM2.5
concentrations
for
CAIR
in
2010
and
2015
using
the
SMAT
technique,
as
described
in
section
VI.
B.
1.
We
compared
the
results
of
the
2010
and
2015
regional
strategy
modeling
to
the
corresponding
results
from
the
2010
and
2015
Base
Cases
to
quantify
the
expected
impacts
of
CAIR.
75
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
The
impacts
of
the
SO2
and
NOx
emissions
reductions
expected
from
CAIR
on
PM2.5
in
2010
and
2015
are
provided
in
Table
VI­
10
and
Table
VI­
11,
respectively.
In
these
tables,
counties
shown
in
bold/
italics
are
projected
to
come
into
attainment
with
CAIR.

Table
VI­
10.
Projected
PM2.5
Concentrations
(

g/
m3)
for
the
2010
Base
Case
and
CAIR
and
the
Impact
of
CAIR
Regional
Controls
in
2010.

State
County
2010
Base
Case
2010
CAIR
Impact
of
CAIR
Alabama
DeKalb
Co
15.23
13.97
­
1.26
Alabama
Jefferson
Co
18.57
17.46
­
1.11
Alabama
Montgomery
Co
15.12
14.10
­
1.02
Alabama
Morgan
Co
15.29
14.11
­
1.18
Alabama
Russell
Co
16.17
15.15
­
1.02
Alabama
Talladega
Co
15.34
14.00
­
1.34
Delaware
New
Castle
Co
16.56
14.84
­
1.72
District
of
Columbia
15.84
13.68
­
2.16
Georgia
Bibb
Co
16.27
15.17
­
1.10
Georgia
Clarke
Co
16.39
14.96
­
1.43
Georgia
Clayton
Co
17.39
16.29
­
1.10
Georgia
Cobb
Co
16.57
15.35
­
1.22
Georgia
DeKalb
Co
16.75
15.70
­
1.05
Georgia
Floyd
Co
16.87
15.87
­
1.00
Georgia
Fulton
Co
18.02
16.98
­
1.04
Georgia
Hall
Co
15.60
14.28
­
1.32
Georgia
Muscogee
Co
15.65
14.57
­
1.08
Georgia
Richmond
Co
15.68
14.64
­
1.04
Georgia
Walker
Co
15.43
14.22
­
1.21
Georgia
Washington
Co
15.31
14.22
­
1.09
Georgia
Wilkinson
Co
16.27
15.22
­
1.05
Illinois
Cook
Co
17.52
16.88
­
0.64
Illinois
Madison
Co
16.66
15.96
­
0.70
Illinois
St.
Clair
Co
16.24
15.54
­
0.70
Indiana
Clark
Co
16.51
15.15
­
1.36
Indiana
Dubois
Co
15.73
14.37
­
1.36
Indiana
Lake
Co
17.26
16.48
­
0.78
Indiana
Marion
Co
16.83
15.54
­
1.29
Indiana
Vanderburgh
Co
15.54
14.26
­
1.28
76
State
County
2010
Base
Case
2010
CAIR
Impact
of
CAIR
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Kentucky
Boyd
Co
15.23
13.38
­
1.85
Kentucky
Bullitt
Co
15.10
13.67
­
1.43
Kentucky
Fayette
Co
15.95
14.17
­
1.78
Kentucky
Jefferson
Co
16.71
15.44
­
1.27
Kentucky
Kenton
Co
15.30
13.72
­
1.58
Maryland
Anne
Arundel
Co
15.26
12.98
­
2.28
Maryland
Baltimore
city
16.96
14.88
­
2.08
Michigan
Wayne
Co
19.41
18.23
­
1.18
Missouri
St.
Louis
City
15.10
14.40
­
0.70
New
Jersey
Union
Co
15.05
13.60
­
1.45
New
York
New
York
Co
16.19
14.95
­
1.24
North
Carolina
Catawba
Co
15.48
14.07
­
1.41
North
Carolina
Davidson
Co
15.76
14.36
­
1.40
North
Carolina
Mecklenburg
Co
15.22
13.92
­
1.30
Ohio
Butler
Co
16.45
15.03
­
1.42
Ohio
Cuyahoga
Co
18.84
17.11
­
1.73
Ohio
Franklin
Co
16.98
15.13
­
1.85
Ohio
Hamilton
Co
18.23
16.61
­
1.62
Ohio
Jefferson
Co
17.94
15.64
­
2.30
Ohio
Lawrence
Co
16.10
14.11
­
1.99
Ohio
Mahoning
Co
15.39
13.40
­
1.99
Ohio
Montgomery
Co
15.41
13.83
­
1.58
Ohio
Scioto
Co
18.13
15.98
­
2.15
Ohio
Stark
Co
17.14
15.08
­
2.06
Ohio
Summit
Co
16.47
14.69
­
1.78
Ohio
Trumbull
Co
15.28
13.50
­
1.78
Pennsylvania
Allegheny
Co
20.55
18.01
­
2.54
Pennsylvania
Beaver
Co
15.78
13.61
­
2.17
Pennsylvania
Berks
Co
15.89
13.56
­
2.33
Pennsylvania
Cambria
Co
15.14
12.72
­
2.42
Pennsylvania
Dauphin
Co
15.17
12.88
­
2.29
Pennsylvania
Delaware
Co
15.61
13.94
­
1.67
Pennsylvania
Lancaster
Co
16.55
14.09
­
2.46
Pennsylvania
Philadelphia
Co
16.65
14.98
­
1.67
Pennsylvania
Washington
Co
15.23
12.99
­
2.24
Pennsylvania
Westmoreland
Co
15.16
12.60
­
2.56
Pennsylvania
York
Co
16.49
14.20
­
2.29
Tennessee
Davidson
Co
15.36
14.26
­
1.10
Tennessee
Hamilton
Co
16.89
15.57
­
1.32
77
State
County
2010
Base
Case
2010
CAIR
Impact
of
CAIR
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Tennessee
Knox
Co
17.44
16.16
­
1.28
Tennessee
Sullivan
Co
15.32
14.01
­
1.31
West
Virginia
Berkeley
Co
15.69
13.43
­
2.26
West
Virginia
Brooke
Co
16.63
14.42
­
2.21
West
Virginia
Cabell
Co
17.03
15.08
­
1.95
West
Virginia
Hancock
Co
17.06
14.89
­
2.17
West
Virginia
Kanawha
Co
17.56
15.27
­
2.29
West
Virginia
Marion
Co
15.32
12.90
­
2.42
West
Virginia
Marshall
Co
15.81
13.46
­
2.35
West
Virginia
Ohio
Co
15.14
12.81
­
2.33
West
Virginia
Wood
Co
16.66
14.14
­
2.52
Table
VI­
11.
Projected
PM2.5
Concentrations
(

g/
m3)
for
the
2015
Base
Case
and
CAIR
and
the
Impact
of
CAIR
Regional
Controls
in
2015.

State
County
2015
Base
Case
2015
CAIR
Impact
of
CAIR
Alabama
DeKalb
Co
15.24
13.46
­
1.78
Alabama
Jefferson
Co
18.85
17.36
­
1.49
Alabama
Montgomery
Co
15.24
13.87
­
1.37
Alabama
Morgan
Co
15.26
13.85
­
1.41
Alabama
Russell
Co
16.10
14.66
­
1.44
Alabama
Talladega
Co
15.22
13.35
­
1.87
Delaware
New
Castle
Co
16.47
14.41
­
2.06
District
of
Columbia
15.57
13.11
­
2.46
Georgia
Bibb
Co
16.41
14.83
­
1.58
Georgia
Chatham
Co
15.06
13.86
­
1.20
Georgia
Clarke
Co
16.15
14.10
­
2.05
Georgia
Clayton
Co
17.46
15.85
­
1.61
Georgia
Cobb
Co
16.51
14.67
­
1.84
Georgia
DeKalb
Co
16.82
15.29
­
1.53
Georgia
Floyd
Co
17.33
15.79
­
1.54
Georgia
Fulton
Co
18.00
16.47
­
1.53
Georgia
Hall
Co
15.36
13.48
­
1.88
Georgia
Muscogee
Co
15.58
14.06
­
1.52
Georgia
Richmond
Co
15.76
14.23
­
1.53
Georgia
Walker
Co
15.37
13.65
­
1.72
Georgia
Washington
Co
15.34
13.67
­
1.67
78
State
County
2015
Base
Case
2015
CAIR
Impact
of
CAIR
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Georgia
Wilkinson
Co
16.54
15.01
­
1.53
Illinois
Cook
Co
17.71
16.95
­
0.76
Illinois
Madison
Co
16.90
16.07
­
0.83
Illinois
St.
Clair
Co
16.49
15.64
­
0.85
Illinois
Will
Co
15.12
14.27
­
0.85
Indiana
Clark
Co
16.37
14.79
­
1.58
Indiana
Dubois
Co
15.66
14.16
­
1.50
Indiana
Lake
Co
17.27
16.36
­
0.91
Indiana
Marion
Co
16.77
15.38
­
1.39
Indiana
Vanderburgh
Co
15.56
14.17
­
1.39
Kentucky
Boyd
Co
15.06
12.95
­
2.11
Kentucky
Fayette
Co
15.62
13.54
­
2.08
Kentucky
Jefferson
Co
16.61
15.13
­
1.48
Kentucky
Kenton
Co
15.09
13.26
­
1.83
Maryland
Baltimore
city
17.04
14.50
­
2.54
Maryland
Baltimore
Co
15.08
12.75
­
2.33
Michigan
Wayne
Co
19.28
17.95
­
1.33
Mississippi
Jones
Co
15.18
14.06
­
1.12
Missouri
St.
Louis
city
15.34
14.50
­
0.84
New
York
New
York
Co
15.76
14.33
­
1.43
North
Carolina
Catawba
Co
15.19
13.45
­
1.74
North
Carolina
Davidson
Co
15.34
13.61
­
1.73
Ohio
Butler
Co
16.32
14.67
­
1.65
Ohio
Cuyahoga
Co
18.60
16.67
­
1.93
Ohio
Franklin
Co
16.64
14.57
­
2.07
Ohio
Hamilton
Co
18.03
16.10
­
1.93
Ohio
Jefferson
Co
17.83
15.26
­
2.57
Ohio
Lawrence
Co
15.92
13.71
­
2.21
Ohio
Mahoning
Co
15.13
12.94
­
2.19
Ohio
Montgomery
Co
15.16
13.33
­
1.83
Ohio
Scioto
Co
17.92
15.55
­
2.37
Ohio
Stark
Co
16.86
14.58
­
2.28
Ohio
Summit
Co
16.14
14.18
­
1.96
Ohio
Trumbull
Co
15.05
13.08
­
1.97
Pennsylvania
Allegheny
Co
20.33
17.47
­
2.86
Pennsylvania
Beaver
Co
15.54
13.09
­
2.45
Pennsylvania
Berks
Co
15.66
12.99
­
2.67
Pennsylvania
Delaware
Co
15.52
13.52
­
2.00
Pennsylvania
Lancaster
Co
16.28
13.33
­
2.95
Pennsylvania
Philadelphia
Co
16.53
14.53
­
2.00
79
State
County
2015
Base
Case
2015
CAIR
Impact
of
CAIR
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Pennsylvania
York
Co
16.22
13.46
­
2.76
Tennessee
Davidson
Co
15.36
14.02
­
1.34
Tennessee
Hamilton
Co
16.82
14.94
­
1.88
Tennessee
Knox
Co
17.34
15.61
­
1.73
Tennessee
Shelby
Co
15.17
14.19
­
0.98
Tennessee
Sullivan
Co
15.37
13.77
­
1.60
West
Virginia
Berkeley
Co
15.32
12.73
­
2.59
West
Virginia
Brooke
Co
16.51
14.05
­
2.46
West
Virginia
Cabell
Co
16.86
14.64
­
2.22
West
Virginia
Hancock
Co
16.97
14.54
­
2.43
West
Virginia
Kanawha
Co
17.17
14.66
­
2.51
West
Virginia
Marshall
Co
15.52
12.87
­
2.65
West
Virginia
Wood
Co
16.69
13.88
­
2.81
As
described
in
section
VI.
B.
1,
we
project
that
79
counties
in
the
East
will
be
nonattainment
for
PM2.5
in
the
2010
Base
Case.
We
estimate
that,
on
average,
the
regional
strategy
will
reduce
PM2.5
in
these
79
counties
by
1.6

g/
m3.
In
over
90
percent
of
the
nonattainment
counties
(
i.
e.,
74
out
of
79
counties),
we
project
that
PM2.5
will
be
reduced
by
at
least
1.0

g/
m3.
In
over
25
percent
of
the
79
nonattainment
counties
(
i.
e.,
23
of
the
79
counties),
we
project
PM2.5
concentrations
will
decline
by
of
more
than
2.0

g/
m3.
Of
the
79
counties
that
are
nonattainment
in
the
2010
Base,
we
project
that
51
counties
will
come
into
attainment
as
a
result
of
the
SO2
and
NOx
emissions
reductions
expected
from
the
regional
controls.
Even
those
28
counties
that
remain
nonattainment
in
2010
after
80
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
implementation
of
the
regional
strategy
will
be
closer
to
attainment
as
a
result
of
these
emissions
reductions.

Specifically,
the
average
reduction
of
PM2.5
in
the
28
residual
nonattainment
counties
is
projected
to
be
1.3

g/
m3.
After
implementation
of
the
regional
controls,
we
project
that
18
of
the
28
residual
nonattainment
counties
in
2010
will
be
within
1.0

g/
m3
of
the
NAAQS
and
12
counties
will
be
within
0.5

g/
m3
of
attainment.

In
2015
we
are
projecting
that
PM2.5
in
the
74
Base
Case
nonattainment
counties
will
be
reduced
by
1.8

g/
m3,
on
average,

as
a
result
of
the
SO2
and
NOx
reductions
in
the
regional
strategy.
In
over
90
percent
of
the
nonattainment
counties
(
i.
e.,
67
of
the
74
counties)
concentrations
of
PM2.5
are
predicted
to
be
reduced
by
at
least
1.0

g/
m3.
In
over
35
percent
of
the
counties
(
i.
e.,
27
of
the
74
counties),
we
project
the
regional
strategy
to
reduce
PM2.5
by
more
than
2.0

g/
m3.
As
a
result
of
the
reductions
in
PM2.5,
56
nonattainment
counties
are
projected
to
come
into
attainment
in
2015.
The
remaining
18
nonattainment
counties
are
projected
to
be
closer
to
attainment
with
the
regional
strategy.
Our
modeling
results
indicate
that
PM2.5
will
be
reduced
in
the
range
of
0.7

g/
m3
to
2.9

g/
m3
in
these
18
counties.
The
average
reduction
across
these
18
residual
nonattainment
counties
is
1.5

g/
m3.

Thus,
the
SO2
and
NOx
emissions
reductions
which
will
result
81
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
from
the
regional
strategy
will
greatly
reduce
the
extent
of
PM2.5
nonattainment
by
2010
and
beyond.
These
emissions
reductions
are
expected
to
substantially
reduce
the
number
of
PM2.5
nonattainment
counties
in
the
East
and
make
attainment
easier
for
those
counties
that
remain
nonattainment
by
substantially
lowering
PM2.5
concentrations
in
these
residual
nonattainment
counties.

2.
Estimated
Impacts
on
8­
Hour
Ozone
Concentrations
and
Attainment
We
determined
the
impacts
on
8­
hour
ozone
of
the
regional
strategy
by
running
the
CAMx
model
for
this
strategy
and
comparing
the
results
to
the
ozone
concentrations
predicted
for
the
2010
and
2015
Base
Cases.
In
brief,
we
ran
the
CAMx
model
for
the
regional
strategy
in
both
2010
and
2015.
The
model
predictions
were
used
to
project
future
8­
hour
ozone
concentrations
for
the
regional
strategy
in
2010
and
2015
using
the
Relative
Reduction
Factor
technique,
as
described
in
section
VI.
B.
1.
We
compared
the
results
of
the
2010
and
2015
regional
strategy
modeling
to
the
corresponding
results
from
the
2010
and
2015
Base
Cases
to
quantify
the
expected
impacts
of
the
regional
controls.

The
results
of
the
regional
strategy
ozone
modeling
are
expressed
in
terms
of
the
expected
reductions
in
projected
8­
hour
concentrations
and
the
implications
for
future
nonattainment.
82
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
The
impacts
of
the
regional
NOx
emissions
reductions
on
8­
hour
ozone
in
2010
and
2015
are
provided
in
Table
VI­
12
and
Table
VI­

13,
respectively.
In
these
tables,
counties
shown
in
bold/
italics
are
projected
to
come
into
attainment
with
the
regional
controls.

Table
VI­
12.
Projected
8­
hour
Concentrations
(
ppb)
for
the
2010
Base
Case
and
CAIR
and
the
Impact
of
CAIR
Regional
Controls
in
2010.

State
County
2010
Base
Case
2010
CAIR
Impact
of
CAIR
Connecticut
Fairfield
Co
92.6
92.2
­
0.4
Connecticut
Middlesex
Co
90.9
90.6
­
0.3
Connecticut
New
Haven
Co
91.6
91.3
­
0.3
District
of
Columbia
District
of
Columbia
85.2
85.0
­
0.2
Delaware
New
Castle
Co
85.0
84.7
­
0.3
Georgia
Fulton
Co
86.5
85.1
­
1.4
Maryland
Anne
Arundel
Co
88.8
88.6
­
0.2
Maryland
Cecil
Co
89.7
89.5
­
0.2
Maryland
Harford
Co
93.0
92.8
­
0.2
Maryland
Kent
Co
86.2
85.8
­
0.4
Michigan
Macomb
Co
85.5
85.4
­
0.1
New
Jersey
Bergen
Co
86.9
86.0
­
0.9
New
Jersey
Camden
Co
91.9
91.6
­
0.3
New
Jersey
Gloucester
Co
91.8
91.3
­
0.5
New
Jersey
Hunterdon
Co
89.0
88.6
­
0.4
New
Jersey
Mercer
Co
95.6
95.2
­
0.4
New
Jersey
Middlesex
Co
92.4
92.1
­
0.3
New
Jersey
Monmouth
Co
86.6
86.4
­
0.2
New
Jersey
Morris
Co
86.5
85.5
­
1.0
New
Jersey
Ocean
Co
100.5
100.3
­
0.2
New
York
Erie
Co
87.3
86.9
­
0.4
New
York
Richmond
Co
87.3
87.1
­
0.2
New
York
Suffolk
Co
91.1
90.8
­
0.3
New
York
Westchester
Co
85.3
84.7
­
0.6
83
State
County
2010
Base
Case
2010
CAIR
Impact
of
CAIR
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
Ohio
Geauga
Co
87.1
86.6
­
0.5
Pennsylvania
Bucks
Co
94.7
94.3
­
0.4
Pennsylvania
Chester
Co
85.7
85.4
­
0.3
Pennsylvania
Montgomery
Co
88.0
87.6
­
0.4
Pennsylvania
Philadelphia
Co
90.3
89.9
­
0.4
Rhode
Island
Kent
Co
86.4
86.2
­
0.2
Texas
Denton
Co
87.4
86.8
­
0.6
Texas
Galveston
Co
85.1
84.6
­
0.5
Texas
Harris
Co
97.9
97.4
­
0.5
Texas
Jefferson
Co
85.6
85.0
­
0.6
Texas
Tarrant
Co
87.8
87.2
­
0.6
Virginia
Arlington
Co
86.2
86.0
­
0.2
Virginia
Fairfax
Co
85.7
85.4
­
0.3
Wisconsin
Kenosha
Co
91.3
91.0
­
0.3
Wisconsin
Ozaukee
Co
86.2
85.8
­
0.4
Wisconsin
Sheboygan
Co
88.3
87.7
­
0.6
Table
VI­
13.
Projected
8­
hour
Concentrations
(
ppb)
for
the
2015
Base
Case
and
CAIR
and
the
Impact
of
CAIR
Regional
Controls
in
2015.

State
County
2015
Base
Case
2015
CAIR
Impact
of
CAIR
Connecticut
Fairfield
Co
91.4
90.6
­
0.8
Connecticut
Middlesex
Co
89.1
88.4
­
0.7
Connecticut
New
Haven
Co
89.8
89.1
­
0.7
Maryland
Anne
Arundel
Co
86.0
84.9
­
1.1
Maryland
Cecil
Co
86.9
85.4
­
1.5
Maryland
Harford
Co
90.6
89.6
­
1.0
Michigan
Macomb
Co
85.1
84.2
­
0.9
New
Jersey
Bergen
Co
85.7
84.5
­
1.2
New
Jersey
Camden
Co
89.5
88.3
­
1.2
84
State
County
2015
Base
Case
2015
CAIR
Impact
of
CAIR
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
New
Jersey
Gloucester
Co
89.6
88.2
­
1.4
New
Jersey
Hunterdon
Co
86.5
85.4
­
1.1
New
Jersey
Mercer
Co
93.5
92.4
­
1.1
New
Jersey
Middlesex
Co
89.8
88.8
­
1.0
New
Jersey
Ocean
Co
98.0
96.9
­
1.1
New
York
Erie
Co
85.2
84.2
­
1.0
New
York
Suffolk
Co
89.9
89.0
­
0.9
Pennsylvania
Bucks
Co
93.0
91.8
­
1.2
Pennsylvania
Montgomery
Co
86.5
84.9
­
1.6
Pennsylvania
Philadelphia
Co
88.9
87.5
­
1.4
Texas
Harris
Co
97.3
96.4
­
0.9
Texas
Jefferson
Co
85.0
84.1
­
0.9
Wisconsin
Kenosha
Co
89.4
88.8
­
0.6
As
described
in
section
VI.
B.
1,
we
project
that
40
counties
in
the
East
would
be
nonattainment
for
8­
hour
ozone
under
the
assumptions
in
the
2010
Base
Case.
Our
modeling
of
the
regional
controls
in
2010
indicates
that
3
of
these
counties
will
come
into
attainment
of
the
8­
hour
ozone
NAAQS
and
that
ozone
in
16
of
the
40
nonattainment
counties
will
be
reduced
by
1
ppb
or
more.
In
addition,
our
modeling
predicts
that
8­
hour
ozone
exceedances
(
i.
e.,
8­
hour
ozone
of
85
ppb
or
higher)
within
nonattainment
areas
are
expected
to
decline
by
5
percent
in
2010
with
CAIR.
Of
the
37
counties
that
are
projected
to
remain
nonattainment
in
2010
after
the
regional
strategy,
nearly
half
(
i.
e.,
16
of
the
37
counties)
are
within
2
ppb
of
attainment.

In
2015,
we
project
that
6
of
the
22
counties
which
are
85
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
nonattainment
for
8­
hour
ozone
in
the
Base
Case
will
come
into
attainment
with
the
regional
strategy.
Ozone
concentrations
in
over
70
percent
(
i.
e.,
16
of
22
counties)
of
the
2015
Base
Case
nonattainment
counties
are
projected
to
be
reduced
by
1
ppb
or
more
as
a
result
of
the
regional
strategy.
Exceedances
of
the
8­
hour
ozone
NAAQS
are
predicted
to
decline
in
nonattainment
areas
by
14
percent
with
regional
controls
in
place
in
2015.

Thus,
the
NOx
emissions
reductions
which
will
result
from
the
regional
strategy
will
help
to
bring
8­
hour
ozone
nonattainment
areas
in
the
East
closer
to
attainment
by
2010
and
beyond.

F.
What
are
the
Estimated
Visibility
Impacts
of
the
Final
Rule?

1.
Methods
for
Calculating
Projected
Visibility
in
Class
I
Areas
The
NPR
contained
example
future
year
visibility
projections
for
the
20
percent
worst
days
and
20
percent
best
days
at
Class
I
areas
that
had
complete
IMPROVE
monitoring
data
in
1996.
Changes
in
future
visibility
were
predicted
by
using
the
REMSAD
model
to
generate
relative
visibility
changes,
then
applying
those
changes
to
measured
current
visibility
data.

Details
of
the
visibility
modeling
and
calculations
can
be
found
in
the
NPR
AQMTSD.
An
example
visibility
calculation
was
given
in
Appendix
M
of
the
NPR
AQMTSD
along
with
the
predicted
improvement
in
visibility
(
in
deciviews)
on
the
20
percent
best
and
worst
days
at
44
Class
I
areas.
The
data
contained
in
Appendix
M
was
for
informational
purposes
only
and
was
not
used
86
18
The
CAIR
scenario
modeled
for
the
visibility
analysis
included
controls
in
Arkansas,
Delaware,
and
New
Jersey.

Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
in
the
significant
contribution
determination
or
control
strategy
development
decisions.

The
SNPR
contained
visibility
calculations
in
support
of
the
"
better­
than­
BART"
analysis.
The
better­
than­
BART
analysis
employed
a
two­
pronged
test
to
determine
if
the
modeled
visibility
improvements
from
the
CAIR
cap
and
trade
program
for
EGU's
were
"
better"
than
the
visibility
improvements
from
a
nationwide
BART
program.
The
analysis
used
the
visibility
calculation
methodology
detailed
in
the
NPR
TSD.
Detailed
results
of
the
SNPR
better­
than­
BART
analysis
are
contained
in
the
SNPR
AQMTSD.
The
better­
than­
BART
analysis
for
the
final
rule
is
addressed
in
section
IX.
C.
2
of
the
preamble.
Additional
information
on
the
visibility
calculation
methodology
is
contained
in
the
NFR
AQMTSD.

2.
Visibility
Improvements
in
Class
I
Areas
For
the
NFR
we
have
modeled
several
new
CAIR18
and
CAIR
+

BART
cases
to
re­
examine
the
better­
than­
BART
two­
pronged
test.

We
have
modeled
an
updated
nationwide
BART
scenario
as
well
as
a
CAIR
in
the
East/
BART
in
the
West
scenario.
The
results
were
analyzed
at
116
Class
I
areas
that
have
complete
IMPROVE
data
for
2001
or
are
represented
by
IMPROVE
monitors
with
complete
data.
Twenty
nine
of
the
Class
I
areas
are
in
the
East
and
87
87
Section
VI
1/
14/
2005
DRAFT
Do
Not
Quote
or
Cite
are
in
the
West.
The
results
of
the
visibility
analysis
are
summarized
in
section
IX.
C.
2.
Detailed
results
for
all
116
Class
I
areas
are
presented
in
the
NFR
AQMTSD.
