APPENDIX
A
CONCEPTUAL
MODEL
AND
EPISODE
SELECTION
FOR
THE
SAN
ANTONIO
EAC
REGION
SAN
ANTONIO
EAC
REGION
ATTAINMENT
DEMONSTRATION
MARCH
2004
A­
2
Conceptual
Model
and
Episode
Selection
for
the
San
Antonio
EAC
Region
Prepared
by:
The
Alamo
Area
Council
of
Governments
Natural
Resources
/
Transportation
Department
San
Antonio,
Texas
September
2000
A­
3
Table
of
Contents
Introduction
                            
A­
4
Preliminary
Concepts
                       ..
A­
5
CAMS
in
the
San
Antonio
Region
               .
A­
5
Design
Value
                        .
A­
6
Elements
of
the
Conceptual
Model                  .
A­
7
Local
Monitored
Data
                    ..
A­
7
Seasonal
Patterns
of
High
Ozone
Occurrences
      
A­
7
Local
Wind
Patterns:
Monitoring
Station
Data
       
A­
9
Regional
Modeling
Data
                   .
A­
10
The
HYSPLIT
Model
                 .
A­
10
Regional
Wind
Patterns:
Back
Trajectories
        
A­
10
Episode
Selection
                         .
A­
14
Episode
Candidates:
Exceedance
Days
            ..
A­
14
Requirements
Limiting
Episode
Selection
            
A­
15
Comparison
of
Back
Trajectories
            .
A­
21
Regional
Considerations
               ..
A­
31
Conclusion
                            .
A­
32
INTRODUCTION
The
U.
S.
Environmental
Protection
Agency
(
EPA)
is
charged
with
the
maintenance
of
regional
air
quality
across
the
United
States
through
a
series
of
standards,
the
National
Ambient
Air
Quality
Standards
(
NAAQS).
When
regions
fail
to
comply
with
these
standards,
the
region
joins
together
with
the
state
and
several
federal
entities
to
create
and
agree
upon
a
State
Implementation
Plan
(
SIP).
The
SIP
is
a
blueprint
for
the
methodology
that
the
region
and
state
will
follow
to
allow
the
region
to
regain
the
federal
air
quality
standards.

Air
quality
analysis
and
the
modeling
of
control
strategies
are
elements
of
a
SIP.
Since
control
strategy
modeling
requires
extensive
technical
analyses
of
control
strategy
impacts
under
all
meteorological
conditions
that
give
rise
to
high
levels
of
ozone
formation,
it
is
important
that
each
photochemical
modeling
episode
be
built
upon
a
time
period
characterized
by
such
meteorological
conditions.
Hence,
careful
selection
of
the
proper
episode
is
important
for
use
in
photochemical
modeling.

The
EPA
suggests
that
a
conceptual
description1,
or
model,
be
developed
to
aid
in
the
selection
of
modeling
episodes.
The
following
paper
represents
a
Preliminary
Conceptual
Model
(
PCM),
developed
and
used
for
episode
selection
for
initial
eight­
hour
modeling.
A
conceptual
model
profiles
or
typifies
the
meteorological
conditions
during
which
high
levels
of
ozone
are
created
for
a
region
through
the
study
of
the
meteorology
accompanying
high
levels
of
ozone.
The
days
which
will
comprise
the
modeling
episode
are
specifically
chosen
because
they
reflect
the
area's
meteorology
during
the
formation
of
high
ozone
levels.
Thus,
a
successful
PCM
will
supply
an
identification
of
the
best
time
periods
for
the
modeler
to
incorporate
into
a
photochemical
model
in
order
to
evaluate
control
strategies.
An
interim
conceptual
model
includes
modifications
made
to
the
PCM
during
the
development
of
the
modeling
protocol
and
base
case
modeling.
The
refined
conceptual
model
will
be
developed
after
initial
modeling
has
been
completed
and
control
strategies
have
been
implemented
2.

The
San
Antonio
area
episode(
s)
will
include
days
during
which
measured
ozone
levels
exceed
the
8­
hour
average
ozone
NAAQS
concentration
standard
of
85
parts
per
billion
(
ppb).
If
during
a
single
day
it
is
found
that
the
8­
hour
average
ozone
level
is
85
ppb
or
above,
while
meteorological
conditions
are
unexceptional,
special
notice
of
that
day
is
taken.
When
several
such
days
occur
in
a
series,
that
set
of
days
is
a
photochemical
model
episode
candidate.
The
process
undertaken
for
identification
of
such
episodes
in
the
San
Antonio
region,
the
analysis
of
the
candidate
episodes,
and
the
rationale
for
their
final
ranking
and
selection
are
the
subject
of
this
report.

1
Pg
18
of
168,
"
DRAFT
GUIDANCE
ON
THE
USE
OF
MODELS
AND
OTHER
ANALYSES
IN
ATTAINMENT
DEMONSTRATIONS
FOR
THE
8­
HOUR
OZONE
NAAQS,"
EPA­
454/
R­
99­
004,
May
1999.
Online:
http://
www.
epa.
gov/
ttn/
scram/
draft8hr.
zip
2
Page
2,
"
Development
of
a
Conceptual
Model
for
Episode
Selection
of
High
Eight­
Hour
Ozone
Events
in
the
Dallas
/
Fort
Worth
area,"
C.
Durrenberger,
P.
Breitenbach,
J.
Red,
D.
Sullivan,
S.
Minto,
TNRCC
A­
5
PRELIMINARY
CONCEPTS
CAMS
in
the
San
Antonio
Region
There
are
currently
four
air
quality
monitors
in
the
San
Antonio
region
that
record
ozone
levels
reported
to
the
public.
The
data
from
these
sites
are
archived
and
displayed
on
the
Internet3
by
the
Texas
Natural
Resource
Conservation
Commission
(
now
known
as
the
Texas
Commission
on
Environmental
Quality),
and
is
quality­
assured
to
EPA
standards.
The
monitoring
equipment
sets
within
this
network
are
called
CAMS,
which
is
an
acronym
for
Continuous
Air
Monitoring
Station4.
Information
about
San
Antonio
CAMS
sites
is
contained
in
the
table
below,
Table
A­
1.
Figure
A­
1,
on
the
following
page,
shows
the
locations
of
the
local
monitoring
sites.

Notice
that
CAMS07
was
deactivated
on
August
11,
1998
and
CAMS58
was
activated
on
August
12,
1998;
the
monitoring
equipment
located
at
CAMS07
was
moved
to
its
present
location
at
CAMS58.
Notice
also
that
only
CAMS23
has
been
active
during
the
entire
1997,
1998
and
1999
ozone
seasons;
data
from
this
monitor
will
be
critical
to
any
change
in
designation
under
the
8­
hour
ozone
NAAQS.

Table
A­
1.
Ozone­
Recording
CAMS
sites
in
the
San
Antonio
(
SA)
Airshed
CAMS
Designation
/
Site
Name
Address;
Location
Description
Data
Measured
First
date
of
data
reporting
(
online);
maintained
by
CAMS23
/
Marshall
High
School
6655
Bluebird
Lane;
northwestern
SA
Ozone,
Weather
Since
September
17,
1996;
TNRCC
CAMS58
/
Camp
Bullis
Near
Wilderness
road;
far
northern
SA
Ozone,
Weather
and
NOx
Since
August
12,
1998;
TNRCC
CAMS59
/
Calaveras
Lake
14620
Laguna
Road;
southeastern
San
Antonio
PM
2.5,
NOx,
Ozone,
and
Weather
Since
May
13,
1998;
University
of
Texas
at
Austin
CAMS678
/
CPS/
Trinity
802
Pecan
Valley
Dr.;
near
eastern
San
Antonio
CO,
SO2,
NOx,
Ozone,
and
Weather
Since
March
4,
1999;
by
Trinity
Consultants
for
CPS
CAMS07
/
San
Antonio
North
C07
522
Pilgrim
Dr.;
near
northern
San
Antonio
CO,
NOx,
Ozone,
and
Weather
Deactivated
on
August
11,
1998
In
addition
to
various
pollutant
readings,
the
weather
data
reported
from
each
of
these
sites
include
location­
specific
temperature,
wind
direction
and
wind
speed.
This
data
is
reported
online
as
hourly­
averaged
values.
Since
promulgation
of
the
8­
hour
ozone
NAAQS
in
1997,
eight­
hour
ozone
reading
averages
are
available
online
as
well.

As
mentioned
in
the
Introduction,
the
8­
hour
average
concentration
of
85
ppb
for
ozone
is
the
single
most
important
air
quality
measurement
for
San
Antonio.
According
to
the
NAAQS,
this
critical
threshold
value
determines
whether
an
area
is
or
is
not
in
attainment
of
the
8­
hour
standard.
If
the
average
of
the
annual
fourth­
highest
eight­
hour
average
for
three
consecutive
years
is
at
or
above
85
ppb
at
any
one
monitor,
that
region
is
not
in
attainment
of
the
NAAQS.

3
"
Air
and
Water
Monitoring,"
on­
line
August
3,
2000:
http://
www.
tnrcc.
state.
tx.
us/
air/
monops/
4
"
What
is
a
CAMS?",
http://
www.
tnrcc.
state.
tx.
us/
cgi­
bin/
monops/
daily_
info?
cams
A­
6
Figure
A­
1.
Monitoring
locations
in
the
San
Antonio
airshed.

Image
courtesy
of
TNRCC5.
In
addition
to
the
ozone
monitors
discussed,
this
image
shows
C140
(
weather
only),
C301
(
PM
2.5
only)
and
C27
(
CO
and
NOx
only)
CAMS
sites.

Design
Value
Another
useful
statistic
is
the
design
value.
In
the
San
Antonio
area,
the
current
design
value
is
88
ppb,
the
average
of
the
annual
fourth­
highest
ozone
readings
recorded
at
CAMS
23
(
Marshall
High
School)
during
the
1997,
1998,
and
1999
ozone
seasons.
The
effectiveness
of
control
strategies
in
helping
a
region
to
regain
attainment
is
measured
against
this
value.
Also,
when
selecting
episode
days,
the
EPA
recommends
that
daily
peak
ozone
8­
hour
averages
be
generally
within
10
ppb
above
the
8­
hour
design
value6.
The
design
value
will
be
discussed
further
in
the
section
treating
Episode
Selection.

5
http://
www.
tnrcc.
state.
tx.
us/
cgi­
bin/
monops/
select_
month?
region13.
gif
6
Conversation
with
Pete
Breitenbach,
TNRCC,
August
18,
2000
A­
7
ELEMENTS
OF
THE
CONCEPTUAL
MODEL
A
conceptual
model
identifies
meteorological
conditions
that
occur
during
days
of
excessive
ozone
formation.
A
high
ozone
day
is
classified
as
a
day
during
which
an
ozone
level
of
85
ppb
or
above,
when
averaged
for
an
eight­
hour
period,
or
125
ppb
or
above
for
a
one­
hour
averaging
period,
is
recorded.
Such
levels
exceed
NAAQS
air
quality
standards.
Days
during
which
such
levels
are
achieved
are
also
called
exceedance
days,
and
are
candidate
days
for
inclusion
in
a
modeling
episode.
(
While
the
one­
hour
average
ozone
NAAQS
carries
the
125
ppb
one­
hour
average
standard,
San
Antonio
is
currently
at
risk
to
lose
only
the
attainment
status
for
the
8­
hour
standard.)

Local
Monitored
Data
Seasonal
Patterns
of
High
Ozone
Occurences
After
compiling
a
list
of
these
ozone
exceedance
days
­­
using
both
the
one­
hour
and
eight­
hour
definitions
for
exceedance
­­
from
TNRCC's
archives,
the
task
of
identifying
patterns
in
the
data
begins.
The
meteorology
determined
for
all
exceedance
days
will,
by
definition,
reflect
all
of
the
meteorological
patterns
that
correspond
to
high
ozone.

The
ozone
season
for
the
region
is
seven
months
long,
lasting
from
April
to
October.
If,
on
a
given
day
from
1990
to
1999,
any
monitor
in
the
San
Antonio
region
showed
an
exceedance
for
either
the
one­
hour
or
the
eight­
hour
ozone
standard
(
125
ppb
and
85
ppb,
respectively),
that
day
was
counted.
Such
counts
were
totaled
by
two­
week
(
halfmonth
periods
and
plotted
in
Figure
2.
No
day
was
counted
more
than
once.

Within
the
ozone
season,
as
shown
in
Figure
A­
2,
there
are
two
prominent
periods
during
which
the
greatest
number
of
exceedances
occurred.
Of
the
57
exceedance
days
counted
for
San
Antonio,
16
(
28.1%)
occurred
between
early
May
and
late
July.
Also,
29
(
50.9%)
occurred
between
early
August
and
late
September.

This
guides
us
in
the
first
consideration.
That
is,
we
should
further
study
episode
candidates
associated
with
each
of
these
two
periods
within
the
ozone
season7.
It
will
likely
be
advisable
that
one
modeling
episode
be
drawn
from
each
period.

7
In
the
more
advanced
conceptual
model,
the
seasonal
periods
identified
should
be
scrutinized
for
underlying
patterns
unique
to
each
season;
perhaps
wind
patterns
are
unique
to
that
season
in
the
target
airshed.
Wind
patterns
may
indict
transport
sources.
Or
perhaps
there
are
sources
identifiable
within
the
emissions
inventory
which
follow
seasonal
activity
patterns.
In
both
cases,
such
identification
may
lead
to
season­
specific
control
strategies.
In
brief,
ozone
day
occurence
patterns
identified
according
to
season
hint
at
further
underlying
factors
affecting
ozone
formation.
Marking
the
patterns
by
the
calendar
is
not
important;
identifying
the
causes
underlying
the
temporal
distribution
of
occurences
is
important.
A­
8
Figure
A­
2.
High
Ozone
Readings
by
Two­
Week
Period
by
Region.

0
10
20
30
40
50
60
Early
Jan.
Late
Jan.
Early
Feb.
Late
Feb.
Early
Ma
r.
Late
Mar.
Early
Apr.
Late
Apr.
Early
May
Late
May
Early
Jun.
Late
Jun.
Early
Jul
.
Late
Jul.
Early
Aug.

L
ate
Aug.
Early
Se
p.
Late
Sep.
Early
Oct.
Late
Oct.
Early
Nov
.
Late
Nov.

E
arly
Dec.

L
ate
Dec.

Years
1990­
1997*
(
unless
stipulated)

Total
#

of
Days
Over
0.12
ppm/

1­

hr
or
0
.08
ppm/

8­

hrs
Houston
Area
Dallas­
Ft.
Worth
Central
Texas
Northeast
Texas
Beaumont­
P.
A.

Corpus
Christi
(
90­
98)
*

Clute
(
90­
98)
*

San
Antonio
Area
(
90­
99)
*
A­
9
Local
Wind
Patterns:
Monitoring
Station
Data
As
prepared
by
TNRCC
data
analysis
staff,
the
following
figure,
Figure
A­
3,
shows
the
wind
patterns
associated
with
days
of
both
low
and
high
levels
of
ozone
formation.
This
is
a
compilation
of
days
within
the
ozone
seasons
(
April
1
through
October
31)
from
1988
to
1997.
The
CAMS
morning
wind
velocities
(
direction
and
speed)
are
averaged
between
7:
00
hours
through
10:
59
hours
Central
Standard
Time
(
CST),
inclusive.
The
afternoon
wind
velocities
are
averaged
between
13:
00
and
16:
59
CST,
inclusive.
The
averages
shown
are
from
5
minute
averages
taken
at
all
CAMS
stations,
averaged
together.

Figure
A­
3.
San
Antonio
Wind
Roses
San
Antonio
Wind
Roses
Low
Ozone
Days,
Morning
Winds
High
Ozone
Days,
Morning
Winds
Low
Ozone
Days,
Afternoon
Winds
High
Ozone
Days,
Afternoon
Winds
Slight
shift
to
the
east
on
high
days.

This
graph
shows
that,
during
low
ozone
days,
16.1%
of
the
velocity
readings
in
the
morning
are
light
and
variable
(
wind
speed
<
0.5
meters
per
second),
while
the
direction
for
the
morning
winds
are
from
the
south,
southeast
or
southeasterly.
In
contrast,
during
high
ozone
days,
26%
of
the
velocity
readings
in
the
morning
are
light
and
variable,
while
the
morning
winds
shift
to
the
east.

In
the
same
manner,
during
low
ozone
days,
the
image
shows
that
11.9%
of
the
velocity
readings
in
the
afternoon
are
light
and
variable,
while
the
direction
for
the
afternoon
winds
are
from
the
southeast.
During
high
ozone
days,
12.8%
of
the
velocity
readings
in
the
afternoon
are
light
and
variable,
while
the
afternoon
winds
shift
to
the
east
again.
A­
10
This
provides
evidence
of
the
wind
directions
one
should
anticipate
seeing
at
the
monitors
when
scrutinizing
meteorological
data
for
candidate
episode
days.
However,
just
as
the
ozone
season
could
be
narrowed
to
the
two
periods
within
which
one
may
select
representative
episodes,
the
Hysplit
model
will
allow
a
refinement
to
the
description
of
wind
directions
and
speeds
beyond
the
monitored,
station­
specific
weather
data.

Regional
Modeling
Data
The
HYSPLIT
Model
The
Texas
Natural
Resource
Conservation
Commission
(
TNRCC)
recommends
back
trajectory
analysis
as
the
preferred
method
in
obtaining
data
necessary
to
track
air
parcels.
Given
a
final
geographic
destination
for
an
air
parcel,
back
trajectories
show
the
path
followed
by
the
parcel
before
reaching
the
destination.
Theoretically,
back
trajectories
effectively
track
air
displacement
over
time,
distance,
and,
consequently,
over
emission
source
areas.

The
TNRCC
recommends
use
of
the
HYSPLIT
model
to
develop
back
trajectories.
The
Air
Resources
Laboratory
of
the
National
Oceanic
and
Atmospheric
Administration
(
NOAA)
maintains
the
HYSPLIT
model.
It
is
available
for
public
use
on
the
Internet
at
their
Realtime
Environmental
Applications
and
Display
sYstem
(
READY)
webpage8.
This
versatile
model
can
be
run
either
as
a
trajectory
(
parcel
displacement)
or
air
dispersion
model,
using
either
forecast
or
archived
meteorological
data.
The
necessary
data
for
creating
the
back
trajectories
used
in
this
conceptual
model
development
is
linked
to
the
online
model.
Point
and
click
operation
of
the
online
model
requires
minimal
data
input
by
the
user.
While
the
meteorological
database
is
not
inexhaustible,
the
model
and
database
is
applicable
across
the
United
States,
which
provides
a
national
reference
for
air
trajectory
and
dispersion
modeling
needs.

Regional
Wind
Patterns:
Back
Trajectories
Earlier,
the
list
of
exceedance
days
was
used
to
identify
the
annual
periods
during
which
ozone
exceedance
days
frequently
occurred,
on
a
seasonal
basis.
That
is,
temporal
patterns
were
identified
for
exceedance
days.
The
HYSPLIT
model
is
first
used
to
estimate
air
parcel
paths
typical
to
ozone
exceedance
days.
By
running
back
trajectories
for
thirty­
two
of
forty
exceedance
days
in
the
San
Antonio
area
from
1993
to
1996,
TNRCC
staff
identified
spatial
patterns
for
exceedance
days,
shown
in
Figure
A­
4.

8
READY
Homepage:
http://
www.
arl.
noaa.
gov/
ready.
html
.
Online
August
3,
2000.
A­
11
Figure
A­
4.
Typification
of
Air
Parcel
Paths
Arriving
in
San
Antonio,
Ozone
Exceedance
Days
1993
­
1998
Figure
A­
4
shows
the
pattern
of
air
parcel
positions
on
their
path
to
the
San
Antonio
International
Airport.
The
HYSPLIT
model
produces
air
parcel
positions
for
every
hour
in
the
model
run
by
latitude
and
longitude.
Figure
4
shows
that,
on
high
ozone
days,
it
is
rare
that
air
arriving
in
San
Antonio
will
have
come
from
the
northwest
or
the
southwest.

A
quantitative
refinement
of
the
above
data
is
presented
next.
In
Figure
A­
5,
the
air
parcel
back
trajectory
locations
have
been
sorted
into
bins
and
counted.
More
specifically,
the
region
of
central
Texas
within
a
250
mile
radius
of
the
San
Antonio
International
Airport
(
SAIA)
has
been
partitioned
into
octants;
northern,
northeastern,
eastern,
southeastern,
etc.
Then,
the
region
has
been
further
subdivided
by
distance
boundaries;
area
within
50
miles
of
SAIA,
50
to
100
miles
of
SAIA,
etc.,
out
to
250
miles
from
SAIA.
Next,
a
count
of
the
air
parcel
locations
that
fall
in
each
bin
were
made,
as
they
are
given
in
the
HYSPLIT
model
output
files.
Finally,
these
raw
counts
were
converted
into
percentages
and
written
into
the
representative
bins.
Note
that
the
percentages
in
bold
font
outside
of
the
250
mile
boundary
are
sums
of
the
percentages
within
the
octant.
That
is,
for
example,
the
image
shows
that
3.5%
of
the
air
parcel
passed
to
the
west
and
within
50
miles
of
SAIA;
0.6%
passed
to
the
west
and
between
A­
12
50
and
100
miles
of
SAIA.
Due
west
of
SAIA,
outside
the
250
mile
boundary,
the
figure
in
bold,
4.1%,
indicates
the
sum
of
all
air
parcels
that
passed
to
the
west
of
SAIA
within
the
western
octant.

Figure
A­
5.
Back
Trajectories
Percentages
by
Direction
for
High
Ozone
Days,
1993­
1998
This
is
extremely
valuable
information.
Just
as
the
exceedance
day
list
was
used
to
identify
the
temporal
occurrence
of
exceedance
days,
this
calculation
shows
clearly
how
frequently
air
parcels
passed
through
a
given
region,
by
distance
and
direction
(
octant),
before
coming
to
San
Antonio
on
a
high
ozone
day.
Industrial
(
point)
sources
can
be
identified
within
the
zones
delineated
in
the
image.
Figure
A­
6
presents
NOx
Point
Sources
in
the
Eastern
Half
of
Texas
by
their
distance,
magnitude
and
direction
from
San
Antonio.
N
A­
13
Figure
A­
6.
NOx
Point
Sources
in
the
Eastern
Half
of
Texas
by
their
distance,
magnitude
and
direction
from
San
Antonio.

Now
that
the
seasonal
time
periods
and
typical
air
movements
prior
to
ozone
exceedances
in
the
San
Antonio
region
have
been
identified,
the
exceedance
day
information
must
be
reviewed.
The
preceding
statistical
work
has
allowed
identification
of
particular
meteorological
parameters
which
candidate
episodes
must
fulfill.
Episodes
must
have
winds
from
the
south,
southeast,
east
and
northeast.
Episodes
should
be
chosen
from
the
two
annual
periods
for
exceedances:
May
to
early
July,
and
late
August
to
late
September.
Depending
on
episode
selection
availability,
air
parcels
traveling
through
the
eastern
octant,
where
some
of
the
larger
point
sources
are
found,
may
A­
14
weigh
favorably
on
episode
selection.
Next,
the
exceedance
day
data
will
be
reviewed
for
formation
of
candidate
episodes.

EPISODE
SELECTION
Episode
Candidates:
Exceedance
Days
San
Antonio
does
not
have
many
episode
candidates,
simply
because
San
Antonio
ozone
levels
are
not
typically
excessive.
The
following
table
(
A­
2)
lists
all
eight­
hour
ozone
exceedance
days
recorded
in
San
Antonio
for
ozone
seasons
1995
through
1999.
While
the
one­
hour
high
values
for
the
same
days
are
listed,
not
every
eight­
hour
exceedance
day
is
a
one­
hour
exceedance
day.
In
fact,
only
three
one­
hour
exceedances
(
one
of
which
was
excused
by
EPA)
exist
on
these
records.
Every
onehour
exceedance
is
listed.

The
years
1995­
1999
alone
are
listed,
since
earlier
years
are
not
considered
feasible
for
emission
inventory
and
photochemical
modeling
development.
A
preference
is
placed
on
modeling
1997
and
more
recent
years,
since
these
are
the
years
during
which
the
8­
hour
ozone
NAAQS
has
been
in
effect.
Note
also
that
the
column
heading
"
Episode
Dates"
refers
either
to
existing
modeling
episode
dates
­­
in
which
case
ramp­
up
days
are
included
in
the
episode
dates
listed
­­
or
refers
to
the
episode
candidate
period
marked
exclusively
by
exceedance
days.
In
the
latter
case,
ramp­
up
days,
which
are
negotiable
but
are
not
part
of
the
analysis
considered
here,
are
not
included
in
the
episode
date
period
listed.

Table
A­
2.
1995­
1999
Ozone
Exceedances
and
Possible
Modeling
Episodes
for
the
AACOG
Region:
Ozone
Readings
from
San
Antonio
Region
Monitors
1995
Ozone
Exceedance
Days
1
Hour
8
Hour
Episode
Dates
Notes
6/
13/
95
105
96
6/
21/
95
100
93
6/
22/
95
97
85
June
18­
22
Existing
UT
Episode
6/
23/
95
111
89
6/
27/
95
105
86
7/
8/
95
109
87
7/
9/
95
99
87
7/
11/
95
109
86
July
7­
12
Existing
AACOG
Episode
9/
3/
95
120
104
August
31­
September
3
Existing
TNRCC
Episode
9/
9/
95
105
94
9/
10/
95
108
91
9/
25/
95
119
108
9/
26/
95
122
101
10/
10/
95
108
90
1996
Ozone
Exceedance
Days
1
Hour
8
Hour
Episode
Dates
Notes
6/
3/
96
130
97
7/
3/
96
106
89
No
Modeling
Episode
Not
sufficient
ozone
exceedances
for
a
1996
modeling
episode
A­
15
1997
Ozone
Exceedance
Days
1
Hour
8
Hour
Episode
Dates
Notes
7/
16/
97
123
95
8/
26/
97
103
95
9/
6/
97
100
88
No
Modeling
Episode
Not
sufficient
ozone
exceedances
for
a
1997
modeling
episode
1998
Ozone
Exceedance
Days
1
Hour
8
Hour
Episode
Dates
Notes
5/
7/
98
140
101
5/
10/
98
107
89
No
Modeling
Episode
Mexican
Forest
Fires;
excused
by
EPA
8/
28/
98
99
89
8/
30/
98
99
92
9/
3/
98
105
87
August
28
­
September
3
9/
4/
98
141
110
9/
16/
98
107
91
10/
9/
98
121
95
No
Modeling
Episode
1999
Ozone
Exceedance
Days
1
Hour
8
Hour
Episode
Dates
Notes
8/
5/
99
120
100
8/
16/
99
109
87
8/
21/
99
109
87
August
16­
21
8/
30/
99
101
85
8/
31/
99
108
95
9/
1/
99
109
91
August
30­
September
1
9/
16/
99
93
85
9/
18/
99
108
96
9/
19/
99
96
91
9/
20/
99
107
86
September
16
 
20
10/
1/
99
99
88
Requirements
Limiting
Episode
Selection
One
criterion
for
episode
selection
is
that
there
be
more
than
two
exceedance
days
in
the
episode.
In
all,
the
episode
should
be
between
three
to
ten
days9
or
so.
Due
to
the
expense
and
time
required
to
model
episodes,
it
is
not
practical
to
model
all
episode
days.
With
these
introductory
guidelines
in
mind,
reinforced
by
the
desirability
of
developing
1997­
1999
episodes,
there
are
no
episode
candidates
in
1995,
1996
or
1997,
although
June
21­
23,
1995
are
three
exceedance
days
for
San
Antonio.
This
is
not
a
strong
candidate
since
the
period
is
in
1995.
This
consideration
tends
to
exclude
August
16­
21,
1999
in
which
only
two
exceedance
days
appear.

Note
also
that
on
September
4,
1998,
a
one­
hour
ozone
high
of
141
ppb
and
an
eighthour
ozone
high
of
110
were
recorded
(
both
at
CAMS
58).
This
day
comes
at
the
end
of
an
episode
candidate,
the
August
28
­
September
3,
1998
period.
Exceedance
days
within
10
ppb
above
the
design
value
of
88
ppb
are
preferred.
The
September
4th
8­
hour
9
Page
2,
"
Development
of
a
Conceptual
Model
for
Episode
Selection
of
High
Eight­
Hour
Ozone
Events
in
the
Dallas
/
Fort
Worth
area,"
C.
Durrenberger,
P.
Breitenbach,
J.
Red,
D.
Sullivan,
S.
Minto,
TNRCC
A­
16
average
value
of
110
ppb
is
22
ppb
above
the
design
value.
This
high
value
tends
to
exclude
September
4th
from
consideration
as
part
of
the
August
28
­
September
3
episode.
The
high
values
for
both
the
one­
and
eight­
hour
averages
recorded
on
September
4th
represent
an
anomaly
and
are
the
highest
in
this
record
set.
On
the
other
hand,
if
the
August
28
­
September
3,
1998
period
is
valuable
as
a
modeling
episode,
it
is
within
the
modeler's
discretion
to
note
the
anomaly
and
choose
to
include
September
4
in
the
episode.
It
may
be
of
interest
to
note
the
response
of
the
model
on
this
day
to
the
control
strategies
considered.
Also,
since
this
exceedance
comes
as
the
very
last
day
in
an
episode,
there
is
no
following
exceedance
day.
There
might
be
conceivably
be
debate
about
the
value
of
an
episode
with
such
an
anomaly
surrounded
by
other
exceedance
days.

A
restriction
on
episode
selection
has
arisen
since
some
meteorological
data
is
missing.
At
this
time,
Eta
Data
Assimilation
System
(
EDAS)
data
is
unavailable
to
create
the
meteorological
modeling
required
for
the
August
30
­
September
1,
1999
candidate
episode10.
There
are
no
current
plans
to
replace
this
data.
Coarse
grid
data
exists
which
may
be
used
to
interpolate
the
missing
data,
but
this
is
not
a
preferred
methodology
and
would
require
additional
expense.
The
August
30
­
September
1,
1999
candidate
episode
is
effectively
removed
from
consideration
for
this
time.

Other
data
is
missing
from
the
same
data
set.
Due
to
a
fire
at
the
National
Weather
Service's
National
Centers
for
Environmental
Prediction
(
NCEP)
computer
facility
at
the
end
of
September
1999,
EDAS
data
was
not
produced
between
October
1
and
November
4,
1999.
There
currently
are
no
plans
to
reproduce
the
missing
EDAS
data11.
EDAS
data
is
used
both
by
the
HYSPLIT
model
and
to
create
the
meteorological
input
files
to
the
photochemical
model.

With
the
above
considerations
in
mind,
August
16­
21,
1999
and
August
30­
September
1,
1999
are
excluded.
A
list
of
two
candidate
episodes
can
be
defined.
They
are
presented
in
Table
A­
3
below.

None
of
the
one­
hour
daily
highs
are
within
15
ppb
of
the
one­
hour
threshold
of
125
ppb.
The
highest
of
the
eight­
hour
averaged
daily
high
values,
96
ppb
for
September
18,
1999,
is
eight
ppb
above
the
design
value
of
88
ppb.
That
is,
none
of
the
values
in
the
remaining
candidate
episodes
are
excessively
elevated,
but
rather
are
near
the
design
value.
This
is
fortunate,
since
very
elevated
daily
high
values
are
anomalies
to
be
avoided,
as
are
the
rare
events
previously
discussed.
These
eight­
hour
values
are
within
10
ppb
above
the
design
value,
as
discussed
in
Section
2.0.

10
EDAS
data
is
missing
for
August
29
at
15Z
­
21Z
and
August
30
at
00Z.
Conversation
with
Pete
Brietenbach
and
Shannon
Minto,
TNRCC,
August
15,
2000.
11
According
to
"
HYSPLIT4
ARCHIVE
TRAJECTORIES,"
http://
www.
arl.
noaa.
gov/
ready­
bin/
traj1file.
pl
A­
17
Table
A­
3.
Candidate
Ozone
Episodes
1998
Ozone
Exceedances
1
Hour
Daily
High
8
Hour
Daily
High
Modeling
Episode
Dates
w/
o
Ramp
Up
#
Days
Required
to
model,
including
Startto
End
Dates
Modeling
Episode
Dates
with
3
Ramp
Up
Days
#
Days
Required
to
model
8/
28/
98
99
89
8/
30/
98
99
92
9/
3/
98
105
87
August
28
­
September
3
7
August
25
­
September
3
10
1999
Ozone
Exceedances
9/
16/
99
93
85
9/
18/
99
108
96
9/
19/
99
96
91
9/
20/
99
107
86
September
16
­
20
5
September
13
­
20
8
A
closely­
related
consideration
is
the
relationship
between
high
one­
hour
averages
and
high
eight­
hour
averages.
The
following
figure,
Figure
A­
7,
shows
the
close
correlation
between
high
one­
and
eight­
hour
values
for
1990
through
1999.
This
graph
was
prepared
by
TNRCC
staff
meteorologists.

Figure
A­
7.
San
Antonio
Area
Maximum
Ozone
1990­
1999
Best­
Fit
Line:
y
=
0.8318x
+
0.667;
R2
=
0.9353
0
25
50
75
100
125
0
25
50
75
100
125
150
Area
1­
Hour
Maximum
(
ppb)
Area
8­
hour
Maximum
(
ppb)

Unhealthy
for
Sensitive
Groups
Unhealthy
Very
Unhealthy
Projected
Unhealthy
Projected
Unhealthy
Sensitive
Projected
Very
Unhealthy
The
best­
fit
line
for
the
data
has
an
R2
value
of
0.9353,
an
indication
that
the
correlation
between
one­
and
eight­
hour
data
is
generally
high.
The
table
below,
Table
A­
4,
gives
A­
18
the
comparison
between
the
best­
fit
line
equation
and
the
observed
values
recorded
in
Table
A­
3.

Table
A­
4.
Observed
and
Predicted
values
correlated
with
Best
Fit
line,
1990­
1999
Observed
1
Hour
Daily
High
Observed
8
Hour
Daily
High
Predicted
8­
hour
high,
based
on
y=
0.8318*
x
+
0.667
Observed
­
Predicted
1995
July
8,
9,
11,
1995
­
Existing
AACOG
Photochem.
Model
Episode
Exceedance
Days
7/
8/
95
109
87
91.3332
­
4.332
7/
9/
95
99
87
83.0152
3.9848
7/
11/
95
109
86
91.3332
­
5.3332
1998
August
28
­
September
3,
1998
8/
28/
98
99
89
83.0152
5.9848
8/
30/
98
99
92
83.0152
8.9848
9/
03/
98
105
87
88.006
­
1.006
9/
04/
98
141
110
117.9508
­
7.9508
1999
September
16
­
20,
1999
9/
16/
99
93
85
78.0244
6.9756
9/
18/
99
108
96
90.5014
5.4986
9/
19/
99
96
91
80.5198
10.4802
9/
20/
99
107
86
89.6696
­
3.6696
The
correlation
between
the
best
fit
line
itself
and
the
one­
and
eight­
hour
observed
values
is
poorest
for
September
19,
1999
and
August
30,
1998.
Underlying
these
differences
is
an
unusually­
long
sustained
ozone
level,
near
in
value
to
the
day's
highest
one­
hour
value,
during
the
eight
hours
used
in
the
eight­
hour
averaging
period
for
that
day.
This
lifts
the
day's
corresponding
eight­
hour
averages
above
the
best­
fit
line,
that
is,
nearer
the
numerically
higher
one­
hour
value
itself,
on
those
two
days.

In
contrast,
the
difference
(
negative)
between
observed
and
predicted
values
on
September
20,
1999,
show
that
the
one­
hour
value
"
spiked;"
the
low
eight­
hour
average
of
86
ppb
required
much
lower
values
in
the
set
to
bring
the
one­
hour
high
of
107
ppb
down
in
the
average.
Yet,
as
can
be
seen
from
the
data
within
the
existing
AACOG
photochemical
model
episode,
July
11,
1995
shows
an
even
more
pronounced
"
spike."
(
The
September
4,
1998
values,
included
here
in
italics,
for
the
sake
of
comparison,
show
this
relationship.)

Judging
from
the
acceptability
of
the
existing
1995
modeling
episode,
the
September
20,
1999
value
is
acceptable
as
well.
In
all,
the
differences
reported
above
are
not
seen
as
valid
reasons
for
disqualifying
the
1998
and
1999
episode
days
from
further
consideration.

Staff
produced
one­
hour
versus
eight­
hour
plots,
like
Figure
5,
for
1995,
1996,
1998,
and
1999.
The
results
are
provided
in
Figures
A­
8
through
A­
11,
respectively.
A­
19
Figure
A­
8.
1995
San
Antonio
1hr
­
8hr
Ozone
Regression
R
2
=
0.9464
0
30
60
90
120
0
30
60
90
120
150
1
Hour
Max
Ozone
(
ppb)
8
H
o
u
r
O
z
o
n
e
(
p
p
b
)

June
18­
22
(
UT
Episode)
July
7­
12
(
AACOG
Episode)
August
31­
September
3
(
TNRCC
Episode)
8­
hr
Ozone
Level
(
85
ppb)

1­
hr
Ozone
Level
(
125
ppb)

Figure
A­
9.
1996
San
Antonio
1hr
­
8hr
Ozone
Regression
R
2
=
0.9298
0
30
60
90
120
0
30
60
90
120
150
1
Hour
Max
Ozone
(
ppb)
8
H
o
u
r
O
z
o
n
e
(
p
p
b
)

June
30­
July
5
(
TNRCC
Episode)
8­
hr
Ozone
Level
(
85
ppb)

1­
hr
Ozone
Level
(
125
ppb)
A­
20
Figure
A­
10.
1998
San
Antonio
1hr
­
8hr
Ozone
Regression
R2
=
0.9435
0
30
60
90
120
0
30
60
90
120
150
1
Hour
Max
Ozone
(
ppb)
8
H
o
u
r
O
z
o
n
e
(
p
p
b
)

August
25­
September
3
1­
hr
Ozone
Level
(
125
ppb)
8­
hr
Ozone
Level
(
85
ppb)

Figure
A­
11.
1999
San
Antonio
1hr
­
8hr
Ozone
Regression
R2
=
0.9344
0
30
60
90
120
0
30
60
90
120
150
1
Hour
Max
Ozone
(
ppb)
8
H
o
u
r
O
z
o
n
e
(
p
p
b
)

August
13­
21
August
27­
September
1
September
13­
19
8­
hr
Ozone
Level
(
85
ppb)

1­
hr
Ozone
Level
(
125
ppb)
A­
21
Comparison
of
Back
Trajectories
The
following
comparison
between
the
Back
Trajectories
for
the
1993­
1998
High
Ozone
Days,
the
existing
1995
Photochemical
Model
Episode,
and
the
1998
&
1999
Episode
Candidates
begins
the
final
analysis
of
the
wind
trajectories.
Figure
5
shows
the
direction
(
by
octant)
and
distance
of
the
air
parcel
trajectories
from
1993
through
1998
for
all
high
ozone
days.
The
distribution
of
air
parcel
locations
in
the
northeast,
east,
southeast
and
southern
octants
represents
83.6%
of
the
total
high
ozone
day
air
parcel
locations
for
the
entire
six
year
period.
In
a
sense,
the
goal
for
this
section
of
the
conceptual
model
development
is
to
incorporate
back
trajectories
for
the
exceedance
days
found
in
the
existing
1995
photochemical
model
together
with
one
or
more
of
the
candidate
episodes
such
that
the
resulting
combination
generally
matches
the
distribution
of
the
1993­
1998
back
trajectory
set.

Each
exceedance
day
of
the
two
candidate
episodes
(
not
including
ramp­
up
days),
August
28,
30
and
September
3,
1998,
and
September
16,
18
­
20,
1999,
was
run
through
the
Hysplit
model
to
determine
a
back
trajectory.
In
all
cases,
the
back
trajectory
ended
at
the
San
Antonio
International
Airport
at
21
UTC
(
4
p.
m.,
Central
Standard,
Daylight
Savings
time),
with
the
exception
of
one
day:
August
31,
1998.
The
back
trajectory
ended
at
12
UTC
(
7
a.
m.)
instead.
This
was
necessary
due
to
missing
EDAS
data
required
in
the
Hysplit
model.
For
the
same
reasons,
all
back
trajectories
were
33
hours
except
for
August
28,
1998
(
19
hours),
and
September
3,
1998
(
19
hours).

By
entering
the
data
from
the
Hysplit
model
as
input
to
GIS
software
(
ARC
INFO),
the
following
graphs
were
developed.
They
indicate
air
parcel
positions
for
each
of
the
episodes.
Graphs
showing
both
raw
position
counts
given
by
the
Hysplit
model
and
percentages
of
the
total
are
given.
A­
22
Back
Trajectories
for
August
28,
30
and
September
3,
1998
A­
23
Back
Trajectories
for
September
16,
18,
19
and
20,
1999
A­
24
In
addition,
the
back
trajectories
were
calculated
for
the
AACOG
Photochemical
model
days
currently
used
for
analysis,
July
10
­
12,
1995.
The
resulting
analysis
is
contained
in
the
graphs
below.

Back
Trajectories
for
July
10
­
12,
1995
Note
that
the
air
parcel
trajectories
associated
with
the
1995
AACOG
Photochemical
Modeling
episode
are
essentially
(
75%)
all
within
the
southern
or
southwestern
octant.
As
shown
in
Figure
A­
5,
very
few
trajectory
positions
in
the
1993­
1998
trajectory
sum
A­
25
were
in
the
southwestern
octant;
the
southwestern
octant
need
not
be
well­
represented
by
the
modeling
episodes.
Ideally,
a
second
modeling
episode
candidate
would
provide
the
missing
back
trajectories
required
in
Figure
A­
5
for
the
southeastern,
eastern
and
northeastern
octants.
Back
trajectories
filling
the
northeast,
east,
and
southeast
octants
are
required
in
a
16.7/
28.4/
19.4
ratio,
or,
roughly,
an
equivalent
ratio
of
26/
44/
30.
The
following
table,
A­
5,
presents
a
comparison
of
the
counts
and
percentages
in
these
three
quadrants,
by
episode.

Table
A­
5.
Comparison
of
Back
Trajectories
without
1995
AACOG
episode
#,
%
of
all
Trajectory
Points
in
NE,
E,
SE
%
Trajectory
Points
in
Northeast
%
Trajectory
Points
in
East
%
Trajectory
Points
in
Southeast
Ratio
NE/
E/
SE
1993­
1998
Back
Trajectories
565,
64.5%
16.7%
28.4%
19.4%
26
/
44
/
30
Aug
28,
30
Sept
3,
1998
146,
70.1%
56.5%
13%
1%
80
/
19
/
1
Sept
16,
18
­
20,
1999
280,
78.9%
8.5%
26.5%
43.9%
11
/
36
/
56
The
1998
scenario
contains
many
trajectory
points
in
the
northeast
octant.
The
September
16
­
20,
1999
scenario
ratio
seems
better
balanced.

The
next
table,
A­
6,
presents
a
similar
comparison
of
the
four
major
octants
which
must
be
accounted
for.
Here,
the
candidate
episode
trajectories
are
combined
with
the
Base
Case
1995
AACOG
Photochemical
Model
(
BC)
to
compare
with
the
1993
­
1998
trajectory
data.
Because
the
attainment
modeling
will
use
both
episodes,
the
1995
Base
Case
plus
at
least
one
additional
episode,
for
demonstration
purposes,
this
shows
the
combined
effectiveness
of
multiple
episode
considerations.

Table
A­
6.
Comparison
of
Back
Trajectories
with
1995
AACOG
episode
#,
%
of
all
Trajectory
Points
in
NE,
E,
SE,
S
%
Trajectory
Points
in
Northeast
%
Trajectory
Points
in
East
%
Trajectory
Points
in
Southeast
%
Trajectory
Points
in
South
Ratio
NE/
E/
SE/
S
1993­
1998
Back
Trajectories
732,
83.6%
16.7%
28.4%
19.4%
19.1%
20
/
34
/
23
/
23
Aug
28,
30
Sept
3,
1998
+
BC
337,
70.5%
25.3%
9.2%
5.3%
30.8%
36
/
13
/
8
/
44
Sept
16,
18
­
20,
1999
+
BC
524,
83.7%
5.4%
17.7%
28.6%
32%
7
/
21
/
34
/
38
Table
A­
6
shows
that
both
combined
episodes
cover
the
four
essential
octants.
This
same
data
is
presented
in
more
detail
in
tables
A­
7
through
A­
9
below.
A­
26
Table
A­
7.
1993
­
1998
Ozone
Exceedance
Days
for
San
Antonio
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
13
0
1
2
2
18
2.05%
Northeast
36
32
30
38
10
146
16.67%
East
104
37
75
32
1
249
28.42%
Southeast
60
35
36
26
13
170
19.41%
South
41
39
52
32
3
167
19.06%
Southwest
12
21
3
1
0
37
4.22%
West
31
5
0
0
0
36
4.11%
Northwest
44
3
2
2
2
53
6.05%

341
172
199
133
31
<=
Counts
38.93%
19.63%
22.72%
15.18%
3.54%
<=
Percent
Table
A­
8.
Base
Case
(
July
10,
11,
and
12,
1995)
plus
Aug
28,
30
Sept
3,
1998
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
18
0
0
0
0
18
3.77%
Northeast
25
25
31
23
17
121
25.31%
East
35
8
1
0
0
44
9.21%
Southeast
14
10
1
0
0
25
5.23%
South
32
8
89
17
1
147
30.75%
Southwest
37
27
10
0
0
74
15.48%
West
27
15
0
0
0
42
8.79%
Northwest
7
0
0
0
0
7
1.46%

195
93
132
40
18
<=
Counts
40.79%
19.46%
27.62%
8.37%
3.77%
<=
Percent
Table
A­
9.
Base
Case
(
July
10,
11,
and
12,
1995)
plus
Sept
16,
18
­
20,
1999
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
8
0
0
0
0
8
1.28%
Northeast
4
8
4
5
13
34
5.43%
East
41
28
19
12
11
111
17.73%
Southeast
71
64
39
5
0
179
28.59%
South
64
23
95
17
1
200
31.95%
Southwest
38
27
10
0
0
75
11.98%
West
13
0
0
0
0
13
2.08%
Northwest
6
0
0
0
0
6
0.96%

245
150
167
39
25
<=
Counts
39.14%
23.96%
26.68%
6.23%
3.99%
<=
Percent
These
tables
show
that
the
combined
episode
using
August
28,
30
and
September
3,
1998,
weighs
heavily
in
the
Northeast
octant
(
25.3%
compared
to
16.7%
in
the
1993
­
1998
set,
a
difference
of
8.6%),
while
the
combined
episode
using
September
16,
18­
20,
1999
is
very
light
(
5.4%
to
16.7%
­­
a
difference
of
11.3%)
in
the
same
octant.
Both
combined
episodes
weigh
heavily
in
the
Southern
octant
(~
31%
to
19%).
However,
the
1998
episode
is
very
light
in
the
East
(
9.2%
to
28.4%
­­
a
difference
of
19.2%)
and
the
Southeast
(
5.2%
to
19.4%
­­
a
difference
of
14.2%).
The
1999
episode
is
better
balanced
in
the
East
(
17.7%
to
28.4%
­­
a
difference
of
10.7%)
and
the
Southeast
(
28.6%
to
19.4%
­­
a
difference
of
9.2%).
In
this
light,
the
September
16,
18­
20,
1999
episode,
combined
with
the
Base
Case
set,
does
represent
the
required
back
A­
27
trajectories
marginally
better
than
does
the
August
28,
30
and
September
3,
1998
combined
set.

Figure
A­
6
shows
that
large
NOx
Point
Sources
exist
to
the
east
of
San
Antonio.
Many
are
based
in
the
Houston
/
Galveston
area.
This
also
argues
in
favor
of
the
September
16,
18­
20,
1999
episode,
which
shows
a
greater
percentage
of
eastern
trajectories
than
does
the
August
28,
30
and
September
3,
1998
trajectory
set.

Combining
the
Base
Case
episode
with
the
1998
and
1999
episodes
described
above
gives
a
good
coverage
of
the
required
back
trajectories.
This
is
shown
in
table
A­
10
below.
Compare
these
values
with
those
in
Table
A­
7.

Table
A­
10.
Base
Case
(
July
10,
11,
and
12,
1995),
Sept
16,
18
­
20,
1999
and
Aug
28,
30
Sept
3,
1998
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
18
0
0
0
0
18
2.16%
Northeast
25
33
35
28
30
151
18.13%
East
68
28
19
12
11
138
16.57%
Southeast
72
65
39
5
0
181
21.73%
South
74
23
95
17
1
210
25.21%
Southwest
46
27
10
0
0
83
9.96%
West
30
15
0
0
0
45
5.40%
Northwest
7
0
0
0
0
7
0.84%

340
191
198
62
42
<=
Counts
40.82%
22.93%
23.77%
7.44%
5.04%
<=
Percent
The
conclusion
for
this
chapter
is
that,
in
consideration
of
the
back
trajectories
provided
by
the
HYSPLIT
model
and
the
point
source
locations,
the
September
16,
18­
20,
1999
will
be
a
preferable
episode
to
model
next.
However,
the
same
data
set
strongly
argues
for
another
episode,
the
August
28,
30
and
September
3,
1998
period
to
follow.
Recall
that,
according
to
Table
A­
3,
the
1999
episode
also
would
require
a
minumum
of
five
days
to
model
(
excluding
ramp­
up
days),
while
the
1998
episode
would
require
seven
days
to
model
(
also
excluding
ramp­
up
days).
Thus
the
1999
episode
might
be
less
expensive
to
model.
A
detailed
trajectory
count,
both
by
combined
and
uncombined
Hysplit
runs,
is
provided
in
Table
A­
11.
A­
28
Table
A­
11.
Total
and
Episode­
specific
Trajectory
Counts.
1993
­
1998
Ozone
Exceedance
Days
for
SA
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
13
0
1
2
2
18
2.05%
Northeast
36
32
30
38
10
146
16.67%
East
104
37
75
32
1
249
28.42%
Southeast
60
35
36
26
13
170
19.41%
South
41
39
52
32
3
167
19.06%

Southwest
12
21
3
1
0
37
4.22%
West
31
5
0
0
0
36
4.11%
Northwest
44
3
2
2
2
53
6.05%

341
172
199
133
31
<=
Counts
38.93%
19.63%
22.72%
15.18%
3.54%
<=
Percent
Base
Case
­­
July
10,
11,
and
12,
1995
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
8
0
0
0
0
8
2.95%
Northeast
4
0
0
0
0
4
1.48%
East
8
8
1
0
0
17
6.27%
Southeast
13
9
1
0
0
23
8.49%
South
22
8
89
17
1
137
50.55%
Southwest
29
27
10
0
0
66
24.35%
West
10
0
0
0
0
10
3.69%
Northwest
6
0
0
0
0
6
2.21%

100
52
101
17
1
<=
Counts
36.90%
19.19%
37.27%
6.27%
0.37%
<=
Percent
August
28,
30
and
September
3,
1998
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
10
0
0
0
0
10
4.83%
Northeast
21
25
31
23
17
117
56.52%
East
27
0
0
0
0
27
13.04%
Southeast
1
1
0
0
0
2
0.97%
South
10
0
0
0
0
10
4.83%
Southwest
8
0
0
0
0
8
3.86%
West
17
15
0
0
0
32
15.46%
Northwest
1
0
0
0
0
1
0.48%

95
41
31
23
17
<=
Counts
45.89%
19.81%
14.98%
11.11%
8.21%
<=
Percent
A­
29
August
30
­
September
1,
1999
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
0
0
0
0
0
0
0.00%
Northeast
4
1
0
0
4
9
3.93%
East
51
43
14
7
0
115
50.22%
Southeast
60
25
18
2
0
105
45.85%
South
0
0
0
0
0
0
0.00%
Southwest
0
0
0
0
0
0
0.00%
West
0
0
0
0
0
0
0.00%
Northwest
0
0
0
0
0
0
0.00%

115
69
32
9
4
<=
Counts
50.22%
30.13%
13.97%
3.93%
1.75%
<=
Percent
September
16,
18
­
20,
1999
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
0
0
0
0
0
0
0.00%
Northeast
0
8
4
5
13
30
8.45%
East
33
20
18
12
11
94
26.48%
Southeast
58
55
38
5
0
156
43.94%
South
42
15
6
0
0
63
17.75%
Southwest
9
0
0
0
0
9
2.54%
West
3
0
0
0
0
3
0.85%
Northwest
0
0
0
0
0
0
0.00%

145
98
66
22
24
<=
Counts
40.85%
27.61%
18.59%
6.20%
6.76%
<=
Percent
A
comparison
may
be
made
between
the
1993
­
1998
ozone
exceedance
days
data
and
combinations
of
the
base
case
with
various
candidate
episode
data.
These
combinations
(
see
Table
A­
12)
help
demonstrate
that
the
back
trajectories
given
by
the
combination
of
base
case
and
various
episodes
will,
to
varying
degrees,
represent
the
back
trajectories
required
to
comprehensively
represent
the
1993
­
1998
ozone
exceedance
days
trajectories.
This
represents
one
goal
of
the
Conceptual
Modeling
exercise,
identification
of
likely
episodes
according
to
the
similarity
of
the
candidate
episode
conditions
­­
in
this
case,
wind
trajectories
­­
compared
to
composite
profile
of
high
ozone
days.
A­
30
Table
A­
12.
Back
Trajectories
Counts
and
Percentages,
Combined
Data
1993
­
1998
Ozone
Exceedance
Days
for
San
Antonio
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
13
0
1
2
2
18
2.05%
Northeast
36
32
30
38
10
146
16.67%
East
104
37
75
32
1
249
28.42%
Southeast
60
35
36
26
13
170
19.41%
South
41
39
52
32
3
167
19.06%
Southwest
12
21
3
1
0
37
4.22%
West
31
5
0
0
0
36
4.11%
Northwest
44
3
2
2
2
53
6.05%

341
172
199
133
31
<=
Counts
38.93%
19.63%
22.72%
15.18%
3.54%
<=
Percent
Base
Case
(
July
10,
11,
and
12,
1995)
plus
Aug
28,
30
Sept
3,
1998
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
18
0
0
0
0
18
3.77%
Northeast
25
25
31
23
17
121
25.31%
East
35
8
1
0
0
44
9.21%
Southeast
14
10
1
0
0
25
5.23%
South
32
8
89
17
1
147
30.75%
Southwest
37
27
10
0
0
74
15.48%
West
27
15
0
0
0
42
8.79%
Northwest
7
0
0
0
0
7
1.46%

195
93
132
40
18
<=
Counts
40.79%
19.46%
27.62%
8.37%
3.77%
<=
Percent
Base
Case
(
July
10,
11,
and
12,
1995)
plus
Sept
16,
18
­
20,
1999
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
8
0
0
0
0
8
1.28%
Northeast
4
8
4
5
13
34
5.43%
East
41
28
19
12
11
111
17.73%
Southeast
71
64
39
5
0
179
28.59%
South
64
23
95
17
1
200
31.95%
Southwest
38
27
10
0
0
75
11.98%
West
13
0
0
0
0
13
2.08%
Northwest
6
0
0
0
0
6
0.96%

245
150
167
39
25
<=
Counts
39.14%
23.96%
26.68%
6.23%
3.99%
<=
Percent
A­
31
Base
Case
(
July
10,
11,
and
12,
1995),
Sept
16,
18
­
20,
1999
and
Aug
28,
30
Sept
3,
1998
0
­
50
Miles
50
­
100
100
­
150
150
­
200
200
­
250
Counts
Percent
North
18
0
0
0
0
18
2.16%
Northeast
25
33
35
28
30
151
18.13%
East
68
28
19
12
11
138
16.57%
Southeast
72
65
39
5
0
181
21.73%
South
74
23
95
17
1
210
25.21%
Southwest
46
27
10
0
0
83
9.96%
West
30
15
0
0
0
45
5.40%
Northwest
7
0
0
0
0
7
0.84%

340
191
198
62
42
<=
Counts
40.82%
22.93%
23.77%
7.44%
5.04%
<=
Percent
Regional
Considerations
An
overview
of
a
data
set
pair
is
included
here
for
the
eastern
half
of
Texas.
A
list
of
exceedance
days
is
included
below
(
tables
A­
13
and
A­
14),
which
covers
the
August
25
­
September
4,
1998
and
the
September
13
­
20,
1999
periods.

Table
A­
13.
Regional
Considerations,
1998
Highest
Daily
Maximum
by
Area,
8­
hour
ozone
average
Exceedance
Days
for
San
Antonio
1998
Dallas­

Fort
Worth
Tyler
Longview­

Marshall
Beaumont­

Port
Arthur
Houston­
Galveston­

Brazoria
Austin
San
Antonio
Victoria
Corpus
Christi
Laredo
McAllen­
Edinburg­

Mission
Brownsville­

Harlingen
8/
25
79
63
63
45
72
44
30
23
23
24
25
19
8/
26
83
68
85
82
111
42
31
17
26
20
23
14
8/
27
90
84
104
87
132
62
56
51
42
29
28
16
X
8/
28
84
87
114
93
114
84
89
55
52
36
33
30
8/
29
84
83
96
87
149
88
84
72
75
52
38
31
X
8/
30
100
85
73
99
118
92
92
48
42
56
33
27
8/
31
96
78
82
46
72
84
74
60
49
33
31
26
9/
1
102
73
73
41
53
79
69
47
53
54
44
41
9/
2
120
99
86
62
75
74
76
62
82
55
64
56
X
9/
3
100
91
107
94
152
101
87
71
76
51
41
36
X
9/
4
92
90
96
97
128
95
110
78
78
50
59
63
A­
32
Table
A­
14.
Regional
Considerations,
1999
Highest
Daily
Maximum
by
Area,
8­
hour
ozone
average
Exceedance
Days
for
San
Antonio
1999
Dallas­
Fort
Worth
Tyler
Longview­
Marshall
Beaumont­
Port
Arthur
Houston­
Galveston­
Brazoria
Austin
San
Antonio
Victoria
Corpus
Christi
Laredo
McAllen­
Edinburg­
Mission
Brownsville­
Harlingen
9/
13
49
43
45
66
82
55
64
67
70
48
47
50
9/
14
69
60
58
64
88
62
66
67
76
61
50
54
9/
15
80
85
75
70
97
78
82
78
82
56
70
68
X
9/
16
78
82
79
89
104
85
85
79
81
65
68
66
9/
17
99
86
75
69
111
99
76
86
81
55
64
57
X
9/
18
99
91
86
101
98
99
96
87
89
61
71
66
X
9/
19
96
91
97
100
120
101
91
84
88
59
71
55
X
9/
20
92
99
110
79
124
87
86
99
75
53
65
54
The
near
non­
attainment
areas
for
the
8­
hour
ozone
NAAQS
in
Texas
are
Austin,
San
Antonio,
Tyler­
Longview­
Marshall,
Victoria
and
Corpus
Christi
for
the
1997­
1999
period.
If
a
combined,
regional
photochemical
model
was
to
be
undertaken
by
these
areas,
the
September
1999
episode
is
preferable,
in
that
there
are
no
8­
hour
exceedances
for
Corpus
Christi
or
Victoria
in
the
1998
data
set.

CONCLUSION
Historical
data
reveals
two
annual
periods
of
likely
high
ozone
exceedances:
May­
July
and
August­
October,
with
the
central
occurences
in
late
June
­
early
July
and
late
August
­
late
September.
The
1995
Base
Case
Photochemical
Model
now
with
AACOG
extends
from
July
7
August
28,
30
and
September
3,
1998
through
July
12th.
The
two
final
episode
candidates
are
represented
by
the
September
16,
18­
20,
1999
exceedance
days
or
the
August
28,
30
and
September
3,
1998
exceedance
days.
The
1995
Base
Case
and
either
of
the
two
final
candidates
represent
the
two
seasonal
periods.

The
wind
patterns
for
these
two
candidate
episodes
do
follow
the
patterns
of
direction
given
in
the
wind
roses
fairly
well.
According
to
the
wind
roses
for
high
ozone
days,
the
morning
winds
are
most
likely
to
come
from
the
south
to
southeast.
According
to
CAMS
23
(
Marshall
High)
data,
the
morning
winds
for
the
episodes
in
question
come
from
(
on
average)
the
south
or
southeast.
And
the
afternoon
winds
do
swing
to
a
more
easterly
direction.
CAMS
23
has
been
in
place
since
1996,
and
so
will
most
closely
match
the
1998­
1997
data
in
the
wind
roses.
Details
of
wind
speed,
resultant
wind
speed,
and
wind
direction
during
the
exceedance
days,
by
monitoring
station,
are
provided
in
tables
A­
15
through
A­
22.
Wind
direction,
the
direction
from
which
the
wind
is
blowing,
is
measured
to
the
nearest
degree
based
on
a
360
degree
compass
with
360
degrees
being
from
the
North
and
180
degrees
being
from
the
South.
A­
33
Table
A­
15.
Wind
Speed
and
Resultant
Wind
Speed
at
CAMS
58,
1998
and
1999
Episodes
Camp
Bullis,
CAMS
58
August
­
September
1998
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Wind
Speed
Average
(
mph)
8­
10:
59
a.
m.
Resultant
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Resultant
Wind
Speed
Average
(
mph)
28­
Aug­
98
82.25
4.10
3.63
5.78
4.68
30­
Aug­
98
87.50
5.13
4.65
5.53
4.80
03­
Sep­
98
87.38
3.88
3.30
3.68
2.75
04­
Sep­
98
110.25
3.13
2.35
4.43
3.20
Average
4.06
3.48
4.85
3.86
September
1999
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Wind
Speed
Average
(
mph)
8­
10:
59
a.
m.
Resultant
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Resultant
Wind
Speed
Average
(
mph)
16­
Sep­
99
78.25
5.10
4.48
7.95
6.98
18­
Sep­
99
96.75
4.03
3.05
6.08
4.50
19­
Sep­
99
91.13
3.73
2.80
5.43
3.85
20­
Sep­
99
81.88
4.23
3.50
5.73
3.83
Average
4.27
3.46
6.29
4.79
Table
A­
16.
Wind
Direction
at
CAMS
58,
1998
and
1999
Episodes
Camp
Bullis,
CAMS
58
August
­
September
1998
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Average
Resultant
Wind
Direction
1­
3:
59
p.
m.
Average
Resultant
Wind
Direction
28­
Aug­
98
82.25
327
109
30­
Aug­
98
87.50
122
120
03­
Sep­
98
87.38
279
201
04­
Sep­
98
110.25
253
160
Average
245
147
September
1999
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Average
Resultant
Wind
Direction
1­
3:
59
p.
m.
Average
Resultant
Wind
Direction
16­
Sep­
99
78.25
194
110
18­
Sep­
99
96.75
236
148
19­
Sep­
99
91.13
246
175
20­
Sep­
99
81.88
258
172
Average
233
151
A­
34
Table
A­
17.
Wind
Speed
and
Resultant
Wind
Speed
at
CAMS
678,
1999
Episode*
CPS/
Trinity
Pecan
Valley
C678
September
1999
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Wind
Speed
Average
(
mph)
8­
10:
59
a.
m.
Resultant
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Resultant
Wind
Speed
Average
(
mph)

16­
Sep­
99
74.00
3.93
3.35
5.58
4.80
18­
Sep­
99
76.25
2.70
1.70
3.88
2.55
19­
Sep­
99
84.88
2.55
1.88
3.85
1.93
20­
Sep­
99
86.88
4.38
3.85
3.53
1.98
Average
3.39
2.69
4.21
2.81
Table
A­
18.
Wind
Direction
at
CAMS
678,
1999
Episode*
CPS/
Trinity
Pecan
Valley
C678
September
1999
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Average
Resultant
Wind
Direction
8­
10:
59
p.
m.
Average
Resultant
Wind
Direction
16­
Sep­
99
74.00
129
90
18­
Sep­
99
76.25
205
102
19­
Sep­
99
84.88
241
147
20­
Sep­
99
86.88
248
157
Average
206
124
*
There
is
no
data
available
at
CPS/
Trinity
Pecan
Valley
C678
for
August
­
September,
1998.

Table
A­
19.
Wind
Speed
and
Resultant
Wind
Speed
at
CAMS
23,
1998
and
1999
Episodes
Marshall
High
C23
August
­
September
1998
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Wind
Speed
Average
(
mph)
8­
10:
59
a.
m.
Resultant
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m..
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Resultant
Wind
Speed
Average
(
mph)

28­
Aug­
98
89.13
2.98
2.33
6.13
5.55
30­
Aug­
98
90.88
5.28
4.93
5.45
4.98
03­
Sep­
98
76.00
5.28
3.78
4.88
2.40
04­
Sep­
98
93.50
3.13
2.70
4.35
2.98
Average
4.16
3.43
5.20
3.98
September
1999
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Wind
Speed
Average
(
mph)
8­
10:
59
a.
m.
Resultant
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m..
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Resultant
Wind
Speed
Average
(
mph)
16­
Sep­
99
85.63
5.85
5.25
7.23
6.45
18­
Sep­
99
92.88
3.73
2.55
4.83
3.23
19­
Sep­
99
89.88
3.58
2.70
4.33
2.60
20­
Sep­
99
84.63
5.08
4.43
4.60
2.80
Average
4.56
3.73
5.24
3.77
A­
35
Table
A­
20.
Wind
Direction
at
CAMS
23,
1998
and
1999
Episodes
Marshall
High
C23
August
­
September
1998
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Average
Resultant
Wind
Direction
1­
3:
59
p.
m.
Average
Resultant
Wind
Direction
28­
Aug­
98
89.13
282
101
30­
Aug­
98
90.88
55
108
03­
Sep­
98
76.00
248
215
04­
Sep­
98
93.50
183
141
Average
192
141
September
1999
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Average
Resultant
Wind
Direction
1­
3:
59
p.
m.
Average
Resultant
Wind
Direction
16­
Sep­
99
85.63
66
83
18­
Sep­
99
92.88
184
133
19­
Sep­
99
89.88
219
169
20­
Sep­
99
84.63
232
122
Average
175
126
Table
A­
21.
Wind
Speed
and
Resultant
Wind
Speed
at
CAMS
59,
1998
and
1999
Episodes
Calaveras
Lake
C59
August
­
September
1998
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Wind
Speed
Average
(
mph)
8­
10:
59
a.
m.
Resultant
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Resultant
Wind
Speed
Average
(
mph)

28­
Aug­
98
66.25
4.58
4.23
5.73
4.95
30­
Aug­
98
74.38
8.00
7.78
5.70
5.03
03­
Sep­
98
78.00
6.03
5.75
4.03
3.35
04­
Sep­
98
78.75
3.70
3.38
4.35
2.88
Average
5.58
5.28
4.95
4.05
September
1999
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Wind
Speed
Average
(
mph)
8­
10:
59
a.
m.
Resultant
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Wind
Speed
Average
(
mph)
1­
3:
59
p.
m.
Resultant
Wind
Speed
Average
(
mph)

16­
Sep­
99
90.75
6.68
6.38
9.33
8.63
18­
Sep­
99
89.13
4.75
4.15
5.05
2.53
19­
Sep­
99
98.88
3.53
2.40
5.40
3.78
20­
Sep­
99
88.88
5.08
4.63
4.00
3.05
Average
5.01
4.39
5.94
4.49
A­
36
Table
A­
22.
Wind
Direction
at
CAMS
59,
1998
and
1999
Episodes
Calaveras
Lake
C59
August
­
September
1998
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Average
Resultant
Wind
Direction
1­
3:
59
p.
m.
Average
Resultant
Wind
Direction
28­
Aug­
98
66.25
232
83
30­
Aug­
98
74.38
46
127
03­
Sep­
98
78.00
256
99
04­
Sep­
98
78.75
197
99
Average
183
102
September
1999
8­
hour
O3
high
(
ppb)
8­
10:
59
a.
m.
Average
Resultant
Wind
Direction
1­
3:
59
p.
m.
Average
Resultant
Wind
Direction
16­
Sep­
99
90.75
35
59
18­
Sep­
99
89.13
31
175
19­
Sep­
99
98.88
169
106
20­
Sep­
99
88.88
259
80
Average
123
105
While
the
complete
list
of
exceedance
days
showed
five
possible
episode
candidates,
several
factors
cut
the
list
down
to
two.
These
factors
are:
lack
of
meteorological
data;
less
than
three
ozone
exceedance
days
within
a
single
identifiable
period;
episode
sequence
was
not
found
during
the
1997
through
1999
ozone
season
period.

For
the
remaining
two
candidates,
September
16­
20,
1999,
or
August
28
­
September
3,
1998,
an
analysis
of
the
back
trajectories
provided
by
the
HYSPLIT
model
was
performed.
Comparing
the
back
trajectories
of
historical
ozone
exceedance
days
with
those
of
the
exceedance
days
from
each
episode
set,
the
September
16­
20,
1999
trajectory
set
more
closely
resembled
the
desired
balance
of
back
trajectories.
Moreover,
the
September
16­
20,
1999
set
showed
a
higher
percentage
of
back
trajectories
from
the
east,
a
valuable
asset,
given
the
large
number
of
NOx
emitters
found
within
250
miles
of
San
Antonio.
However,
a
combination
of
the
September
16­
20,
1999
and
the
August
28
­
September
3,
1998
episodes
well
filled
the
back
trajectory
requirements.

Finally,
if
regional
modeling
considerations
are
weighed,
the
September
16­
20,
1999
is
the
most
practical
next
modeling
episode
for
most
of
the
other
near
non­
attainment
areas
to
attempt.
Of
the
two
candidates,
exceedance
days
for
more
of
the
five
near
nonattainment
areas
occured
during
the
September
16­
20,
1999
period.
For
these
reasons,
the
September
16­
20,
1999
is
advised
as
the
next
photochemical
modeling
episode
to
be
undertaken
for
photochemical
modeling
in
the
San
Antonio
region.
