APPENDIX
E
1999
BASE
CASE
DEVELOPMENT
AND
PERFORMANCE
ANALYSES
SAN
ANTONIO
EAC
REGION
ATTAINMENT
DEMONSTRATION
MARCH
2004
E­
2
Appendix
E
Table
of
Contents
Page
Introduction
                              .
E­
9
Model
Inputs
                              
E­
11
Land
Use
Data
                         ..
E­
11
Dry
Deposition
Algorithms
                     
E­
11
Chemistry
Data
                         ..
E­
14
Boundary
and
Initial
Conditions
                  .
E­
14
CAMx
Model
Options
                       
E­
17
September
13­
20,
1999
Photochemical
Model
Performance
         .
E­
17
Model
Performance:
1­
Hour
Average
Ozone
Concentrations
     ..
E­
18
Ozone
Metrics
                      ..
E­
18
Time
Series
Plots
                     .
E­
32
Scatter
Plots
                       .
E­
35
Weekend/
Weekday
Comparisons
              .
E­
37
Model
Performance:
8­
Hour
Average
Ozone
Concentrations
      
E­
39
Ozone
Metrics
                      ..
E­
39
Graphic
Analyses
                     .
E­
70
Scatter/
Q­
Q
Plots                  ..
E­
70
Tile
Plots
                     ..
E­
74
Diagnostic
Evaluations
                   
E­
78
Zero­
out
Runs:
Urban
Areas
             .
E­
78
Incremental
Removal
of
VOC
and
NOx
Precursors
   ..
E­
79
Comparisons
between
UT
Austin
and
AACOG
1999
Base
Cases
       .
E­
83
References
                              ..
E­
85
E­
3
Appendix
E
List
of
Tables
Page
Table
E­
1
Meteorological
and
Emissions
Inputs
to
CAMx
Photochemical
Model
Test
Runs
                     ..
E­
12
Table
E­
2
Summary
of
Chemistry
Data
for
the
September
13­
20,
1999
CAMx
Model
                       .
E­
14
Table
E­
3
Boundary
and
Initial
Conditions
used
by
in
the
Original
Model
and
the
Final
Model
used
by
San
Antonio
and
Austin
for
their
Early
Action
Compacts
                     ..
E­
16
Table
E­
4
Summary
of
Options
for
the
September
13­
20,
1999
CAMx
Model..
E­
17
Table
E­
5
1­
hour
Statistics
for
CAMS
3,
September
13
 
20,
1999     
E­
20
Table
E­
6
1­
hour
Statistics
for
CAMS
38,
September
13
 
20,
1999
    
E­
20
Table
E­
7
1­
hour
Statistics
for
Austin
Monitors
(
CAMS
3
&
38),
September
13
 
20,
1999
                       .
E­
20
Table
E­
8
1­
hour
Statistics
for
CAMS
23,
September
13
 
20,
1999
    .
E­
20
Table
E­
9
1­
hour
Statistics
for
CAMS
58,
September
13
 
20,
1999
    .
E­
21
Table
E­
10
1­
hour
Statistics
for
CAMS
59,
September
13
 
20,
1999
    .
E­
21
Table
E­
11
1­
hour
Statistics
for
CAMS
678,
September
13
 
20,
1999
   ..
E­
21
Table
E­
12
1­
hour
Statistics
for
San
Antonio
Monitors
(
CAMS
23,
58,
59
&
678)
September
13
 
20,
1999
                  
E­
21
Table
E­
13
1­
hour
Statistics
for
CAMS
62,
September
13
 
20,
1999
    .
E­
22
Table
E­
14
1­
hour
Statistics
for
CAMS
601,
September
13
 
20,
1999    
E­
22
Table
E­
15
1­
hour
Statistics
for
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
&
678)
September
13
 
20,
1999
         .
E­
22
Table
E­
16
1­
hour
Statistics
for
Corpus
Christi
(
CAMS
4
&
21),
September
13
 
20,
1999
                       
E­
22
Table
E­
17
1­
hour
Statistics
for
CAMS
87,
September
13
 
20,
1999
    .
E­
23
Table
E­
18
8­
hour
Statistics
for
CAMS
3,
September
13
 
20,
1999     
E­
40
Table
E­
19
8­
hour
Statistics
for
CAMS
38,
September
13
 
20,
1999
    .
E­
40
Table
E­
20
8­
hour
Statistics
for
Austin
Downwind
Monitors
(
CAMS
3
&
38),
September
13
 
20,
1999
                  
E­
41
Table
E­
21
8­
hour
Statistics
for
CAMS
23,
September
13
 
20,
1999
    .
E­
41
Table
E­
22
8­
hour
Statistics
for
CAMS
58,
September
13
 
20,
1999
    .
E­
41
Table
E­
23
8­
hour
Statistics
for
CAMS
59,
September
13
 
20,
1999
    .
E­
41
Table
E­
24
8­
hour
Statistics
for
CAMS
678,
September
13
 
20,
1999    
E­
42
Table
E­
25
8­
hour
Statistics
for
San
Antonio
Downwind
Monitors
(
CAMS
23
&
58),
September
13
 
20,
1999               .
E­
42
Table
E­
26
8­
hour
Statistics
for
San
Antonio
Upwind
Monitors
(
CAMS
59
&
678),
September
13
 
20,
1999
              .
E­
42
Table
E­
27
8­
hour
Statistics
for
all
San
Antonio
Monitors
(
CAMS
23,
58,
59,
&
678),
September
13
 
20,
1999             
E­
42
Table
E­
28
8­
hour
Statistics
for
CAMS
601,
September
13
 
20,
1999    
E­
43
Table
E­
29
8­
hour
Statistics
for
CAMS
62,
September
13
 
20,
1999
    .
E­
43
Table
E­
30
8­
hour
Statistics
for
Central
Texas
Downwind
Monitors
(
CAMS
3,
23,
38,
58)
September
13
 
20,
1999
             
E­
43
Table
E­
31
8­
hour
Statistics
for
Central
Texas
Upwind
Monitors
(
CAMS
59,
62,
601,
678),
September
13
 
20,
1999
           ..
E­
43
E­
4
Appendix
E
List
of
Tables
(
continued)
Page
Table
E­
32
8­
hour
Statistics
for
Central
Texas
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999
           .
E­
44
Table
E­
33
8­
hour
Statistics
for
Corpus
Christi
Monitors
(
CAMS
4
&
21),
September
13
 
20,
1999
                 ..
E­
44
Table
E­
34
8­
hour
Statistics
for
CAMS
87,
September
13
 
20,
1999
    .
E­
44
Table
E­
35
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
3,
September
13
 
20,
1999    
E­
47
Table
E­
36
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
38,
September
13
 
20,
1999
   .
E­
48
Table
E­
37
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
Austin
Monitors
 
CAMS
3
&
38,
September
13
 
20,
1999
                  
E­
49
Table
E­
38
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
23,
September
13
 
20,
1999
   .
E­
50
Table
E­
39
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
58,
September
13
 
20,
1999
   .
E­
51
Table
E­
40
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
59,
September
13
 
20,
1999
   .
E­
52
Table
E­
41
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
678,
September
13
 
20,
1999   
E­
53
Table
E­
42
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
San
Antonio
Downwind
Monitors
(
CAMS
23
&
58),
September
13
 
20,
1999
          
E­
54
Table
E­
43
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
San
Antonio
Upwind
Monitors
(
CAMS
59
&
678),
September
13
 
20,
1999
              .
E­
55
Table
E­
44
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
San
Antonio
Monitors
(
CAMS
23,
58,
59
&
678),
September
13
 
20,
1999
              .
E­
56
Table
E­
45
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
62,
September
13
 
20,
1999
   
E­
57
Table
E­
46
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
601,
September
13
 
20,
1999   
E­
58
Table
E­
47
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
Central
Texas
Downwind
Monitors
(
CAMS
3,
23,
38,
58),
September
13
 
20,
1999
        
E­
59
Table
E­
48
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
Central
Texas
Upwind
Monitors
(
CAMS
59,
62,
601,
678),
September
13
 
20,
1999            .
E­
60
Table
E­
49
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999
      ..
E­
61
Table
E­
50
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
Corpus
Christi
Monitors
 
CAMS
4
&
21,
September
13
 
20,
1999
                  
E­
62
E­
5
Appendix
E
List
of
Tables
(
continued)
Page
Table
E­
51
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
87,
September
13
 
20,
1999
   .
E­
63
Table
E­
52
Comparison
of
Predicted
Peak
8­
hour
Concentrations
for
Final
UT
and
AACOG
Base
Case
Runs
           
E­
84
E­
6
Appendix
E
List
of
Figures
Page
Figure
E­
1
Nested
Grid
System
for
the
1999
South
Texas
Photochemical
Model
Simulation
                     ..
E­
9
Figure
E­
2
Long­
term
Palmer
Drought
Severity
Index
for
September
18,
1999.
E­
13
Figure
E­
3
Map
Showing
the
Delineation
of
Boundary
Segments
for
the
Photochemical
Model
used
by
Austin
and
San
Antonio
for
their
Early
Action
Compacts
                   
E­
17
Figure
E­
4
1­
hour
Unpaired
Peak
Accuracy,
CAMS
23,
September
13
 
20,
1999
                           
E­
23
Figure
E­
5
1­
hour
Unpaired
Peak
Accuracy,
CAMS
58,
September
13
 
20,
1999
                           
E­
24
Figure
E­
6
1­
hour
Unpaired
Peak
Accuracy,
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999    
E­
24
Figure
E­
7
1­
hour
Average
Paired
Peak
Accuracy
at
CAMS
23,
September
13­
20,
1999
                        
E­
25
Figure
E­
8
1­
hour
Average
Paired
Peak
Accuracy
at
CAMS
58,
September
13­
20,
1999
                        
E­
25
Figure
E­
9
1­
hour
Average
Paired
Peak
Accuracy
at
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13­
20,
1999
 
E­
26
Figure
E­
10
1­
hour
Normalized
Bias,
CAMS
23,
September
13
 
20,
1999
  .
E­
26
Figure
E­
11
1­
hour
Normalized
Bias,
CAMS
58,
September
13
 
20,
1999
  .
E­
27
Figure
E­
12
1­
hour
Normalized
Bias,
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999
      .
E­
27
Figure
E­
13
1­
hour
Normalized
Error,
CAMS
23,
September
13
 
20,
1999
  
E­
28
Figure
E­
14
1­
hour
Normalized
Error,
CAMS
58,
September
13
 
20,
1999
  
E­
28
Figure
E­
15
1­
hour
Normalized
Error,
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999
      .
E­
29
Figure
E­
16
1­
hour
Timing
Bias
at
CAMS
23,
September
13
 
20,
1999
   
E­
29
Figure
E­
17
1­
hour
Timing
Bias
at
CAMS
58,
September
13
 
20,
1999
   
E­
30
Figure
E­
18
1­
hour
Timing
Bias
at
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999
        .
E­
30
Figure
E­
19
1­
hour
Fractional
Bias
at
CAMS
23,
September
13
 
20,
1999
  
E­
31
Figure
E­
20
1­
hour
Fractional
Bias
at
CAMS
58,
September
13
 
20,
1999
  
E­
31
Figure
E­
21
1­
hour
Fractional
Bias
at
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678,
September
13
 
20,
1999
       
E­
32
Figure
E­
22
Observed
versus
Predicted
1­
hour
Average
Ozone
Concentrations
at
CAMS
23,
September
13
 
20,
1999
            
E­
33
Figure
E­
23
Observed
versus
Predicted
1­
hour
Average
Ozone
Concentrations
at
CAMS
58,
September
13
 
20,
1999
            
E­
33
Figure
E­
24
Observed
versus
Predicted
1­
hour
Average
Ozone
Concentrations
at
CAMS
59,
September
13
 
20,
1999
            
E­
34
Figure
E­
25
Observed
versus
Predicted
1­
hour
Average
Ozone
Concentrations
at
CAMS
678,
September
13
 
20,
1999
           .
E­
34
Figure
E­
26
Scatter
Plot
of
1­
hour
Observed/
Predicted
Data
Pairs
at
CAMS
23.
E­
35
Figure
E­
27
Scatter
Plot
of
1­
hour
Observed/
Predicted
Data
Pairs
at
CAMS
58.
E­
35
Figure
E­
28
Scatter
Plot
of
1­
hour
Observed/
Predicted
Data
Pairs
at
CAMS
59.
E­
36
Figure
E­
29
Scatter
Plot
of
1­
hour
Observed/
Predicted
Data
Pairs
at
CAMS
678
                           ..
E­
36
E­
7
Appendix
E
List
of
Figures
(
continued)
Page
Figure
E­
30
Comparison
of
Observed/
predicted
1­
hour
Concentrations
at
CAMS
23
and
Daily
NOx
EI
for
Bexar
County,
September
13
 
20,
1999
                        .
E­
37
Figure
E­
31
Comparison
of
Observed/
predicted
1­
hour
Concentrations
at
CAMS
58
and
Daily
NOx
EI
for
Bexar
County,
September
13
 
20,
1999
                        
E­
38
Figure
E­
32
Comparison
of
Observed/
predicted
1­
hour
Concentrations
at
Four
San
Antonio
CAMS
and
Daily
NOx
EI
for
Bexar
County,
September
13
 
20,
1999
                  
E­
38
Figure
E­
33
Normalized
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
1,
September
13­
20,
1999
      
E­
64
Figure
E­
34
Normalized
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
2,
September
13­
20,
1999      .
E­
65
Figure
E­
35
Normalized
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
3,
September
13­
20,
1999
      
E­
66
Figure
E­
36
Fractional
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
1,
September
13­
20,
1999
      
E­
67
Figure
E­
37
Fractional
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
2,
September
13­
20,
1999
      
E­
68
Figure
E­
38
Fractional
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
3,
September
13­
20,
1999
      
E­
69
Figure
E­
39
Scatter
/
Q­
Q
Plots
for
CAMS
23
Calculated
using
Three
Methodologies
                      ..
E­
71
Figure
E­
40
Scatter
/
Q­
Q
Plots
for
CAMS
58
Calculated
using
Three
Methodologies                       
E­
72
Figure
E­
41
Scatter
/
Q­
Q
Plots
for
the
Eight
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678)
Calculated
using
Three
Methodologies                       
E­
73
Figure
E­
42
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Wednesday,
September
15,
1999
     
E­
75
Figure
E­
43
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Thursday,
September
16,
1999      ..
E­
75
Figure
E­
44
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Friday,
September
17,
1999
       ..
E­
76
Figure
E­
45
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Saturday,
September
18,
1999
      .
E­
76
Figure
E­
46
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Sunday,
September
19,
1999
       
E­
77
Figure
E­
47
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Monday,
September
20,
1999
      ..
E­
77
Figure
E­
48
Predicted
Reduction
in
Ozone
Concentrations
(%)
in
the
SAER
after
Removing
Austin,
Corpus
Christi,
and
Houston
Anthropogenic
NOx
and
VOC
Emissions
from
the
Photochemical
Model
      ..
E­
79
Figure
E­
49
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
(
4­
county
SAER)
NOx
and
VOC
Emissions
from
Simulation
17b,
September
15,
1999
                  
E­
80
E­
8
Appendix
E
List
of
Figures
(
continued)
Page
Figure
E­
50
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
(
4­
county
SAER)
NOx
and
VOC
Emissions
from
Simulation
17b,
September
16,
1999
                  
E­
81
Figure
E­
51
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
(
4­
county
SAER)
NOx
and
VOC
Emissions
from
Simulation
17b,
September
17,
1999
                  
E­
81
Figure
E­
52
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
(
4­
county
SAER)
NOx
and
VOC
Emissions
from
Simulation
17b,
September
18,
1999
                  
E­
82
Figure
E­
53
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
(
4­
county
SAER)
NOx
and
VOC
Emissions
from
Simulation
17b,
September
19,
1999
                  
E­
82
Figure
E­
54
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
(
4­
county
SAER)
NOx
and
VOC
Emissions
from
Simulation
17b,
September
20,
1999
                  
E­
83
E­
9
INTRODUCTION
The
elevated
ozone
episode
that
occurred
between
September
15
 
20,
1999
in
Southcentral
Texas
was
modeled
using
the
Comprehensive
Air
quality
Model
with
extensions
(
CAMx).
This
is
an
alternative
model
for
developing
air
quality
simulations
in
accordance
with
EPA's
(
On­
line,
no
date)
report,
Summary
Descriptions
of
Alternative
Air
Quality
Models.
CAMx
is
a
Eulerian
photochemical
grid
model
that
makes
use
of
a
two­
way
nested
grid
structure,
with
the
coarsest
grid
(
36­
km)
covering
a
wide
regional
domain,
a
12­
km
grid
that
incorporates
Eastern
Texas
including
the
nonattainment
areas
of
Dallas/
Fort
Worth,
Houston/
Galveston,
and
Beaumont/
Port
Arthur,
and
a
fine
grid
(
4­
km)
extending
over
four
Texas
near
nonattainment
areas
as
shown
in
figure
E­
1.

Figure:
E­
1.
Nested
Grid
System
for
the
1999
South
Texas
Photochemical
Model
Simulation.
E­
10
The
original
September
1999
simulation
was
developed
by
ENVIRON
International
Corporation
and
documented
in
their
report
Development
of
a
Joint
CAMx
Photochemical
Modeling
Database
for
the
Four
Southern
Texas
Near
Non­
Attainment
Areas
(
Emery,
Tai,
Wilson,
and
Yarwood,
2002).
Their
final
and
best
performing
run
(
developed
without
arbitrary
meteorological
or
emissions
inputs),
labeled
CAMx
Run
13,
produced
promising
results.
However,
CAMx
Run
13
exhibited
some
problems
that
were
prevalent
in
earlier
simulations,
including
consistent
under­
predictions
of
maximum
and
mean
ozone
concentrations.
ENVIRON
staff
recommended
performing
comprehensive
QA/
QC
procedures
on
model
settings
and
inputs,
particularly
meteorological
and
emissions
inputs,
and
incorporating
refinements
to
the
model
where
appropriate.

ENVIRON
and
the
University
of
Texas
at
Austin
conducted
extensive
analyses
of
the
meteorological
model
used
as
CAMx
input
for
the
September
1999
episode.
Development
of
the
original
meteorological
model,
labeled
Met
3c,
was
described
in
ENVIRON's
report
(
Emery
et
al.,
2002).
As
a
product
of
the
QA/
QC
analyses
conducted
by
the
ENVIRON/
UT
team,
numerous
improvements
were
made
to
the
meteorological
model.
Some
revised
runs,
including
Met
6f
and
Met
5d,
attenuated,
although
not
necessarily
eliminated,
certain
problems
associated
with
Met
3c,
including
over­
prediction
of
wind
speeds
at
night
and
under
predictions
during
the
daytime
and
over
prediction
of
early
morning
temperatures.
Most
of
the
refinements
the
ENVIRON/
UT
team
incorporated
into
the
meteorological
model
are
documented
in
the
report
Revised
Meteorological
Modeling
of
the
September
13­
20,
1999
Texas
Ozone
Episode
(
Emery,
Tai,
McGaughey,
and
Allen,
2003).
Development
of
the
meteorological
model
used
as
input
to
the
final
photochemical
model
run,
labeled
Met
5g,
is
described
in
appendix
B
of
the
SAER
SIP
revision.

Emissions
inputs
were
also
reviewed
and
refined
for
the
September
1999
baseline
simulation.
One
of
the
most
significant
refinements
made
to
the
1999
modeling
EI
was
use
of
MOBILE6
to
estimate
on­
road
emissions
for
some
urban
counties
in
Texas.
MOBILE6
was
not
available
when
the
original
September
1999
model
was
created.
Texas
Transportation
Institute,
under
contract
with
the
TCEQ,
recalculated
on­
road
files
for
18
NNA
counties,
including
Bexar
County,
using
the
newly­
released
MOBILE6
model.
The
original
MOBILE6
file,
referred
to
as
version
1,
was
later
refined
to
incorporate
an
improved
methodology
to
account
for
heavy­
duty
diesel
vehicle
VMT.
The
refined
MOBILE6
on­
road
file
for
the
18
NNA
counties
is
labeled
version
2.
(
See
Appendix
C
for
a
detailed
description
of
TTI's
on­
road
estimation
methodology.)
In
addition
to
enhanced
mobile
on­
road
estimations,
the
State
provided
AACOG
with
refined
EIs
for
specific
geographic
locations
including
other
NNA
areas
(
Austin,
Corpus
Christi
and
Victoria),
Houston,
and
the
remainder
of
the
State
of
Texas.
1
AACOG
was
also
provided
a
refined
point
source
file
(
TCEQ
2003),
which
contained
additional
VOC
emissions
for
Houston,
to
account
for
the
results
of
a
study
conducted
as
part
of
the
TexAQS
2000
project.
However,
staff
decided
to
omit
the
revised
Houston
point
source
file
in
the
final
1999
baseline
simulation
for
several
reasons.
First,
a
sensitivity
run
(
CAMx
Run
17c)
in
which
the
file
was
included
as
model
input
indicated
1
The
updated
Texas
EI
was
developed
for
the
2000
Houston
attainment
SIP
and
includes
refined
area
and
non­
road
emission
estimations.
Since
the
Houston
EI
was
developed
for
a
different
time
period,
the
emissions
were
backcast
to
September
1999
using
projection
ratios
developed
from
EGAS
and
NONROAD
models.
E­
11
the
increased
VOC
emissions
for
Houston
point
sources
had
an
insignificant
impact
on
ozone
concentrations
in
San
Antonio
during
the
September
1999
episode.
Second,
as
of
the
timeframe
when
the
final
baseline
run
(
CAMx
Run
18)
was
developed
and
tested,
the
State
had
yet
to
receive
approval
from
the
EPA
to
use
the
modified
point
source
file
in
the
Houston
attainment
demonstration
SIP.
Third,
the
refined
VOC
database
did
not
include
a
2007
projection.
Using
the
refined
data
for
only
the
1999
base
case
would
create
inconsistencies
in
the
treatment
of
the
base
case
and
projection
case
EIs.
And
fourth,
the
modified
EI
was
designed
for
a
2000
episode,
not
1999.

Table
E­
1
provides
a
description
of
the
meteorological
and
emissions
inputs
to
the
final
set
of
sensitivity
and
baseline
runs
conducted
for
the
September
episode
beginning
with
the
first
1999
base
case
submitted
by
ENVIRON,
CAMx
Run
13,
through
development
of
the
final
baseline
run,
CAMx
Run
18.

MODEL
INPUTS
Methodologies
used
to
develop
the
major
inputs
to
the
photochemical
model
are
described
in
detail
in
appendix
B
Development
of
the
1999
Meteorological
Model,
Appendix
C
On­
Road
Mobile
Emissions
Inventory
Development,
and
appendix
D
1999
Base
Case
Modeling
Emissions
Inventory
Development
of
the
SAER
attainment
demonstration
SIP.
This
appendix
focuses
on
development
of
other
inputs
to
the
model
simulation
and
selection
of
CAMx
model
options.
The
following
sections
on
development/
determination
of
land
use
data,
dry
deposition
algorithms,
chemistry
data,
boundary/
initial
conditions,
and
CAMx
model
options
were
provided
by
The
University
of
Texas
at
Austin
and
ENVIRON
(
September
17,
2003)
in
their
draft
report
Development
of
the
September
13­
20,
1999
Base
Case
Photochemical
Model
for
Austin's
Early
Action
Compact.
Except
where
noted,
these
sections
focus
on
model
inputs
and
options
for
the
final
baseline
run,
CAMx
Run
18.

Land
Use
Data
ENVIRON
(
Jimenez
et
al.,
2002)
developed
land
use
grid
data
to
characterize
surface
boundary
conditions
for
the
September
1999
episode.
This
information
was
developed
from
the
same
data
used
to
generate
spatial
emission
surrogates.
ENVIRON
created
software
to
process
the
raw
spatial
surrogate
data
into
the
eleven
land
use
categories
used
by
CAMx,
to
grid
the
data
to
the
36,
12,
and
4km
CAMx
grids,
and
to
write
the
results
in
an
appropriate
format
for
input
to
CAMx.

Dry
Deposition
Algorithms
Dry
deposition
algorithms
in
CAMx
are
based
on
the
regional­
scale
deposition
model
developed
by
Wesely
et
al.
(
1989).
These
algorithms
have
been
widely
used
in
both
field
applications
and
air
quality
models.

ENVIRON
and
UT
reviewed
the
Palmer
drought
severity
index,
shown
in
figure
E­
2
for
eastern
Texas
and
found
a
moderate
level
of
drought
stress
during
the
episode.
Although
the
MM5
models
accounted
for
reduced
soil
moisture,
the
original
dry
deposition
algorithm
in
the
September
13­
20,
1999
CAMx
model
did
not
account
for
vegetation
moisture
stress.
Because
of
the
potential
influence
of
drought
stress
on
the
uptake
of
pollutants
through
plant
stomata
and
the
importance
of
dry
deposition
as
a
E­
12
Table
E­
1.
Meteorological
and
Emissions
Inputs
to
CAMx
Photochemical
Model
Test
Runs.

CAMx
Run:
Run
13
Run
13b
Run
13c
Run
13d
Run
13e
Run
14
Run
15
Run
16
Run
17
Run
17b
Run
17c
Run
17d
Run
18
Met
3c
 
 
 
 
 
Met
3c
+
Increased
Mixing
layer
 
Mer
6f
 
Met
5d
 
 
Met
5g
 
 
 
 
 
Updated
Texas
EI
 
 
 
 
Updated
Austin
EI
 
 
 
 
Updated
2007
Regional
EI*
 
Updated
Victoria's
EI
 
BC/
IC
60
 
 
 
 
 
 
 
Mobile5
 
 
 
 
 
 
 
 
Mobile6
version
1
 
 
 
Mobile6
version
1x1.4**
 
Mobile6
version
2
(
VMT
Upgrade)
 
Modified
dry
deposition
 
 
 
 
 
 
 
Additional
VOC
from
Houston's
Point
Sources
 
*
The
Austin,
Corpus
Christi,
San
Antonio,
and
Victoria
portions
of
the
August
2000
Texas
EI
were
removed
from
the
model
and
replaced
with
refined
area­
specific
September
1999
emissions
for
those
areas.
For
the
remainder
of
the
State,
the
August
2000
Texas
EI
was
backcast
to
September
1999.

**
Version
1x1.4
is
a
sensitivity
run
conducted
prior
to
receiving
the
MOBILE6
version
2
EI
file.
It
was
employed
to
test
the
anticipated
impact
of
increasing
HDD
VMT
(
Version
2)
by
multiplying
HDD
VMT
by
1.4
(
increasing
HDD
VMT
by
40%).
E­
13
physical
removal
process
for
ozone
and
other
secondary
pollutants,
ENVIRON
initiated
changes
to
the
CAMx
deposition
algorithms
for
the
September
13­
20,
1999
episode.

Wesely
supplies
minimum
bulk
stomatal
resistances
by
season
and
land
use
type.
High
resistances
(
9999)
represent
no
deposition
through
the
stomata.
Vegetation
moisture
stress
codes
(
0=
unstressed;
1=
stressed,
2=
extremely
stressed)
essentially
define
The
current
formulation
of
Wesely's
dry
deposition
model
in
CAMx
can
be
manipulated
to
account
for
drought
stress
through
the
use
of
vegetation
moisture
stress
codes.
Factors
by
which
minimum
bulk
stomatal
resistances
are
increased/
decreased
to
reflect
drought
conditions:
if
istress
=
1
then
the
stomatal
resistance
is
increased
by
a
factor
of
two;
if
istress
=
2,
then
the
stomatal
resistance
is
increased
by
a
factor
of
10.
ENVIRON
increased
drought
stress
codes
over
land
use
categories
by
one
to
reflect
summer
drought
conditions
for
those
land
use
categories
that
did
not
already
have
very
high
minimum
bulk
stomatal
resistances.
With
the
approval
of
the
TCEQ
and
the
U.
S.
EPA,
these
modifications
were
adopted
by
San
Antonio
and
Austin
for
the
September
13­
20,
1999
photochemical
model.

Figure
E­
2.
Long­
term
Palmer
Drought
Severity
Index
for
September
18,
1999
(
NOAA
Climate
Prediction
Center,
2003).
E­
14
Chemistry
Data
Chemistry
data,
developed
by
ENVIRON
(
Emery
et
al.,
2002)
for
the
September
1999
episode,
are
summarized
in
Table
E­
2.

Table
E­
2.
Summary
of
Chemistry
Data
for
the
September
13­
20,
1999
CAMx
Model.
Input
Data/
Specification
Description
Chemistry
Parameters
CB4
with
current
radical
termination
reactions
and
isoprene
mechanism
Photolysis
rates
TUV
version
4
Albedo/
Haze/
Ozone
File
 
Surface
UV
albedo
from
land
use
grid
data
 
Total
ozone
column
data
from
satellite
data
from
the
Total
Ozone
Mapping
Spectrometer
(
TOMS)
 
Haze
optical
depth
field
assumed
spatially
and
temporally
constant
at
0.1.

Boundary
and
Initial
Conditions
A
number
of
sensitivity
studies
focusing
on
boundary
and
initial
conditions
were
conducted
following
the
initial
CAMx
model
performance
evaluation
of
the
September
13­
20,
1999
episode.
The
results
of
the
model
performance
evaluation
and
CAMx
simulations
conducted
by
ENVIRON
suggested
that
the
September
1999
episode
was
promising,
but
required
additional
refinement
to
improve
performance.
Daily
peak
and
daily
mean
ozone
concentrations
were
under
predicted
by
10­
20%
and
10­
30%,
respectively,
at
each
ambient
monitoring
site.
Although
the
unpaired
peak
accuracy
and
normalized
gross
error
met
EPA
criteria
on
most
days,
the
relative
bias
failed
to
meet
EPA
criteria
in
all
near
non­
attainment
areas
on
at
least
one
episode
day.

EPA
default
boundary
and
initial
conditions,
which
were
used
in
the
original
modeling,
are
shown
in
the
last
column
of
Table
E­
3.
AACOG
initiated
sensitivity
studies
that
focused
on
increasing
ozone
concentrations
from
40
ppb
to
60
ppb
along
all
boundaries
of
the
36­
km
domain
and
in
the
initial
conditions
supplied
to
CAMx.
Model
performance
improved
significantly
and
indeed,
EPA
performance
criteria
for
unpaired
peak
accuracy,
normalized
bias,
and
gross
error,
were
met
on
most
days
in
Central
Texas.
The
AACOG
and
the
University
of
Texas,
on
behalf
of
CAPCO,
then
undertook
a
number
of
sensitivity
studies
to
further
elucidate
the
influence
of
boundary
and
initial
conditions
on
model
performance:

 
Increase
all
boundary
conditions
to
60
ppb
 
Increase
ozone
concentrations
along
the
northern
and
eastern
boundaries
from
40
ppb
to
60
ppb.
 
Increase
ozone
concentrations
along
the
northern
boundary
from
40
ppb
to
60
ppb.
 
Increase
ozone
concentrations
along
the
eastern
boundary
from
40
ppb
to
60
ppb.
 
Increase
initial
ozone
concentrations
from
40
ppb
to
60
ppb.
E­
15
In
conjunction
with
the
sensitivity
studies,
UT
also
examined
whether
the
increase
in
ozone
concentrations
upwind
and
along
the
boundaries
could
be
supported
by
ambient
monitoring
data
during
the
episode.
UT
developed
time
series
from
relevant
U.
S.
EPA
AIRS
monitors
as
well
as
from
the
IMPROVE
network.
In
total,
data
from
sixteen
states
(
Oklahoma,
Kansas,
Nebraska,
Missouri,
Iowa,
Indiana,
Illinois,
Arkansas,
West
Virginia,
Ohio,
Kentucky,
Tennessee,
North
Carolina,
Georgia,
South
Carolina,
Alabama)
were
examined.
Although
the
sensitivity
studies
led
to
variable
degrees
of
improvement
in
model
performance,
the
ambient
data
did
not
support
increasing
ozone
concentrations
above
40
ppb
along
most
boundaries
throughout
the
episode.
The
exception
was
the
area
of
domain
that
encompassed
Tennessee
and
North
Carolina,
which
experienced
ozone
concentrations
in
excess
of
60
ppb
on
most
episode
days.

The
TCEQ
suggested
that
UT
examine
the
effects
of
boundary
and
initial
conditions
used
for
the
September
1993
photochemical
model
for
Houston's
State
Implementation
Plan
on
model
performance.
These
boundary
and
initial
conditions
are
currently
being
used
for
the
August
13­
22,
1999
episode
for
the
Dallas/
Fort
Worth
area,
for
the
Longview/
Tyler/
Marshall
area,
and
for
Oklahoma
(
Yarwood,
2003).

All
of
the
model
applications
described
above
suffered
from
a
tendency
to
underpredict
regional
ozone
levels,
which
prompted
a
review
of
the
boundary
conditions.
In
particular,
total
VOC
levels
of
only
4.4
ppb
may
be
too
low
in
areas
of
the
regional
modeling
domain
that
are
over
land.
Boundary
condition
values
shown
in
Columns
1­
3
of
Table
E­
3
were
originally
developed
for
the
TCEQ's
regional
modeling
of
the
September
1993
episode
(
Yocke
et
al.,
1996).
These
values
varied
by
boundary
segment,
as
shown
in
Figure
E­
3
and
were
based
on
several
data
sources.
Concentrations
along
the
East/
Northeastern
Boundary
were
based
on
EPA's
guidance
for
UAM
modeling
(
EPA,
1991)
with
CO
reduced
from
350
ppb
to
200
ppb
and
higher
biogenic
VOCs
(
ISOP,
MEOH
and
ETOH)
based
on
measurements
at
Kinterbish,
AL
for
the
Rural
Oxidants
in
the
Southern
Environment
study
(
Goldan
et
al.,
1995).
Western
boundary
concentrations
were
based
on
EPA's
UAM
modeling
guidance
(
EPA,
1991)
with
CO
reduced
from
350
ppb
to
200
ppb
and
were
consistent
with
data
from
Niwot
Ridge,
CO
(
Watkins
et
al.,
1995).
Southern
Boundary
concentrations
were
based
on
the
GMAQS
(
Gulf
of
Mexico
Air
Quality
Study)
sponsored
by
the
Minerals
Management
Service
(
MMS,
1995).
Initial
conditions
were
identical
to
those
in
Column
3.

UT
conducted
a
sensitivity
study
using
the
September
1993
Houston/
Galveston
boundary
and
initial
conditions
and
found
improved
model
performance
for
the
September
13­
20,
1999
episode
in
Central
Texas.
The
negative
bias
predicted
by
the
original
model
was
considerably
reduced,
and
this
metric,
now
fell
within
the
range
of
EPA
performance
criteria.
Model
performance
statistics
will
be
described
in
detail
below,
but
improved
ozone
predictions
were
observed
throughout
the
regional
domain,
including
in
the
Houston/
Galveston
area.
Because
of
the
significantly
improved
model
performance
and
the
robust
technical
basis
of
these
data,
Austin
and
San
Antonio,
in
collaboration
with
the
TCEQ,
decided
to
use
these
boundary
and
initial
conditions
for
the
photochemical
modeling
for
their
Early
Action
Compacts.
E­
16
Table
E­
3.
Boundary
and
Initial
Conditions
used
by
in
the
Original
Model
and
the
Final
Model
used
by
San
Antonio
and
Austin
for
their
Early
Action
Compacts.
(
Initial
conditions
were
identical
to
concentrations
along
the
western
boundary.
The
EAC
boundary
and
initial
conditions
are
identical
to
those
used
in
the
September
1993
Houston/
Galveston
model
for
the
State
Implementation
Plan.)

Species
NE
Boundary
below
1700m
(
ppb)
(
EAC)
West
Boundary
below
1700
m
(
ppb)
(
EAC)
SE
Boundary
and
Above
1700m
(
ppb)
(
EAC)
Default
Initial
and
Boundary
Conditions
used
in
original
modeling*.

O3
40
40
40
40
CO
200
200
100
100
NO
0.1
0.1
0.1
0.000049
NO2
1
1
1
0.08555
HNO3
3
3
1
1.525
HNO2
0.001
0.001
0.001
0.000728
ALD2
0.555
0.555
0.05
0.1051
ETH
0.51
0.51
0.15
0.005315
HCHO
2.1
2.1
0.05
1.068
OLE
0.3
0.3
0.05
PAR
14.9
14.9
7.6
3.078
TOL
0.18
0.18
0.0786
0.006043
XYL
0.0975
0.0975
0.0688
ISOP
3.6
0.1
0.001
PAN
0.1
0.1
0.1
0.03834
H2O2
3
3
1
2.263
MEOH
8.5
0.001
0.001
ETOH
1.1
0.001
0.001
*
EPA
Guidance
E­
17
Figure
E­
3.
Map
Showing
the
Delineation
of
Boundary
Segments
for
the
Photochemical
Model
used
by
Austin
and
San
Antonio
for
their
Early
Action
Compacts.*

*
below
1700
m
CAMx
Model
Options
CAMx
model
options,
established
by
ENVIRON
(
Emery
et
al.,
2002),
are
summarized
in
Table
E­
4.

Table
E­
4.
Summary
of
Options
for
the
September
13­
20,
1999
CAMx
Model.
Input
Data/
Specification
Description
Advection
Scheme
Piece
Parabolic
Method
(
PPM)
Plume­
in­
Grid
Model
Selected
for
major
NOx
sources
>
10
tons/
day
in
4
km
grid
>
25
tons/
day
in
12­
and
36­
km
grids
Chemical
Mechanism
CMC
fast
solver
SEPTEMBER
13­
20,
1999
PHOTOCHEMICAL
MODEL
PERFORMANCE
Performance
analyses
were
conducted
on
several
versions
of
the
baseline
runs
as
a
means
of
comparing
the
results
of
refinements
made
to
the
model.
EPA
recommends
various
types
of
performance
tests
in
their
8­
hour
guidance:
graphics,
ozone
metrics,
precursor
concentrations,
observational
models,
weekend/
weekday
comparisons,
ratios
E­
18
of
indicator
species,
and
retrospective
analyses.
Analyses
using
observational
models,
ratios
of
indicator
species,
and
retrospective
studies
were
not
conducted
on
the
1999
model,
due
to
a
lack
of
necessary
data
&
tools.

For
example,
the
use
of
observational
models
is
suggested
in
cases
where
an
extensive
monitoring
network
exists
and
precursor
and
indicator
species
are
measured
using
instruments
with
appropriate
sensitivity
(
EPA,
1999).
During
the
September
1999
time
period,
just
three
regulatory
ozone
monitors
were
operational
in
the
entire
San
Antonio
EAC
region.
Also,
there
were
no
(
e.
g.,
SO2)
species
being
measured
in
the
region.
The
sole
ozone
precursor
monitored
by
San
Antonio
area
CAMS
stations
was
NOx:
CAMS
27
located
in
downtown
San
Antonio,
CAMS
59
located
at
Calaveras
Lake
in
southeastern
Bexar
County,
and
CAMS
62
located
northeast
of
San
Antonio
in
Caldwell
County.
However,
the
NOx
levels
were
employed
for
precursor
concentration
analyses,
provided
in
section
3.6.3
of
the
Executive
Summary;
all
other
performance
evaluations
are
provided
in
the
sections
that
follow.

Model
performance
was
evaluated
using
statistical
and
graphical
metrics
in
accordance
with
EPA
guidance
(
1999)
for
both
1­
hour
and
8­
hour
attainment
demonstrations.
The
following
sections
provide
the
results
of
the
1­
hour
and
8­
hour
performance
tests
conducted
on
the
final
September
1999
base
case,
CAMx
Run
18.

During
the
September
1999
episode,
peak
ozone
concentrations
in
the
SAER
were
measured
at
CAMS
23
and
58.
Therefore,
metrics
results
for
these
monitors
are
of
particular
interest.
Because
of
their
importance,
all
statistics
(
1­
hour
and
8­
hour)
for
CAMS
23
and
CAMS
58
are
included
in
this
appendix.
Some
test
results
(
1­
hour
time
series
plots
and
8­
hour
scatter
and
Q­
Q
plots)
for
other
monitors
within
the
4­
km
subdomain
were
omitted
from
this
appendix
for
the
sake
of
brevity.
These
tests/
results
will
be
provided
to
the
EPA
and
the
TCEQ
on
compact
disc
with
submission
of
this
SIP
revision.

Model
Performance:
1­
Hour
Averaged
Ozone
Concentrations
EPA
recommends
conducting
a
series
of
1­
hour
graphical
performance
procedures
and
statistical
performance
tests
as
part
of
the
performance
evaluation
process
(
EPA
1991).
When
evaluating
statistical
test
results,
monitoring
network
density
should
be
considered.
Since,
individually,
the
San
Antonio
and
Austin
area
networks
are
sparse,
statistical
measurements
were
conducted
on
groups
as
well
as
individual
monitors.

Ozone
Metrics
The
1­
hour
statistics
were
determined
using
a
program
developed
by
ENVIRON:
"
camxpost."
This
program
calculates
unpaired
peak
accuracy
(
UPA),
average
paired
peak
accuracy
(
APPA),
peak
timing
bias
(
PTB),
normalized
bias
(
NB),
fractional
bias,
normalized
error
(
NE),
and
fractional
error
(
FE).
Statistical
metrics
and
associated
EPA
performance
criteria
for
1­
hour
averaged
ozone
concentrations
include
Statistical
Performance
Measure
Performance
Criteria
Unpaired
highest
prediction
accuracy
±
20%
Normalized
bias
±
15%
Gross
error
of
all
pairs
>
60
ppb
+
35%
Average
paired
peak
accuracy
­­
Bias
in
peak
timing
­­
E­
19
The
results
of
the
camxpost
program
are
presented
for
each
monitor
and
for
groups
of
monitors
(
averages
for
Austin
monitors,
San
Antonio
monitors,
and
the
eight
Central
Texas
monitors)
within
the
4­
km
subdomain
in
tables
E­
5
through
E­
17.
Yellowhighlighted
values
in
these
tables
represent
statistics
that
fall
outside
EPA's
performance
criteria
on
primary
episode
days
(
September
15
 
20,
1999).
While
it
is
not
necessary
to
conduct
these
tests
on
model
initialization
days
(
September
13­
14,
1999),
statistics
for
the
initialization
period
are
included
for
comparison
purposes.
Model
initialization
statistics
that
fall
outside
performance
thresholds
are
listed
in
bold
type.
Columns
where
data
are
missing
represent
days
in
which
predicted
measurements
were
less
than
60
ppb.

The
1­
hour
statistical
results
are
also
provided
in
graphic
form
for
some
monitors.
Figures
E­
4
through
E­
21
present
bar
graphs
of
1­
hour
statistics
for
CAMS
23,
CAMS
58,
and
the
averaged
data
for
eight
Central
Texas
monitors.
Monitor
numbers,
monitor
locations,
and
descriptions
of
monitor
groups
are
as
follows:

CAMS
#
Monitor
Name/
Group
Name
Location
AIRS
#
3
Murchison
Travis
County
48­
453­
0014
4
Corpus
Christi
West
Nueces
County
48­
355­
0025
21
Corpus
Christi
Tuloso
Nueces
County
48­
355­
0026
23
San
Antonio
Northwest
Bexar
County
48­
029­
0032
38
Audubon
Travis
County
48­
453­
0020
58
Camp
Bullis
Bexar
County
48­
029­
0052
59
Calaveras
Lake
Bexar
County
48­
029­
0059
62
San
Marcos
Caldwell
County
48­
055­
0062
87
Victoria
Victoria
County
48­
469­
0003
601
Fayette
Fayette
County
48­
149­
0001
678
CPS
Pecan
Valley
Bexar
County
48­
029­
0055
3,
38
Austin
Monitors
Travis
County
23,
58,
59,
678
San
Antonio
Monitors
Bexar
County
3,
23,
38,
58,
59,
62,
601,
678
Central
Texas
Monitors
Bexar,
Caldwell,
Fayette,
Travis
Performance
statistics
for
the
San
Antonio
area
monitors
were
quite
good
overall,
although
the
statistical
results
for
CAMS
59
tended
to
exhibit
a
negative
bias.
Unpaired
peak
accuracy
was
also
somewhat
problematic,
particularly
at
CAMS
678.
When
results
of
all
four
San
Antonio
monitors
were
averaged
(
table
E­
12),
all
daily
1­
hour
statistical
measurements
fell
within
acceptable
bounds
with
the
exception
of
unpaired
peak
accuracy
on
September
19th.

Normalized
bias
was
also
a
performance
issue
for
the
Austin
area.
Furthermore,
when
1­
hour
results
for
the
two
Austin
monitoring
locations
were
averaged,
the
problem
only
improved
slightly.
One­
hour
statistics
for
the
coastal
areas,
Corpus
Christi
and
Victoria,
exhibited
a
similar
negative
bias.

Table
E­
15
provides
statistical
metrics
for
1­
hour
averaged
ozone
concentrations
in
Central
Texas,
which
includes
results
for
the
Austin,
San
Antonio,
San
Marcos,
and
Fayette
County
monitors.
As
demonstrated,
model
performance
met
EPA
acceptance
criteria
for
each
day
of
the
primary
episode.
E­
20
Table
E­
5.
1­
hour
Statistics
for
CAMS
3,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
N/
A
21.1%
13.1%
5.2%
5.1%
4.8%
3.2%
APPA
N/
A
N/
A
1.5%
­
5.0%
­
9.3%
­
3.9%
­
1.7%
­
28.5%
PTB
N/
A
N/
A
­
1
0
0
1
2
1
NB
N/
A
N/
A
1.2%
­
7.5%
15.0%
­
10.9%
­
5.2%
­
23.7%
FB
N/
A
N/
A
1.1%
­
7.9%
­
16.5%
­
12.2%
­
5.8%
­
28.5%
NE
N/
A
N/
A
4.2%
7.5%
15.0%
11.4%
10.2%
23.7%
FE
N/
A
N/
A
4.1%
7.9%
16.5%
12.7%
10.6%
28.5%

Table
E­
6.
1­
hour
Statistics
for
CAMS
38,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
12.4%
13.8%
­
16.9%
­
5.4%
15.0%
­
3.8%
26.8%
APPA
N/
A
­
9.1%
­
7.2%
­
20.5%
­
12.0%
­
25.6%
8.9%
22.0%
PTB
N/
A
­
1
­
1
­
2
1
1
1
­
10
NB
N/
A
­
7.1%
­
7.1%
­
12.8%
­
15.8%
­
19.4%
­
15.8%
­
19.4%
FB
N/
A
­
7.4%
­
7.6%
­
13.8%
­
18.0%
­
22.1%
­
17.7%
­
22.1%
NE
N/
A
7.1%
8.1%
12.8%
16.3%
19.4%
15.8%
19.4%
FE
N/
A
7.4%
8.5%
13.8%
18.4%
22.1%
17.7%
22.1%

Table
E­
7.
1­
hour
Statistics
for
Austin
Monitors
(
CAMS
3
&
38),
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
12.4%
13.8%
­
16.9%
­
5.4%
­
15.0%
­
3.8%
3.2%
APPA
N/
A
­
9.1%
­
2.8%
­
12.7%
­
10.6%
­
14.7%
­
5.3%
­
25.2%
PTB
N/
A
­
1
­
1
­
1
1
1
2
­
6
NB
N/
A
­
7.1%
­
4.2%
­
10.5%
­
15.5%
­
15.7%
­
11.0%
­
21.3%
FB
N/
A
­
7.4%
­
4.5%
­
11.3%
­
17.4%
­
17.8%
­
12.4%
­
24.8%
NE
N/
A
7.1%
6.7%
10.5%
15.8%
15.9%
13.3%
21.3%
FE
N/
A
7.4%
7.0%
11.3%
17.7%
18.0%
14.6%
24.8%

Table
E­
8.
1­
hour
Statistics
for
CAMS
23,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
1.3%
7.9%
9.0%
2.6%
15.0%
8.7%
26.3%
6.7%
APPA
­
16.7%
­
12.5%
­
9.9%
­
16.7%
­
5.3%
3.9%
­
3.2%
­
0.3%
PTB
­
1
0
­
1
­
3
2
1
­
2
­
3
NB
­
20.7%
­
14.3%
­
12.6%
­
21.2%
­
11.0%
­
3.2%
­
5.7%
6.8%
FB
­
23.6%
­
15.5%
­
13.7%
­
24.5%
­
11.8%
­
3.7%
­
6.0%
4.6%
NE
20.7%
14.3%
12.6%
21.2%
11.0%
6.7%
6.7%
19.8%
FE
23.6%
15.5%
13.7%
24.5%
11.8%
7.1%
7.0%
19.2%
E­
21
Table
E­
9.
1­
hour
Statistics
for
CAMS
58,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
5.7%
20.8%
14.0%
15.0%
15.0%
2.6%
23.6%
13.1%
APPA
­
12.9%
­
4.2%
­
16.1%
­
2.7%
0.6%
­
12.3%
20.1%
­
9.6%
PTB
­
3
­
3
­
4
­
2
1
0
0
­
4
NB
­
24.5%
­
6.2%
­
9.9%
­
7.9%
0.9%
­
10.4%
4.8%
2.1%
FB
­
28.0%
­
6.7%
­
10.5%
­
8.7%
0.8%
­
11.0%
4.2%
0.8%
NE
24.5%
6.8%
9.9%
10.2%
4.7%
10.4%
9.0%
14.4%
FE
28.0%
7.3%
10.5%
11.0%
4.7%
11.0%
8.6%
14.5%

Table
E­
10.
1­
hour
Statistics
for
CAMS
59,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
8.8%
14.2%
9.9%
1.4%
5.4%
32.0%
23.6%
16.6%
APPA
­
20.7%
­
20.5%
­
27.3%
­
13.8%
­
17.8%
­
21.1%
­
17.6%
2.6%
PTB
1
­
3
2
5
7
2
1
­
2
NB
­
23.3%
­
19.3%
­
16.8%
­
10.2%
­
10.6%
­
18.5%
­
14.4%
­
16.6%
FB
­
26.5%
­
21.4%
­
18.8%
­
11.3%
­
11.6%
­
20.6%
­
15.7%
­
20.1%
NE
23.3%
19.3%
16.8%
12.5%
11.3%
18.5%
15.4%
18.0%
FE
26.5%
21.4%
18.8%
13.6%
12.3%
20.6%
16.8%
21.4%

Table
E­
11.
1­
hour
Statistics
for
CAMS
678,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
1.0%
25.3%
27.0%
14.1%
33.5%
32.0%
30.4%
5.7%
APPA
­
7.9%
­
10.2%
­
11.3%
­
13.0%
­
1.0%
­
7.9%
­
4.6%
­
10.1%
PTB
1
­
1
1
­
2
1
3
­
1
­
1
NB
­
8.2%
­
12.2%
­
9.6%
­
7.0%
1.6%
­
6.2%
­
5.7%
­
4.5%
FB
­
8.7%
­
13.0%
­
10.2%
­
7.6%
1.6%
­
6.7%
­
5.9%
­
4.8%
NE
8.2%
12.2%
9.6%
8.2%
2.7%
6.5%
5.7%
6.1%
FE
8.7%
13.0%
10.2%
8.7%
2.6%
7.0%
5.9%
6.4%

Table
E­
12.
1­
hour
Statistics
for
San
Antonio
Monitors
(
CAMS
23,
58,
59
&
678),
September
13
 
20,
1999.
San
Antonio
Monitors
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
8.8%
7.9%
9.0%
2.6%
5.4%
2.6%
23.6%
5.7%
APPA
­
14.6%
­
11.9%
­
16.1%
­
11.5%
­
5.9%
­
9.4%
­
1.3%
­
4.3%
PTB
­
1
­
2
­
1
­
1
3
2
­
1
­
3
NB
­
19.8%
­
12.9%
­
12.9%
­
11.5%
­
5.6%
­
10.2%
­
5.7%
­
4.0%
FB
­
22.4%
­
14.1%
­
14.1%
­
12.9%
­
6.1%
­
11.1%
­
6.4%
­
6.0%
NE
19.8%
13.1%
12.9%
13.0%
8.0%
11.0%
9.6%
14.9%
FE
22.4%
14.3%
14.1%
14.4%
8.5%
11.9%
10.0%
15.9%
E­
22
Table
E­
13.
1­
hour
Statistics
for
CAMS
62,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
N/
A
21.1%
15.4%
7.7%
19.2%
22.9%
30.1%
APPA
N/
A
N/
A
­
9.2%
5.3%
­
14.6%
­
11.6%
­
3.5%
1.2%
PTB
N/
A
N/
A
0
1
5
­
11
­
1
0
NB
N/
A
N/
A
­
4.5%
4.9%
­
10.0%
­
11.6%
­
6.5%
2.6%
FB
N/
A
N/
A
­
4.7%
4.5%
­
11.4%
­
12.5%
­
6.9%
2.6%
NE
N/
A
N/
A
6.3%
6.5%
14.2%
11.6%
7.6%
3.3%
FE
N/
A
N/
A
6.5%
6.3%
15.3%
12.5%
8.0%
3.2%

Table
E­
14.
1­
hour
Statistics
for
CAMS
601,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
4.5%
N/
A
9.3%
29.5%
­
6.1%
3.6%
13.8%
24.4%
APPA
­
18.7%
N/
A
­
5.5%
2.0%
­
14.4%
­
10.5%
­
5.7%
4.0%
PTB
­
1
N/
A
­
1
­
3
2
­
1
­
2
3
NB
­
18.5%
N/
A
­
3.8%
­
0.9%
­
17.6%
­
11.8%
­
12.5%
­
5.8%
FB
­
20.3%
N/
A
­
4.1%
­
1.2%
­
19.7%
­
12.9%
­
14.5%
­
6.8%
NE
18.5%
N/
A
5.2%
6.7%
17.6%
11.9%
16.4%
10.5%
FE
20.3%
N/
A
5.4%
6.7%
19.7%
13.0%
18.1%
11.3%

Table
E­
15.
1­
hour
Statistics
for
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
8.8%
7.9%
­
9.0%
­
16.9%
­
5.4%
­
15.0%
­
3.8%
5.7%
APPA
­
15.4%
­
11.3%
­
10.6%
­
8.0%
­
9.2%
­
11.1%
­
3.1%
­
7.8%
PTB
­
1
­
2
­
1
­
1
2
­
1
0
­
2
NB
­
19.5%
­
11.7%
­
8.5%
­
8.2%
­
11.0%
­
12.1%
­
8.5%
­
9.1%
FB
­
22.0%
­
12.7%
­
9.2%
­
9.2%
­
12.3%
­
13.4%
­
9.6%
­
11.2%
NE
19.5%
11.9%
9.5%
11.0%
12.7%
12.5%
11.7%
14.9%
FE
22.0%
12.9%
10.2%
11.8%
13.9%
13.8%
12.5%
16.6%

Table
E­
16.
1­
hour
Statistics
for
Corpus
Christi
(
CAMS
4
&
21),
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
13.0%
­
11.9%
­
0.9%
2.7%
­
1.5%
6.9%
9.6%
­
0.7%
APPA
­
31.8%
­
24.7%
24.9%
­
18.6%
­
15.2%
­
3.1%
4.9%
­
15.0%
PTB
­
1
­
2
0
­
6
­
4
­
2
3
2
NB
­
29.4%
­
23.6%
­
21.5%
­
23.0%
­
21.7%
­
14.3%
­
8.3%
­
25.4%
FB
­
35.5%
­
27.2%
­
24.2%
­
26.1%
­
24.7%
­
16.7%
­
10.5%
­
31.4%
NE
29.4%
23.6%
21.5%
23.0%
21.7%
15.6%
16.2%
25.4%
FE
35.5%
27.2%
24.2%
26.1%
24.7%
18.0%
17.9%
31.4%
E­
23
Table
E­
17.
1­
hour
Statistics
for
CAMS
87,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
13.8%
1.9%
8.4%
6.2%
3.0%
0.7%
­
0.5%
­
16.8%
APPA
­
44.8%
18.5%
­
17.3%
­
3.5%
­
6.6%
15.1%
­
3.9%
­
20.4%
PTB
0
­
6
­
1
­
2
­
1
1
­
3
1
NB
­
31.5%
­
13.3%
­
7.6%
­
3.7%
­
9.7%
­
21.5%
­
18.7%
­
20.0%
FB
­
37.8%
­
14.4%
­
8.2%
­
4.0%
­
10.8%
­
24.6%
­
24.4%
­
23.3%
NE
31.5%
13.3%
11.2%
5.3%
10.4%
21.5%
20.0%
20.0%
FE
37.8%
14.4%
11.9%
5.5%
11.5%
24.6%
25.6%
23.3%

Figure
E­
4.
1­
hour
Unpaired
Peak
Accuracy,
CAMS
23,
September
13
 
20,
1999.

Unpaired
Peak
Accuracy
for
AACOG
Modeled
Values
at
CAMS
23,
Sept.
13­
20,
1999
­
30%
­
20%
­
10%
0%
10%
20%
30%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
20%

EPA
Guidance
­
20%
E­
24
Figure
E­
5.
1­
hour
Unpaired
Peak
Accuracy,
CAMS
58,
September
13
 
20,
1999.

Figure
E­
6.
1­
hour
Unpaired
Peak
Accuracy,
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999.

Unpaired
Peak
Accuracy
for
AACOG
Modeled
Values
in
the
Central
Texas
Region,
Sept.
13­
20,
1999
­
30%
­
20%
­
10%
0%
10%
20%
30%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
20%

EPA
Guidance
­
20%
Unpaired
Peak
Accuracy
for
AACOG
Modeled
Values
at
CAMS
58,
Sept.
13­
20,
1999
­
30%
­
20%
­
10%
0%
10%
20%
30%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
20%

EPA
Guidance
­
20%
E­
25
Figure
E­
7.
1­
hour
Average
Paired
Peak
Accuracy
at
CAMS
23,
September
13­
20,
1999.

Figure
E­
8.
1­
hour
Average
Paired
Peak
Accuracy
at
CAMS
58,
September
13­
20,
1999.
Average
Paired
Peak
Accuracy
for
AACOG
Modeled
Values
at
CAMS
23,
Sept.
13­
20,
1999
­
40%
­
30%
­
20%
­
10%
0%
10%
20%
30%
40%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
Average
Paired
Peak
Accuracy
for
AACOG
Modeled
Values
at
CAMS
58,
Sept.
13­
20,
1999
­
40%
­
30%
­
20%
­
10%
0%
10%
20%
30%
40%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
E­
26
Figure
E­
9.
1­
hour
Average
Paired
Peak
Accuracy
at
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13­
20,
1999.

Figure
E­
10.
1­
hour
Normalized
Bias
at
CAMS
23,
September
13
 
20,
1999.
Average
Paired
Peak
Accuracy
for
AACOG
Modeled
Values
in
the
Central
Texas
Region,
Sept.
13­
20,
1999
­
40%
­
30%
­
20%
­
10%
0%
10%
20%
30%
40%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
Normalized
Bias
for
AACOG
Modeled
Values
at
CAMS
23,
Sept.
13­
20,
1999
­
35%
­
25%
­
15%
­
5%
5%
15%
25%
35%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
15%
EPA
Guidance
­
15%
E­
27
Figure
E­
11.
1­
hour
Normalized
Bias
at
CAMS
58,
September
13
 
20,
1999.

Figure
E­
12.
1­
hour
Normalized
Bias
at
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999.

Normalized
Bias
for
AACOG
Modeled
Values
in
the
Central
Texas
Region,
Sept.
13­
20,
1999
­
35%
­
25%
­
15%
­
5%
5%
15%
25%
35%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
15%
EPA
Guidance
­
15%
Normalized
Bias
for
AACOG
Modeled
Values
at
CAMS
58,
Sept.
13­
20,
1999
­
35%
­
25%
­
15%
­
5%
5%
15%
25%
35%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
15%
EPA
Guidance
­
15%
E­
28
Figure
E­
13.
1­
hour
Normalized
Error
at
CAMS
23,
September
13
 
20,
1999.

Figure
E­
14.
1­
hour
Normalized
Error
at
CAMS
58,
September
13
 
20,
1999.
Normalized
Error
for
AACOG
Modeled
Values
at
CAMS
23,
Sept.
13­
20,
1999
0%
5%
10%
15%
20%
25%
30%
35%
40%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
35%

Normalized
Error
for
AACOG
Modeled
Values
at
CAMS
58,
Sept.
13­
20,
1999
0%
5%
10%
15%
20%
25%
30%
35%
40%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
35%
E­
29
Figure
E­
15.
1­
hour
Normalized
Error
at
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999.

Figure
E­
16.
1­
hour
Timing
Bias
at
CAMS
23,
September
13
 
20,
1999.
Normalized
Error
for
AACOG
Modeled
Values
in
the
Central
Texas
Region,
Sept.
13­
20,
1999
0%
5%
10%
15%
20%
25%
30%
35%
40%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
35%

Bias
in
time
for
AACOG
Modeled
Values
at
CAMS
23,
Sept.
13­
20,
1999
­
6
­
5
­
4
­
3
­
2
­
1
0
1
2
3
4
5
9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Difference
in
Time
(
Hours)

Run
18.
sos.
e
E­
30
Figure
E­
17.
1­
hour
Timing
Bias
at
CAMS
58,
September
13
 
20,
1999.

Figure
E­
18.
1­
hour
Timing
Bias
at
Central
Texas
Monitors
23
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999.
Bias
in
time
for
AACOG
Modeled
Values
at
CAMS
58,
Sept.
13­
20,
1999
­
6
­
5
­
4
­
3
­
2
­
1
0
1
2
3
4
5
9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Difference
in
Time
(
Hours)

Run
18.
sos.
e
Bias
in
time
for
AACOG
Modeled
Values
in
the
Central
Texas
Region,
Sept.
13­
20,
1999
­
6
­
5
­
4
­
3
­
2
­
1
0
1
2
3
4
5
9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Difference
in
Time
(
Hours)

Run
18.
sos.
e
E­
31
Figure
E­
19.
1­
hour
Fractional
Bias
at
CAMS
23,
September
13
 
20,
1999.

Figure
E­
20.
1­
hour
Fractional
Bias
at
CAMS
58,
September
13
 
20,
1999.
Fractional
Bias
for
AACOG
Modeled
Values
at
CAMS
23,
Sept.
13­
20,
1999
­
35%
­
25%
­
15%
­
5%
5%
15%
25%
35%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
20%
EPA
Guidance
­
20%

Fractional
Bias
for
AACOG
Modeled
Values
at
CAMS
58,
Sept.
13­
20,
1999
­
35%
­
25%
­
15%
­
5%
5%
15%
25%
35%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
20%
EPA
Guidance
­
20%
E­
32
Figure
E­
21.
1­
hour
Fractional
Bias
at
Central
Texas
Monitors
23
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999.

Time
Series
Plots
In
addition
to
conducting
1­
hour
statistical
measurements,
staff
produced
time
series
plots
of
observed
versus
predicted
1­
hour
ozone
concentrations.
These
plots
provide
an
indication
of
the
model's
ability
to
replicate
peak
predictions,
the
presence
of
bias
within
the
diurnal
cycles,
as
well
as
biases
in
timing
of
the
predicted
daily
maxima.
Time
series
plots
for
the
four
San
Antonio
area
monitors
are
provided
in
figures
E­
22
through
E­
25.
These
graphs
indicate
that
diurnal
trends
were
replicated
quite
well,
although
peak
concentrations
were
slightly
under­
predicted
at
CAMS
23,
58,
and
678
during
several
episode
days.
Peak
concentrations
at
CAMS
59
were
consistently
under
predicted,
except
for
the
final
day
of
the
episode,
September
20th.
Fractional
Bias
for
AACOG
Modeled
Values
in
the
Central
Texas
Region,
Sept.
13­
20,
1999
­
35%
­
25%
­
15%
­
5%
5%
15%
25%
35%

9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
Percent
Difference
Run
18.
sos.
e
EPA
Guidance
+
20%
EPA
Guidance
­
20%
E­
33
Figure
E­
22.
Observed
versus
Predicted
1­
hour
Average
Ozone
Concentrations
at
CAMS
23,
September
13
 
20,
1999.

Figure
E­
23.
Observed
versus
Predicted
1­
hour
Average
Ozone
Concentrations
at
CAMS
58,
September
13
 
20,
1999.
0
20
40
60
80
100
120
9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
O
z
o
n
e
(
p
p
b
)
Min/
Max
Run.
18.
sos.
e
Predicted
Observed
0
20
40
60
80
100
120
9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
O
z
o
n
e
(
p
p
b
)
Min/
Max
Observed
Run.
18.
sos.
e
Predicted
E­
34
Figure
E­
24.
Observed
versus
Predicted
1­
hour
Average
Ozone
Concentrations
at
CAMS
59,
September
13
 
20,
1999.

0
20
40
60
80
100
120
9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
O
z
o
n
e
(
p
p
b
)
Min/
Max
Observed
Run.
18.
sos.
e
Predicted
Figure
E­
25.
Observed
versus
Predicted
1­
hour
Average
Ozone
Concentrations
at
CAMS
678,
September
13
 
20,
1999.

0
20
40
60
80
100
120
9/
13/
1999
9/
14/
1999
9/
15/
1999
9/
16/
1999
9/
17/
1999
9/
18/
1999
9/
19/
1999
9/
20/
1999
Date
O
z
o
n
e
(
p
p
b
)
Min/
Max
Observed
Run.
18.
sos.
e
Predicted
E­
35
Scatter
Plots
Scatter
plots
of
1­
hour
observed
(
x)
and
predicted
(
y)
data
for
the
four
SAER
monitors
are
provided
in
figures
E­
26
through
E­
29.
The
observed/
predicted
data
points
for
each
monitor
follows
the
1:
1
reference
line
fairly
well
and
each
plot
exhibits
moderate,
positive
correlation
coefficients.
Some
outlier
data
pairs
are
evident
in
each
chart,
however.

Figure
E­
26.
Scatter
Plot
of
1­
hour
Observed/
Predicted
Data
Pairs
at
CAMS
23.

y
=
0.6508x
+
19.189
R
2
=
0.7325
0
20
40
60
80
100
120
0
20
40
60
80
100
120
O3
Observed
(
ppb)
O
3
P
r
e
d
i
c
t
e
d
(
p
p
b
)

Figure
E­
27.
Scatter
Plot
of
1­
hour
Observed/
Predicted
Data
Pairs
at
CAMS
58.

y
=
0.5182x
+
32.088
R
2
=
0.6597
0
20
40
60
80
100
120
0
20
40
60
80
100
120
O3
Observed
(
ppb)
O
3
P
r
e
d
i
c
t
e
d
(
p
p
b
)
E­
36
Figure
E­
28.
Scatter
Plot
of
1­
hour
Observed/
Predicted
Data
Pairs
at
CAMS
59.

y
=
0.5277x
+
23.99
R
2
=
0.6597
0
20
40
60
80
100
120
0
20
40
60
80
100
120
O3
Observed
(
ppb)
O
3
P
r
e
d
i
c
t
e
d
(
p
p
b
)

Figure
E­
29.
Scatter
Plot
of
1­
hour
Observed/
Predicted
Data
Pairs
at
CAMS
678.

y
=
0.5152x
+
30.363
R
2
=
0.6785
0
20
40
60
80
100
120
0
20
40
60
80
100
120
O3
Observed
(
ppb)
O
3
P
r
e
d
i
c
t
e
d
(
p
p
b
)
E­
37
Weekend/
Weekday
Comparisons
Weekend/
weekday
analyses
may
be
useful
for
determining
whether
the
model
responds
appropriately
to
changes
in
precursor
emission
rates.
Typically,
EIs
for
weekdays
are
very
different
than
weekends.
For
example,
vehicle
miles
traveled
(
VMT)
are
generally
lower
on
weekends
than
weekdays.
In
areas
such
as
San
Antonio
where
mobile
sources
are
the
primary
source
of
NOx
emissions,
this
means
the
NOx
EIs
for
Saturday
and
Sunday
are
the
lowest
of
the
week.

Figure
E­
30
provides
a
comparison
between
predicted/
observed
1­
hour
average
ozone
concentrations
at
CAMS
23
and
the
daily
NOx
EI
for
Bexar
County.
As
shown,
observed
and
predicted
concentrations
track
fairly
closely
throughout
the
episode.
Both
curves
show
a
rise
in
ozone
concentrations
on
Saturday
as
the
result
of
lower
NOx
emissions
(
NOx
reduction
disbenefit).
Higher
ozone
concentrations
were
predicted
at
CAMS
58
(
figure
E­
31)
on
Saturday
and
Sunday,
although
actual
1­
hour
measurements
fell
on
Sunday.

Figure
E­
32
aggregates
the
predicted
peak
concentrations
(
September
13
 
20,
1999)
of
the
four
San
Antonio
area
monitors
into
a
single
chart
and
compares
the
data
to
the
daily
NOx
EI
for
Bexar
County.
Unlike
the
predictions
at
CAMS
23
and
58,
the
peak
predictions
at
CAMS
59
and
678
do
not
rise
as
sharply
between
Friday,
September
17th
and
Saturday,
September
18th.
These
results
are
consistent
with
expectations
since
CAMS
59
and
678
are
upwind
monitors.

Figure
E­
30.
Comparison
of
Observed/
predicted
1­
hour
Concentrations
at
CAMS
23
and
Daily
NOx
EI
for
Bexar
County,
September
13
 
20,
1999.

0
50
100
150
200
250
300
Mon
13­
Sep
Tue
14­
Sep
Wed
15­
Sep
Thu
16­
Sep
Fri
17­
Sep
Sat
18­
Sep
Sun
19­
Sep
Mon
20­
Sep
Total
NOx
(
tons/
day)

0
25
50
75
100
125
Ozone
(
ppb)
Total
NOx
Observed
CAMS
23
Predicted
CAMS
23
E­
38
Figure
E­
31.
Comparison
of
Observed/
predicted
1­
hour
Concentrations
at
CAMS
58
and
Daily
NOx
EI
for
Bexar
County,
September
13
 
20,
1999.

Figure
E­
32.
Comparison
of
Observed/
predicted
1­
hour
Concentrations
at
Four
San
Antonio
CAMS
and
Daily
NOx
EI
for
Bexar
County,
September
13
 
20,
1999.
0
50
100
150
200
250
300
Mon
13­
Sep
Tue
14­
Sep
Wed
15­
Sep
Thu
16­
Sep
Fri
17­
Sep
Sat
18­
Sep
Sun
19­
Sep
Mon
20­
Sep
Total
NOx
(
tons/
day)

0
25
50
75
100
125
Ozone
(
ppb)
Total
NOx
Observed
CAMS
58
Predicted
CAMS
58
0
50
100
150
200
250
300
Mon
13­
Sep
Tue
14­
Sep
Wed
15­
Sep
Thu
16­
Sep
Fri
17­
Sep
Sat
18­
Sep
Sun
19­
Sep
Mon
20­
Sep
Total
NOx
(
tons/
day)

0
25
50
75
100
125
Ozone
(
ppb)
Total
NOx
CAMS
58
CAMS
23
CAMS
678
CAMS
59
Adjusted
VMT
E­
39
Model
Performance:
8­
Hour
Average
Ozone
Concentrations
In
their
draft
8­
hour
guidance,
the
U.
S.
EPA
recommends
conducting
a
variety
of
tests
to
evaluate
a
photochemical
model.
In
broad
terms,
these
evaluations
include
performance
tests
and
diagnostic
analyses.
Both
types
of
evaluations
were
conducted
on
the
September
1999
simulation
with
excellent
results.
The
performance
tests,
used
to
determine
how
well
the
model
predicted
8­
hour
concentrations
are
described
in
the
following
sections
on
ozone
metrics,
graphic
analyses,
and
tile
plots
Ozone
Metrics
Metrics
are
used
to
evaluate
how
closely
predicted
ozone
concentrations
match
observations,
both
in
terms
of
spatial
and
temporal
distributions.
To
evaluate
the
performance
of
the
1999
base
case,
staff
conducted
statistical
tests
from
EPA's
1­
hour
and
8­
hour
guidance
documents.
To
apply
the
1­
hour
metrics
to
8­
hour
data,
staff
utilized
ENVIRON's
"
camxpost"
program,
described
in
the
1­
hour
statistics
section.
The
program
provided
unpaired
peak
accuracy
(
UPA),
average
paired
peak
accuracy
(
APPA),
peak
timing
bias
(
PTB),
normalized
bias
(
NB),
fractional
bias,
normalized
error
(
NE),
and
fractional
error
(
FE)
for
peak
8­
hour
data.
Results
of
these
8­
hour
tests
are
provided
in
tables
E­
18
through
E­
34.
Whenever
daily
modeled
predictions
were
less
than
60
ppb,
statistical
tests
were
not
performed.
These
days
are
indicated
by
"
N/
A"
in
the
appropriate
columns.

Yellow­
highlighted
values
in
these
tables
represent
statistics
that
fall
outside
EPA's
performance
criteria
on
primary
episode
days
(
September
15
 
20,
1999.
Although
it
is
not
necessary
to
conduct
these
tests
on
model
initialization
days
(
September
13­
14,
1999),
statistics
for
the
initialization
period
are
included
for
comparison
purposes.
Model
initialization
statistics
that
fall
outside
performance
thresholds
are
listed
in
bold
type.
Columns
where
data
are
missing
represent
days
in
which
predicted
measurements
were
less
than
60
ppb.

EPA
recommends
grouping
monitored
data
in
terms
of
location,
i.
e.,
downwind,
upwind,
and
city
center,
as
a
means
of
developing
useful
comparisons.
Averaged
metrics
tests
were
conducted
for
groups
of
monitors
when
possible.
2
The
Central
Texas
monitoring
network
is
relatively
sparse;
consequently,
the
Austin
and
San
Antonio
areas,
in
conjunction
with
the
TCEQ
and
the
U.
S.
EPA
Region
6,
recommended
evaluating
performance
based
on
averaged
data
from
all
Central
Texas
stations.
As
a
result,
metrics
tests
were
applied
to
averaged
results
at
all
eight
Central
Texas
monitors,
as
well
as
groups
of
monitors.
Monitor
numbers,
locations,
and
monitor
groups
are
as
follows:

2
Austin
has
two
monitors,
both
of
which
are
downwind.
San
Antonio
has
two
downwind,
two
upwind,
and
no
city
center
monitors.
As
a
consequence,
there
are
no
average
statistics
for
city
center
for
Austin,
San
Antonio
or
Central
Texas.
E­
40
CAMS
#
Monitor
Name/
Group
Name
Location
AIRS
#
3
Murchison
Travis
County
48­
453­
0014
4
Corpus
Christi
West
Nueces
County
48­
355­
0025
21
Corpus
Christi
Tuloso
Nueces
County
48­
355­
0026
23
San
Antonio
Northwest
Bexar
County
48­
029­
0032
38
Audubon
Travis
County
48­
453­
0020
58
Camp
Bullis
Bexar
County
48­
029­
0052
59
Calaveras
Lake
Bexar
County
48­
029­
0059
62
San
Marcos
Caldwell
County
48­
055­
0062
87
Victoria
Victoria
County
48­
469­
0003
601
Fayette
Fayette
County
48­
149­
0001
678
CPS
Pecan
Valley
Bexar
County
48­
029­
0055
3,
38
Austin
Downwind
Monitors
Travis
County
23,
58
San
Antonio
Downwind
Monitors
Bexar
County
59,
678
San
Antonio
Upwind
Monitors
Bexar
County
23,
58,
59,
678
San
Antonio
Monitors
Bexar
County
3,
23,
38,58
Central
Texas
Downwind
Mntrs
Bexar,
Travis
59,
62,
601,
678
Central
Texas
Upwind
Mntrs
Bexar,
Caldwell,
Fayette
3,
23,
38,
58,
59,
62,
601,
678
Central
Texas
Monitors
Bexar,
Caldwell,
Fayette,
Travis
Applying
the
same
performance
criteria
recommended
for
1­
hour
statistics
to
peak
8­
hour
observed/
predicted
data
comparisons,
the
results
are
excellent.
Although
a
few
individual
monitors
exhibit
results
that
fall
outside
EPA
thresholds
(
such
as
the
negative
bias
exhibited
by
the
coastal
monitors),
outcomes
for
monitor
groups
generally
fell
within
acceptable
ranges.
The
sole
exception
was
unpaired
peak
accuracy
on
September
18th
when
testing
paired
data
for
San
Antonio
upwind
monitors
and
Central
Texas
upwind
monitors.

Table
E­
18.
8­
hour
Statistics
for
CAMS
3,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
N/
A
18.9%
13.0%
­
0.9%
­
1.3%
­
0.9%
8.8%
APPA
N/
A
N/
A
1.2%
­
6.7%
­
14.3%
­
8.7%
­
4.4%
­
19.0%
PTB
N/
A
N/
A
0
1
0
1
0
0
NB
N/
A
N/
A
3.1%
­
4.8%
­
12.9%
­
10.9%
­
3.3%
­
18.2%
FB
N/
A
N/
A
3.0%
­
5.0%
­
13.8%
­
12.1%
­
3.5%
­
20.2%
NE
N/
A
N/
A
3.1%
5.6%
12.9%
11.4%
5.6%
18.2%
FE
N/
A
N/
A
3.0%
5.8%
13.8%
12.6%
5.7%
20.2%

Table
E­
19.
8­
hour
Statistics
for
CAMS
38,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
9.9%
3.3%
­
8.4%
­
11.5%
­
13.0%
­
11.1%
26.7%
APPA
N/
A
­
7.1%
­
9.5%
­
12.7%
­
13.7%
­
19.0%
­
18.3%
­
16.3%
PTB
N/
A
0
1
0
0
1
1
­
1
NB
N/
A
­
6.8%
­
6.0%
­
10.4%
­
15.4%
­
17.0%
­
10.7%
­
17.3%
FB
N/
A
­
7.1%
­
6.2%
­
11.0%
­
16.9%
­
19.0%
­
11.7%
­
19.1%
NE
N/
A
6.8%
6.0%
10.4%
15.4%
17.0%
10.9%
17.3%
FE
N/
A
7.1%
6.3%
11.0%
16.9%
19.0%
11.9%
19.1%
E­
41
Table
E­
20.
8­
hour
Statistics
for
Austin
Downwind
Monitors
(
CAMS
3
&
38),
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
9.9%
3.3%
­
8.4%
­
11.5%
­
13.0%
­
11.1%
8.8%
APPA
N/
A
­
7.1%
­
4.1%
­
9.7%
­
14.0%
­
13.8%
­
11.4%
­
17.6%
PTB
N/
A
0
1
1
0
1
1
­
1
NB
N/
A
­
6.8%
­
3.7%
­
8.7%
­
14.4%
­
14.8%
­
7.8%
­
17.7%
FB
N/
A
­
7.1%
­
3.9%
­
9.2%
­
15.7%
­
16.5%
­
8.5%
­
19.6%
NE
N/
A
6.8%
5.3%
9.0%
14.4%
15.0%
8.9%
17.7%
FE
N/
A
7.1%
5.5%
9.4%
15.7%
16.7%
9.5%
19.6%

Table
E­
21.
8­
hour
Statistics
for
CAMS
23,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
3.5%
1.1%
­
1.2%
13.6%
6.3%
13.4%
10.5%
APPA
N/
A
­
14.7%
­
12.9%
­
20.8%
0.0%
­
0.3%
­
5.0%
4.0%
PTB
N/
A
­
1
0
0
1
0
0
0
NB
N/
A
­
14.0%
­
9.3%
­
16.6%
­
7.9%
­
1.4%
­
2.0%
8.0%
FB
N/
A
­
15.1%
­
9.9%
­
18.3%
­
8.3%
­
1.5%
­
2.0%
7.6%
NE
N/
A
14.0%
9.3%
16.6%
7.9%
3.6%
3.6%
8.0%
FE
N/
A
15.1%
9.9%
18.3%
8.3%
3.6%
3.7%
7.6%

Table
E­
22.
8­
hour
Statistics
for
CAMS
58,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
9.0%
5.0%
8.1%
16.0%
2.0%
11.8%
14.2%
APPA
N/
A
­
6.3%
­
11.1%
­
5.9%
0.7%
­
11.1%
7.9%
­
0.8%
PTB
N/
A
0
0
0
1
0
0
­
1
NB
N/
A
­
5.3%
­
6.7%
­
3.1%
3.1%
­
7.8%
11.2%
1.2%
FB
N/
A
­
5.5%
­
7.0%
­
3.2%
3.0%
­
8.2%
10.4%
1.2%
NE
N/
A
5.3%
7.1%
4.3%
3.2%
7.8%
11.2%
2.9%
FE
N/
A
5.5%
7.4%
4.4%
3.2%
8.2%
10.4%
2.8%

Table
E­
23.
8­
hour
Statistics
for
CAMS
59,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
9.6%
1.9%
2.0%
­
6.1%
1.1%
21.6%
14.2%
10.8%
APPA
­
20.2%
­
18.6%
­
21.2%
­
13.3%
­
12.9%
­
19.7%
­
15.5%
­
7.2%
PTB
0
1
2
4
1
1
0
1
NB
­
19.6%
­
17.2%
­
16.8%
­
9.6%
­
6.8%
­
17.0%
­
9.6%
­
11.2%
FB
­
21.8%
­
18.9%
­
18.5%
­
10.3%
­
7.5%
­
18.7%
­
10.2%
­
12.1%
NE
19.6%
17.2%
16.8%
10.2%
10.1%
17.0%
9.6%
11.2%
FE
21.8%
18.9%
18.5%
11.0%
10.7%
18.7%
10.2%
12.1%
E­
42
Table
E­
24.
8­
hour
Statistics
for
CAMS
678,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
N/
A
12.6%
9.8%
31.1%
29.5%
20.1%
7.6%
APPA
N/
A
N/
A
­
10.1%
­
7.5%
2.6%
­
5.9%
­
4.8%
­
1.7%
PTB
N/
A
N/
A
1
0
­
1
0
0
1
NB
N/
A
N/
A
­
6.3%
­
5.2%
3.7%
­
4.0%
­
0.4%
0.1%
FB
N/
A
N/
A
­
6.5%
­
5.3%
3.6%
­
4.1%
­
0.6%
­
0.2%
NE
N/
A
N/
A
6.3%
5.2%
3.7%
4.1%
5.6%
6.7%
FE
N/
A
N/
A
6.5%
5.3%
3.6%
4.2%
5.6%
6.6%

Table
E­
25.
8­
hour
Statistics
for
San
Antonio
Downwind
Monitors
(
CAMS
23
&
58),
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
3.5%
1.1%
­
1.2%
13.6%
2.0%
11.8%
10.5%
APPA
N/
A
10.5%
­
12.0%
­
13.3%
­
4.1%
­
5.7%
1.5%
1.6%
PTB
N/
A
­
1
0
0
1
0
0
­
1
NB
N/
A
­
10.1%
­
7.7%
­
10.3%
­
2.4%
­
4.4%
4.6%
4.2%
FB
N/
A
­
10.8%
­
8.1%
­
11.2%
­
2.6%
­
4.7%
4.2%
4.2%
NE
N/
A
10.1%
8.0%
10.9%
5.6%
5.6%
7.4%
5.1%
FE
N/
A
10.8%
8.4%
11.8%
5.7%
5.8%
7.1%
4.9%

Table
E­
26.
8­
hour
Statistics
for
San
Antonio
Upwind
Monitors
(
CAMS
59
&
678),
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
9.6%
1.9%
2.0%
­
6.1%
1.1%
21.6%
14.2%
7.6%
APPA
­
20.2%
­
18.6%
­
15.6%
­
10.4%
­
5.2%
­
12.8%
­
10.2%
­
4.4%
PTB
0
1
2
2
0
1
0
1
NB
­
19.6%
­
17.2%
­
13.5%
­
8.5%
­
3.8%
­
12.4%
­
6.1%
­
6.2%
FB
­
21.8%
­
18.9%
­
14.7%
­
9.1%
­
4.3%
­
13.6%
­
6.5%
­
6.8%
NE
19.6%
17.2%
13.5%
9.0%
8.3%
12.5%
8.1%
9.2%
FE
21.8%
18.9%
14.7%
9.6%
8.7%
13.6%
8.4%
9.7%

Table
E­
27.
8­
hour
Statistics
for
all
San
Antonio
Monitors
(
CAMS
23,
58,
59
&
678)
September
13
 
20,
1999.
San
Antonio
Monitors
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
9.6%
3.5%
1.1%
­
1.2%
1.1%
2.0%
11.8%
7.6%
APPA
­
20.2%
­
13.2%
­
13.8%
­
11.9%
­
4.7%
­
9.3%
­
4.3%
­
1.4%
PTB
0
0
1
1
1
0
0
0
NB
­
19.6%
­
12.7%
­
10.9%
­
9.3%
­
3.1%
­
8.5%
­
0.9%
­
1.3%
FB
­
21.8%
­
13.7%
­
11.8%
­
10.0%
­
3.5%
­
9.2%
­
1.3%
­
1.8%
NE
19.6%
12.7%
11.0%
9.8%
6.9%
9.1%
7.7%
7.3%
FE
21.8%
13.7%
11.9%
10.5%
7.2%
9.8%
7.8%
7.4%
E­
43
Table
E­
28.
8­
hour
Statistics
for
CAMS
601,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
N/
A
6.3%
28.9%
­
0.6%
3.2%
11.5%
17.0%
APPA
N/
A
N/
A
­
2.2%
1.2%
­
14.4%
­
5.8%
­
3.0%
1.7%
PTB
N/
A
N/
A
­
1
1
­
1
1
­
3
0
NB
N/
A
N/
A
­
3.5%
1.2%
­
15.4%
­
12.2%
­
11.1%
­
3.2%
FB
N/
A
N/
A
­
3.6%
1.2%
­
16.9%
­
13.2%
­
12.2%
­
3.3%
NE
N/
A
N/
A
3.5%
1.4%
15.4%
12.2%
11.5%
3.6%
FE
N/
A
N/
A
3.6%
1.4%
16.9%
13.2%
12.7%
3.7%

Table
E­
29.
8­
hour
Statistics
for
CAMS
62,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
N/
A
15.1%
13.1%
2.9%
13.0%
10.1%
24.5%
APPA
N/
A
N/
A
­
6.1%
6.8%
­
13.3%
­
11.0%
­
6.7%
3.3%
PTB
N/
A
N/
A
­
1
1
6
1
0
0
NB
N/
A
N/
A
­
2.1%
6.9%
­
7.7%
­
8.3%
­
2.8%
4.2%
FB
N/
A
N/
A
­
2.1%
6.6%
­
8.6%
­
8.7%
­
3.0%
4.1%
NE
N/
A
N/
A
3.3%
6.9%
11.8%
8.3%
5.8%
4.2%
FE
N/
A
N/
A
3.3%
6.6%
12.5%
8.7%
5.9%
4.1%

Table
E­
30.
8­
hour
Statistics
for
Central
Texas
Downwind
Monitors
(
CAMS
3,
23,
38,
58)
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
N/
A
3.5%
1.1%
­
8.4%
­
11.5%
­
13.0%
­
11.1%
8.8%
APPA
N/
A
­
9.4%
­
8.1%
­
11.5%
­
9.1%
­
9.8%
­
5.0%
­
8.0%
PTB
N/
A
0
0
0
1
1
0
­
1
NB
N/
A
­
9.3%
­
5.4%
­
9.5%
­
10.1%
­
10.9%
­
3.0%
­
8.5%
FB
N/
A
­
10.8%
­
8.1%
­
11.2%
­
2.6%
­
4.7%
4.2%
4.0%
NE
N/
A
9.3%
6.4%
9.9%
11.2%
11.4%
8.3%
12.4%
FE
N/
A
9.9%
6.7%
10.5%
12.1%
12.6%
8.6%
13.4%

Table
E­
31.
8­
hour
Statistics
for
Central
Texas
Upwind
Monitors
(
CAMS
59,
62,
601,
678)
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
9.6%
1.9%
2.0%
­
6.1%
­
0.6%
21.6%
14.2%
7.6%
APPA
­
20.2%
­
18.6%
­
9.6%
­
3.0%
­
8.4%
­
10.1%
­
7.1%
3.2%
PTB
0
1
0
2
1
1
­
1
1
NB
­
19.6%
­
17.2%
­
8.5%
­
3.9%
­
8.5%
­
11.0%
­
6.9%
­
3.1%
FB
­
21.8%
­
18.9%
­
9.2%
­
4.3%
­
9.4%
­
11.9%
­
7.6%
­
3.4%
NE
19.6%
17.2%
8.8%
6.9%
12.2%
11.0%
8.9%
6.5%
FE
21.8%
18.9%
9.5%
7.3%
13.0%
11.9%
9.3%
6.7%
E­
44
Table
E­
32.
8­
hour
Statistics
for
Central
Texas
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
9.6%
3.5%
1.1%
­
8.4%
­
11.5%
­
13.0%
­
11.1%
8.8%
APPA
­
20.2%
­
11.7%
­
9.0%
­
7.3%
­
9.3%
­
10.2%
­
6.2%
­
4.5%
PTB
0
0
0
1
1
1
0
0
NB
­
19.6%
­
11.6%
­
7.1%
­
6.7%
­
9.5%
­
11.0%
­
4.9%
5.9%
FB
21.8%
­
12.5%
­
7.6%
­
7.3%
­
10.4%
­
12.1%
­
5.5%
­
6.7%
NE
19.6%
11.6%
7.7%
8.4%
11.5%
11.3%
8.5%
9.5%
FE
21.8%
12.5%
8.2%
8.9%
12.3%
12.4%
9.0%
10.1%

Table
E­
33.
8­
hour
Statistics
for
Corpus
Christi
Monitors
(
CAMS
4
&
21),
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
4.1%
­
5.5%
0.8%
­
5.3%
­
3.6%
­
1.4%
1.6%
­
1.2%
APPA
­
25.7%
­
21.7%
­
22.2%
­
22.6%
­
19.6%
­
10.1%
­
6.1%
­
22.6%
PTB
0
1
1
0
­
2
­
1
2
2
NB
­
27.2%
­
22.4%
­
19.9%
­
23.4%
­
18.8%
­
14.1%
­
7.6%
­
24.2%
FB
­
31.7%
­
25.5%
­
22.1%
­
26.5%
­
20.9%
­
15.7%
­
8.4%
­
27.8%
NE
27.2%
22.4%
19.9%
23.4%
18.8%
14.3%
9.2%
24.2%
FE
31.7%
25.5%
22.1%
26.5%
20.9%
15.9%
9.9%
27.8%

Table
E­
34.
8­
hour
Statistics
for
CAMS
87,
September
13
 
20,
1999.
Statistic
13­
Sep
14­
Sep
15­
Sep
16­
Sep
17­
Sep
18­
Sep
19­
Sep
20­
Sep
UPA
­
4.5%
9.8%
13.2%
10.0%
8.2%
0.6%
0.4%
­
15.9%
APPA
­
31.2%
­
12.9%
­
6.4%
­
3.1%
­
5.3%
­
15.2%
­
4.3%
­
18.9%
PTB
0
­
1
­
1
0
1
­
1
0
1
NB
­
28.8%
­
12.8%
­
6.0%
­
1.3%
­
7.9%
­
18.8%
­
16.4%
­
15.6%
­
33.6%
­
13.7%
­
6.3%
­
1.4%
­
8.3%
­
20.8%
­
19.4%
­
17.6%
NE
28.8%
12.8%
6.3%
2.7%
7.9%
18.8%
16.4%
17.2%
FE
33.6%
13.7%
6.6%
2.8%
8.3%
20.8%
19.4%
19.1%

A
variety
of
analyses
are
specifically
recommended
by
the
EPA
in
their
draft
guidance
on
8­
hour
attainment
demonstrations
(
EPA
1999).
These
include:
 
Bias
between
spatially
paired
means
of
observations
and
predictions
of
8­
hour
daily
maximum
ozone
concentrations,
with
predicted
values
based
on
grid
cells
`
near'
a
monitor.
 
Correlation
coefficient
and
scatter
plot
for
average
observed
and
predicted
8­
hour
daily
maximum
ozone
concentrations.
 
Temporal
correlation
coefficient
of
observed
and
nearby
predicted
8­
hour
daily
maximum
ozone
concentrations,
which
are
spatially
averaged.
If
the
monitoring
network
is
sufficiently
large,
concentrations
should
be
grouped
into
upwind,
downwind,
and
center
city
locations.
 
Quantile­
quantile
plots
 
Fractional
bias
E­
45
The
EPA­
recommended
performance
goal
for
each
test
is:

Test
Goal
Bias
between
predicted/
observed
mean
8­
hour
(
and
1­
hour)
daily
maxima
near
each
monitor
20%
most
monitors
(
8­
hr
comparisons
only)

Fractional
bias
between
predicted/
Observed
mean
8­
hour
(
and
1­
hour)
daily
maxima
near
each
monitor
20%
most
monitors
(
8­
hr
comparisons
only)

Correlation
coefficients,
all
data,
temporally
paired
means,
spatially
paired
means
Moderate
to
large
positive
correlation
Bias
(
8­
hour
daily
maxima
and
1­
hour
observed/
predicted),
all
monitors
5
 
15%

Gross
error
(
8­
hour
daily
maxima
and
1­
hour
observed/
predicted),
all
monitors
30
 
35%

Staff
conducted
metric,
scatter
plot,
and
quantile­
quantile
(
Q­
Q)
plot
analyses
using
a
FORTRAN
program
developed
by
ENVIRON
International,
Inc.
Due
to
some
uncertainty
as
to
the
most
appropriate
means
of
calculating
these
evaluations,
ENVIRON's
statistical
program
performs
the
calculations
using
three
different
methodologies:
1)
The
predicted
daily
maximum
ozone
concentration
within
grid
cells
near
a
monitor;
3
2)
The
predicted
daily
maximum
ozone
concentration
within
grid
cells
near
a
monitor
that
is
closest
in
magnitude
to
the
observed
daily
maximum
at
the
monitor;
and
3)
A
bilinear
interpolation
of
predicted
daily
maximum
ozone
concentration
around
the
monitor
location.

Normalized
bias,
fractional
bias,
normalized
error,
and
fractional
error
calculations
for
all
monitors
and
monitoring
groups
in
the
4­
km
subdomain
are
provided
in
tables
E­
35
through
E­
51.
Each
table
presents
the
statistical
results
for
all
three
methodologies.

To
compare
8­
hour
normalized
bias
statistics
more
readily,
line
graphs
were
created
of
downwind,
upwind,
and
coastal
monitors
for
each
methodology
(
figures
E­
33
through
E­
35).
Episode
day
is
designated
by
a
number
(
1
=
9/
13/
99,
2
=
9/
14/
99,
3
=
9/
15/
99,
4
=
9/
16/
99,
5
=
9/
17/
99,
6
=
9/
18/
99,
7
=
9/
19/
99,
and
8
=
9/
20/
99)
in
each
graph.
The
model
initialization
period
(
days
1
and
2)
are
separated
by
a
dashed
black
line
and
performance
goals
are
designated
by
dashed
red
lines.

Method
1
for
calculating
normalized
bias
yields
very
good
results.
Discounting
the
model
initialization
period,
the
only
days
when
EPA
performance
goals
were
not
met
using
3
In
accordance
with
EPA
guidance,
grid
cells
`
near'
a
monitor
were
defined
as
a
7x7
array
of
cells
(
U.
S.
EPA,
1999).
E­
46
method
1were
September
15th
(
CAMS
59
 
San
Antonio
upwind
monitor),
and
September
19th
(
CAMS
23
and
58
 
San
Antonio
downwind
monitors).

Model
performance
based
on
the
statistical
metrics
was
best
when
calculated
using
method
2.
The
±
20%
performance
goal
for
normalized
bias
was
only
exceeded
once
using
this
method:
September
15th
at
CAMS
59.

Method
3
yielded
the
most
incidences
where
performance
goals
were
not
met.
With
the
exception
of
CAMS
23,
the
normalized
bias
statistics
for
each
downwind
monitor
failed
performance
goals
during
at
least
one
primary
episode
day.
These
include
CAMS
38
(
Austin)
on
September
16th,
18th,
and
20th,
CAMS
3
(
Austin)
on
September
20th,
and
CAMS
58
(
San
Antonio)
on
September
19th.
Both
Austin
monitors
(
CAMS
3
&
38)
are
downwind
monitors
and
both
exhibited
a
negative
bias
during
the
modeling
episode.
Normalized
bias
calculated
using
method
3
yielded
better
results
for
the
upwind
monitors.
The
only
days
when
performance
goals
were
not
met
were
September
15th
and
18th,
both
at
CAMS
59
(
San
Antonio).
In
addition,
all
three
coastal
monitors
failed
the
normalized
bias
goal
with
method
3
on
at
least
one
episode
day:
CAMS
4
(
Corpus
Christi)
on
September
15th
and
16th,
CAMS
21
(
Corpus
Christi)
on
September
15th,
and
CAMS
87
(
Victoria)
on
September
20th.

Line
graphs
of
fractional
bias
statistics
were
also
developed
for
each
methodology.
These
graphs
are
presented
in
figures
E­
36
through
E­
38.
Results
of
the
fractional
bias
calculations
are
very
similar
to
those
for
normalized
bias,
although
the
incidence
of
results
that
failed
to
meet
performance
goals
increased
somewhat
with
fractional
bias.
For
the
daily
peak
predicted/
observed
statistics
(
September
15
 
20,
1999)
using
Method
1,
for
example,
the
performance
goal
for
normalized
bias
was
met
on
95%
of
the
days
and
the
performance
goal
for
fractional
bias
was
met
on
94%
of
the
days.
Using
method
2,
the
±
20%
goal
for
normalized
and
fractional
bias
was
met
on
98%
and
97%
of
the
days,
respectively.
The
poorest
performance
was
associated
with
evaluating
the
model
using
method
3.
Even
using
this
strictest
methodology,
however,
the
performance
goals
for
normalized
and
fractional
bias
were
passed
on
85%
of
the
primary
episode
days.
E­
47
Table
E­
35.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
3,
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
53.00
56.21
6.06
5.88
6.06
5.88
9/
14/
1999
56.00
70.02
25.04
22.25
25.04
22.25
9/
15/
1999
78.00
89.62
14.90
13.86
14.90
13.86
9/
16/
1999
73.00
79.73
9.22
8.81
9.22
8.81
9/
17/
1999
98.00
102.26
4.35
4.25
4.35
4.25
9/
18/
1999
97.00
98.17
1.21
1.20
1.21
1.20
9/
19/
1999
101.00
104.52
3.49
3.43
3.49
3.43
9/
20/
1999
102.00
103.21
1.19
1.18
1.19
1.18
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
53.00
52.95
­
0.09
­
0.09
0.09
0.09
9/
14/
1999
56.00
55.79
­
0.38
­
0.38
0.38
0.38
9/
15/
1999
78.00
78.05
0.06
0.06
0.06
0.06
9/
16/
1999
73.00
73.03
0.04
0.04
0.04
0.04
9/
17/
1999
98.00
97.40
­
0.61
­
0.61
0.61
0.61
9/
18/
1999
97.00
97.00
0.00
0.00
0.00
0.00
9/
19/
1999
101.00
100.89
­
0.11
­
0.11
0.11
0.11
9/
20/
1999
102.00
103.21
1.19
1.18
1.19
1.18
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
53.00
51.49
­
2.85
­
2.89
2.85
2.89
9/
14/
1999
56.00
58.35
4.20
4.11
4.20
4.11
9/
15/
1999
78.00
79.62
2.08
2.06
2.08
2.06
9/
16/
1999
73.00
68.84
­
5.70
­
5.87
5.70
5.87
9/
17/
1999
98.00
88.69
­
9.50
­
9.97
9.50
9.97
9/
18/
1999
97.00
90.65
­
6.55
­
6.77
6.55
6.77
9/
19/
1999
101.00
98.81
­
2.17
­
2.19
2.17
2.19
9/
20/
1999
102.00
72.45
­
28.97
­
33.88
28.97
33.88
E­
48
Table
E­
36.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
38,
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
57.00
56.73
­
0.47
­
0.47
0.47
0.47
9/
14/
1999
66.00
64.87
­
1.71
­
1.73
1.71
1.73
9/
15/
1999
83.00
86.06
3.69
3.62
3.69
3.62
9/
16/
1999
100.00
82.48
­
17.52
­
19.20
17.52
19.20
9/
17/
1999
109.00
102.26
­
6.18
­
6.38
6.18
6.38
9/
18/
1999
120.00
98.17
­
18.19
­
20.01
18.19
20.01
9/
19/
1999
110.00
104.63
­
4.88
­
5.00
4.88
5.00
9/
20/
1999
83.00
72.89
­
12.18
­
12.97
12.18
12.97
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
57.00
56.73
­
0.47
­
0.47
0.47
0.47
9/
14/
1999
66.00
64.87
­
1.71
­
1.73
1.71
1.73
9/
15/
1999
83.00
83.16
0.19
0.19
0.19
0.19
9/
16/
1999
100.00
82.48
­
17.52
­
19.20
17.52
19.20
9/
17/
1999
109.00
102.26
­
6.18
­
6.38
6.18
6.38
9/
18/
1999
120.00
98.17
­
18.19
­
20.01
18.19
20.01
9/
19/
1999
110.00
104.63
­
4.88
­
5.00
4.88
5.00
9/
20/
1999
83.00
72.89
­
12.18
­
12.97
12.18
12.97
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
57.00
55.57
­
2.51
­
2.54
2.51
2.54
9/
14/
1999
66.00
59.72
­
9.52
­
9.99
9.52
9.99
9/
15/
1999
83.00
77.19
­
7.00
­
7.25
7.00
7.25
9/
16/
1999
100.00
79.95
­
20.05
­
22.28
20.05
22.28
9/
17/
1999
109.00
95.70
­
12.20
­
12.99
12.20
12.99
9/
18/
1999
120.00
87.69
­
26.93
­
31.11
26.93
31.11
9/
19/
1999
110.00
99.42
­
9.62
­
10.10
9.62
10.10
9/
20/
1999
83.00
64.17
­
22.69
­
25.59
22.69
25.59
E­
49
Table
E­
37.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
Austin
monitors
 
CAMS
3
&
38
(
both
are
downwind
monitors),
September
13
 
20,
1999.

Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
55.00
56.47
2.79
2.70
3.27
3.18
9/
14/
1999
61.00
67.45
11.66
10.26
13.37
11.99
9/
15/
1999
80.50
87.84
9.29
8.74
9.29
8.74
9/
16/
1999
86.50
81.11
­
4.15
­
5.19
13.37
14.01
9/
17/
1999
103.50
102.26
­
0.92
­
1.06
5.27
5.32
9/
18/
1999
108.50
98.17
­
8.49
­
9.41
9.70
10.61
9/
19/
1999
105.50
104.58
­
0.70
­
0.79
4.18
4.21
9/
20/
1999
92.50
88.05
­
5.50
­
5.90
6.68
7.07
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
55.00
54.84
­
0.28
­
0.28
0.28
0.28
9/
14/
1999
61.00
60.33
­
1.04
­
1.05
1.04
1.05
9/
15/
1999
80.50
80.61
0.13
0.13
0.13
0.13
9/
16/
1999
86.50
77.76
­
8.74
­
9.58
8.78
9.62
9/
17/
1999
103.50
99.83
­
3.40
­
3.50
3.40
3.50
9/
18/
1999
108.50
97.59
­
9.10
­
10.01
9.10
10.01
9/
19/
1999
105.50
102.76
­
2.50
­
2.56
2.50
2.56
9/
20/
1999
92.50
88.05
­
5.50
­
5.90
6.68
7.07
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
55.00
53.53
­
2.68
­
2.72
2.68
2.72
9/
14/
1999
61.00
59.04
­
2.66
­
2.94
6.86
7.05
9/
15/
1999
80.50
78.41
­
2.46
­
2.60
4.54
4.65
9/
16/
1999
86.50
74.40
­
12.87
­
14.07
12.87
14.07
9/
17/
1999
103.50
92.20
­
10.85
­
11.48
10.85
11.48
9/
18/
1999
108.50
89.17
­
16.74
­
18.94
16.74
18.94
9/
19/
1999
105.50
99.12
­
5.89
­
6.15
5.89
6.15
9/
20/
1999
92.50
68.31
­
25.83
­
29.73
25.83
29.73
E­
50
Table
E­
38.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
23,
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
66.00
56.73
­
14.05
­
15.11
14.05
15.11
9/
14/
1999
75.00
73.79
­
1.61
­
1.63
1.61
1.63
9/
15/
1999
92.00
97.21
5.66
5.51
5.66
5.51
9/
16/
1999
93.00
88.59
­
4.74
­
4.86
4.74
4.86
9/
17/
1999
82.00
91.47
11.55
10.92
11.55
10.92
9/
18/
1999
102.00
110.88
8.71
8.34
8.71
8.34
9/
19/
1999
94.00
118.52
26.09
23.08
26.09
23.08
9/
20/
1999
106.00
112.50
6.13
5.95
6.13
5.95
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
66.00
56.73
­
14.05
­
15.11
14.05
15.11
9/
14/
1999
75.00
73.79
­
1.61
­
1.63
1.61
1.63
9/
15/
1999
92.00
92.01
0.01
0.01
0.01
0.01
9/
16/
1999
93.00
88.59
­
4.74
­
4.86
4.74
4.86
9/
17/
1999
82.00
82.29
0.35
0.35
0.35
0.35
9/
18/
1999
102.00
102.16
0.16
0.16
0.16
0.16
9/
19/
1999
94.00
94.69
0.73
0.73
0.73
0.73
9/
20/
1999
106.00
105.99
­
0.01
­
0.01
0.01
0.01
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
66.00
54.88
­
16.85
­
18.40
16.85
18.40
9/
14/
1999
75.00
65.41
­
12.79
­
13.66
12.79
13.66
9/
15/
1999
92.00
82.80
­
10.00
­
10.53
10.00
10.53
9/
16/
1999
93.00
77.74
­
16.41
­
17.88
16.41
17.88
9/
17/
1999
82.00
77.80
­
5.12
­
5.26
5.12
5.26
9/
18/
1999
102.00
105.79
3.72
3.65
3.72
3.65
9/
19/
1999
94.00
90.81
­
3.39
­
3.45
3.39
3.45
9/
20/
1999
106.00
105.71
­
0.27
­
0.27
0.27
0.27
E­
51
Table
E­
39.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
58,
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
61.00
57.07
­
6.44
­
6.66
6.44
6.66
9/
14/
1999
67.00
70.82
5.70
5.54
5.70
5.54
9/
15/
1999
88.00
87.19
­
0.92
­
0.92
0.92
0.92
9/
16/
1999
83.00
86.38
4.07
3.99
4.07
3.99
9/
17/
1999
82.00
91.67
11.79
11.14
11.79
11.14
9/
18/
1999
108.00
110.88
2.67
2.63
2.67
2.63
9/
19/
1999
96.00
118.52
23.46
21.00
23.46
21.00
9/
20/
1999
100.00
106.71
6.71
6.49
6.71
6.49
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
61.00
57.07
­
6.44
­
6.66
6.44
6.66
9/
14/
1999
67.00
67.01
0.01
0.01
0.01
0.01
9/
15/
1999
88.00
87.19
­
0.92
­
0.92
0.92
0.92
9/
16/
1999
83.00
83.06
0.07
0.07
0.07
0.07
9/
17/
1999
82.00
81.93
­
0.09
­
0.09
0.09
0.09
9/
18/
1999
108.00
108.53
0.49
0.49
0.49
0.49
9/
19/
1999
96.00
95.72
­
0.29
­
0.29
0.29
0.29
9/
20/
1999
100.00
99.51
­
0.49
­
0.49
0.49
0.49
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
61.00
53.22
­
12.75
­
13.62
12.75
13.62
9/
14/
1999
67.00
63.68
­
4.96
­
5.08
4.96
5.08
9/
15/
1999
88.00
73.64
­
16.32
­
17.77
16.32
17.77
9/
16/
1999
83.00
80.83
­
2.61
­
2.65
2.61
2.65
9/
17/
1999
82.00
82.77
0.94
0.93
0.94
0.93
9/
18/
1999
108.00
94.55
­
12.45
­
13.28
12.45
13.28
9/
19/
1999
96.00
115.42
20.23
18.37
20.23
18.37
9/
20/
1999
100.00
90.40
­
9.60
­
10.08
9.60
10.08
E­
52
Table
E­
40.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
59,
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
72.00
62.13
­
13.71
­
14.72
13.71
14.72
9/
14/
1999
67.00
56.70
­
15.37
­
16.65
15.37
16.65
9/
15/
1999
91.00
70.87
­
22.12
­
24.87
22.12
24.87
9/
16/
1999
85.00
76.10
­
10.47
­
11.05
10.47
11.05
9/
17/
1999
82.00
73.04
­
10.93
­
11.56
10.93
11.56
9/
18/
1999
84.00
77.42
­
7.83
­
8.15
7.83
8.15
9/
19/
1999
96.00
90.71
­
5.51
­
5.67
5.51
5.67
9/
20/
1999
97.00
103.84
7.05
6.81
7.05
6.81
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
72.00
62.13
­
13.71
­
14.72
13.71
14.72
9/
14/
1999
67.00
56.70
­
15.37
­
16.65
15.37
16.65
9/
15/
1999
91.00
70.87
­
22.12
­
24.87
22.12
24.87
9/
16/
1999
85.00
76.10
­
10.47
­
11.05
10.47
11.05
9/
17/
1999
82.00
73.04
­
10.93
­
11.56
10.93
11.56
9/
18/
1999
84.00
77.42
­
7.83
­
8.15
7.83
8.15
9/
19/
1999
96.00
90.71
­
5.51
­
5.67
5.51
5.67
9/
20/
1999
97.00
97.01
0.01
0.01
0.01
0.01
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
72.00
57.06
­
20.75
­
23.15
20.75
23.15
9/
14/
1999
67.00
52.09
­
22.25
­
25.04
22.25
25.04
9/
15/
1999
91.00
66.07
­
27.40
­
31.74
27.40
31.74
9/
16/
1999
85.00
73.35
­
13.71
­
14.71
13.71
14.71
9/
17/
1999
82.00
67.48
­
17.71
­
19.43
17.71
19.43
9/
18/
1999
84.00
65.96
­
21.48
­
24.06
21.48
24.06
9/
19/
1999
96.00
79.28
­
17.42
­
19.08
17.42
19.08
9/
20/
1999
97.00
99.84
2.93
2.89
2.93
2.89
E­
53
Table
E­
41.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
678,
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
67.00
63.05
­
5.90
­
6.07
5.90
6.07
9/
14/
1999
64.00
60.99
­
4.70
­
4.82
4.70
4.82
9/
15/
1999
79.00
78.00
­
1.27
­
1.27
1.27
1.27
9/
16/
1999
81.00
76.83
­
5.15
­
5.28
5.15
5.28
9/
17/
1999
70.00
75.71
8.16
7.84
8.16
7.84
9/
18/
1999
84.00
88.35
5.18
5.05
5.18
5.05
9/
19/
1999
91.00
96.07
5.57
5.42
5.57
5.42
9/
20/
1999
107.00
108.27
1.19
1.18
1.19
1.18
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
67.00
63.05
­
5.90
­
6.07
5.90
6.07
9/
14/
1999
64.00
60.99
­
4.70
­
4.82
4.70
4.82
9/
15/
1999
79.00
78.00
­
1.27
­
1.27
1.27
1.27
9/
16/
1999
81.00
76.83
­
5.15
­
5.28
5.15
5.28
9/
17/
1999
70.00
70.14
0.20
0.20
0.20
0.20
9/
18/
1999
84.00
85.89
2.25
2.22
2.25
2.22
9/
19/
1999
91.00
91.06
0.07
0.07
0.07
0.07
9/
20/
1999
107.00
107.02
0.02
0.02
0.02
0.02
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
67.00
61.47
­
8.25
­
8.61
8.25
8.61
9/
14/
1999
64.00
56.38
­
11.91
­
12.66
11.91
12.66
9/
15/
1999
79.00
69.83
­
11.61
­
12.32
11.61
12.32
9/
16/
1999
81.00
70.67
­
12.75
­
13.62
12.75
13.62
9/
17/
1999
70.00
69.50
­
0.71
­
0.72
0.71
0.72
9/
18/
1999
84.00
77.14
­
8.17
­
8.51
8.17
8.51
9/
19/
1999
91.00
85.83
­
5.68
­
5.85
5.68
5.85
9/
20/
1999
107.00
96.44
­
9.87
­
10.38
9.87
10.38
E­
54
Table
E­
42.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
San
Antonio
downwind
monitors
(
CAMS
23
&
58),
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
63.50
56.90
­
10.24
­
10.88
10.24
10.88
9/
14/
1999
71.00
72.31
2.04
1.96
3.66
3.58
9/
15/
1999
90.00
92.20
2.37
2.29
3.29
3.22
9/
16/
1999
88.00
87.49
­
0.33
­
0.43
4.41
4.42
9/
17/
1999
82.00
91.57
11.67
11.03
11.67
11.03
9/
18/
1999
105.00
110.88
5.69
5.49
5.69
5.49
9/
19/
1999
95.00
118.52
24.77
22.04
24.77
22.04
9/
20/
1999
103.00
109.61
6.42
6.22
6.42
6.22
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
63.50
56.90
­
10.24
­
10.88
10.24
10.88
9/
14/
1999
71.00
70.40
­
0.80
­
0.81
0.81
0.82
9/
15/
1999
90.00
89.60
­
0.45
­
0.46
0.47
0.47
9/
16/
1999
88.00
85.83
­
2.33
­
2.39
2.41
2.46
9/
17/
1999
82.00
82.11
0.13
0.13
0.22
0.22
9/
18/
1999
105.00
105.35
0.32
0.32
0.32
0.32
9/
19/
1999
95.00
95.21
0.22
0.22
0.51
0.51
9/
20/
1999
103.00
102.75
­
0.25
­
0.25
0.25
0.25
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
63.50
54.05
­
14.80
­
16.01
14.80
16.01
9/
14/
1999
71.00
64.55
­
8.87
­
9.37
8.87
9.37
9/
15/
1999
90.00
78.22
­
13.16
­
14.15
13.16
14.15
9/
16/
1999
88.00
79.29
­
9.51
­
10.26
9.51
10.26
9/
17/
1999
82.00
80.29
­
2.09
­
2.16
3.03
3.10
9/
18/
1999
105.00
100.17
­
4.37
­
4.82
8.08
8.46
9/
19/
1999
95.00
103.12
8.42
7.46
11.81
10.91
9/
20/
1999
103.00
98.06
­
4.94
­
5.18
4.94
5.18
E­
55
Table
E­
43.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
San
Antonio
upwind
monitors
(
CAMS
59
&
678),
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
69.50
62.59
­
9.80
­
10.40
9.80
10.40
9/
14/
1999
65.50
58.85
­
10.04
­
10.73
10.04
10.73
9/
15/
1999
85.00
74.44
­
11.69
­
13.07
11.69
13.07
9/
16/
1999
83.00
76.47
­
7.81
­
8.17
7.81
8.17
9/
17/
1999
76.00
74.38
­
1.38
­
1.86
9.54
9.70
9/
18/
1999
84.00
82.89
­
1.33
­
1.55
6.51
6.60
9/
19/
1999
93.50
93.39
0.03
­
0.12
5.54
5.54
9/
20/
1999
102.00
106.06
4.12
4.00
4.12
4.00
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
69.50
62.59
­
9.80
­
10.40
9.80
10.40
9/
14/
1999
65.50
58.85
­
10.04
­
10.73
10.04
10.73
9/
15/
1999
85.00
74.44
­
11.69
­
13.07
11.69
13.07
9/
16/
1999
83.00
76.47
­
7.81
­
8.17
7.81
8.17
9/
17/
1999
76.00
71.59
­
5.36
­
5.68
5.56
5.88
9/
18/
1999
84.00
81.66
­
2.79
­
2.96
5.04
5.19
9/
19/
1999
93.50
90.89
­
2.72
­
2.80
2.79
2.87
9/
20/
1999
102.00
102.02
0.01
0.01
0.01
0.01
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
69.50
59.27
­
14.50
­
15.88
14.50
15.88
9/
14/
1999
65.50
54.24
­
17.08
­
18.85
17.08
18.85
9/
15/
1999
85.00
67.95
­
19.50
­
22.03
19.50
22.03
9/
16/
1999
83.00
72.01
­
13.23
­
14.17
13.23
14.17
9/
17/
1999
76.00
68.49
­
9.21
­
10.07
9.21
10.07
9/
18/
1999
84.00
71.55
­
14.82
­
16.29
14.82
16.29
9/
19/
1999
93.50
82.56
­
11.55
­
12.46
11.55
12.46
9/
20/
1999
102.00
98.14
­
3.47
­
3.75
6.40
6.63
E­
56
Table
E­
44.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
San
Antonio
monitors
(
CAMS
23,
58,
59,
678),
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
66.50
59.75
­
10.02
­
10.64
10.02
10.64
9/
14/
1999
68.25
65.58
­
4.00
­
4.39
6.85
7.16
9/
15/
1999
87.50
83.32
­
4.66
­
5.39
7.49
8.14
9/
16/
1999
85.50
81.98
­
4.07
­
4.30
6.11
6.30
9/
17/
1999
79.00
82.97
5.14
4.58
10.61
10.36
9/
18/
1999
94.50
96.88
2.18
1.97
6.10
6.04
9/
19/
1999
94.25
105.96
12.40
10.96
15.16
13.79
9/
20/
1999
102.50
107.83
5.27
5.11
5.27
5.11
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
66.50
59.75
­
10.02
­
10.64
10.02
10.64
9/
14/
1999
68.25
64.62
­
5.42
­
5.77
5.43
5.78
9/
15/
1999
87.50
82.02
­
6.07
­
6.76
6.08
6.77
9/
16/
1999
85.50
81.15
­
5.07
­
5.28
5.11
5.32
9/
17/
1999
79.00
76.85
­
2.61
­
2.77
2.89
3.05
9/
18/
1999
94.50
93.50
­
1.23
­
1.32
2.68
2.76
9/
19/
1999
94.25
93.05
­
1.25
­
1.29
1.65
1.69
9/
20/
1999
102.50
102.38
­
0.12
­
0.12
0.13
0.13
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
66.50
56.66
­
14.65
­
15.95
14.65
15.95
9/
14/
1999
68.25
59.39
­
12.98
­
14.11
12.98
14.11
9/
15/
1999
87.50
73.09
­
16.33
­
18.09
16.33
18.09
9/
16/
1999
85.50
75.65
­
11.37
­
12.22
11.37
12.22
9/
17/
1999
79.00
74.39
­
5.65
­
6.12
6.12
6.58
9/
18/
1999
94.50
85.86
­
9.60
­
10.55
11.45
12.38
9/
19/
1999
94.25
92.84
­
1.57
­
2.50
11.68
11.69
9/
20/
1999
102.50
98.10
­
4.20
­
4.46
5.67
5.91
E­
57
Table
E­
45.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
62,
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
57.00
60.41
5.98
5.81
5.98
5.81
9/
14/
1999
N/
A
N/
A
N/
A
N/
A
N/
A
N/
A
9/
15/
1999
78.00
81.86
4.95
4.83
4.95
4.83
9/
16/
1999
70.00
76.98
9.97
9.50
9.97
9.50
9/
17/
1999
89.00
91.89
3.25
3.20
3.25
3.20
9/
18/
1999
83.00
76.84
­
7.42
­
7.71
7.42
7.71
9/
19/
1999
86.00
89.41
3.97
3.89
3.97
3.89
9/
20/
1999
78.00
82.58
5.87
5.70
5.87
5.70
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
57.00
56.97
­
0.05
­
0.05
0.05
0.05
9/
14/
1999
N/
A
N/
A
N/
A
N/
A
N/
A
N/
A
9/
15/
1999
78.00
78.06
0.08
0.08
0.08
0.08
9/
16/
1999
70.00
70.07
0.10
0.10
0.10
0.10
9/
17/
1999
89.00
88.86
­
0.16
­
0.16
0.16
0.16
9/
18/
1999
83.00
76.84
­
7.42
­
7.71
7.42
7.71
9/
19/
1999
86.00
86.12
0.14
0.14
0.14
0.14
9/
20/
1999
78.00
78.12
0.15
0.15
0.15
0.15
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
57.00
55.72
­
2.25
­
2.27
2.25
2.27
9/
14/
1999
N/
A
N/
A
N/
A
N/
A
N/
A
N/
A
9/
15/
1999
78.00
70.27
­
9.91
­
10.43
9.91
10.43
9/
16/
1999
70.00
73.52
5.03
4.91
5.03
4.91
9/
17/
1999
89.00
75.97
­
14.64
­
15.80
14.64
15.80
9/
18/
1999
83.00
72.92
­
12.14
­
12.93
12.14
12.93
9/
19/
1999
86.00
83.16
­
3.30
­
3.36
3.30
3.36
9/
20/
1999
78.00
78.12
0.15
0.15
0.15
0.15
E­
58
Table
E­
46.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
CAMS
601,
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
64.07
55.73
­
13.02
­
13.92
13.02
13.92
9/
14/
1999
56.23
58.65
4.30
4.21
4.30
4.21
9/
15/
1999
72.90
74.12
1.67
1.66
1.67
1.66
9/
16/
1999
70.94
76.39
7.68
7.40
7.68
7.40
9/
17/
1999
103.29
94.71
­
8.31
­
8.67
8.31
8.67
9/
18/
1999
87.12
88.14
1.17
1.16
1.17
1.16
9/
19/
1999
85.65
87.36
2.00
1.98
2.00
1.98
9/
20/
1999
87.12
97.28
11.66
11.02
11.66
11.02
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
64.07
55.73
­
13.02
­
13.92
13.02
13.92
9/
14/
1999
56.23
56.42
0.34
0.34
0.34
0.34
9/
15/
1999
72.90
72.26
­
0.88
­
0.88
0.88
0.88
9/
16/
1999
70.94
71.30
0.51
0.51
0.51
0.51
9/
17/
1999
103.29
94.71
­
8.31
­
8.67
8.31
8.67
9/
18/
1999
87.12
86.16
­
1.10
­
1.11
1.10
1.11
9/
19/
1999
85.65
86.12
0.55
0.55
0.55
0.55
9/
20/
1999
87.12
87.05
­
0.08
­
0.08
0.08
0.08
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
64.07
52.79
­
17.61
­
19.31
17.61
19.31
9/
14/
1999
56.23
54.16
­
3.68
­
3.75
3.68
3.75
9/
15/
1999
72.90
67.46
­
7.46
­
7.75
7.46
7.75
9/
16/
1999
70.94
72.28
1.89
1.87
1.89
1.87
9/
17/
1999
103.29
88.34
­
14.47
­
15.60
14.47
15.60
9/
18/
1999
87.12
78.53
­
9.86
­
10.37
9.86
10.37
9/
19/
1999
85.65
80.53
­
5.98
­
6.16
5.98
6.16
9/
20/
1999
87.12
90.63
4.03
3.95
4.03
3.95
E­
59
Table
E­
47.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
Central
Texas
Downwind
Monitors
(
CAMS
3,
23,
38,
58),
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
59.25
56.69
­
3.73
­
4.09
6.75
7.03
9/
14/
1999
66.00
69.88
6.85
6.11
8.52
7.79
9/
15/
1999
85.25
90.02
5.83
5.52
6.29
5.98
9/
16/
1999
87.25
84.30
­
2.24
­
2.81
8.89
9.22
9/
17/
1999
92.75
96.92
5.38
4.98
8.47
8.17
9/
18/
1999
106.75
104.53
­
1.40
­
1.96
7.69
8.05
9/
19/
1999
100.25
111.55
12.04
10.62
14.48
13.13
9/
20/
1999
97.75
98.83
0.46
0.16
6.55
6.65
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
59.25
55.87
­
5.26
­
5.58
5.26
5.58
9/
14/
1999
66.00
65.37
­
0.92
­
0.93
0.93
0.94
9/
15/
1999
85.25
85.10
­
0.16
­
0.16
0.30
0.30
9/
16/
1999
87.25
81.79
­
5.54
­
5.99
5.59
6.04
9/
17/
1999
92.75
90.97
­
1.63
­
1.68
1.81
1.86
9/
18/
1999
106.75
101.47
­
4.39
­
4.84
4.71
5.16
9/
19/
1999
100.25
98.98
­
1.14
­
1.17
1.50
1.53
9/
20/
1999
97.75
95.40
­
2.87
­
3.07
3.47
3.66
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
59.25
53.79
­
8.74
­
9.36
8.74
9.36
9/
14/
1999
66.00
61.79
­
5.77
­
6.16
7.86
8.21
9/
15/
1999
85.25
78.31
­
7.81
­
8.37
8.85
9.40
9/
16/
1999
87.25
76.84
­
11.19
­
12.17
11.19
12.17
9/
17/
1999
92.75
86.24
­
6.47
­
6.82
6.94
7.29
9/
18/
1999
106.75
94.67
­
10.55
­
11.88
12.41
13.70
9/
19/
1999
100.25
101.12
1.26
0.66
8.85
8.53
9/
20/
1999
97.75
83.18
­
15.38
­
17.46
15.38
17.46
E­
60
Table
E­
48.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
Central
Texas
Upwind
Monitors
(
CAMS
59,
62,
601,
678),
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
65.02
60.33
­
6.66
­
7.23
9.65
10.13
9/
14/
1999
62.41
58.78
­
5.26
­
5.75
8.13
8.56
9/
15/
1999
80.23
76.21
­
4.19
­
4.91
7.50
8.16
9/
16/
1999
76.74
76.58
0.51
0.14
8.32
8.31
9/
17/
1999
86.07
83.84
­
1.96
­
2.30
7.66
7.81
9/
18/
1999
84.53
82.69
­
2.23
­
2.41
5.40
5.52
9/
19/
1999
89.66
92.44
3.21
2.97
5.97
5.80
9/
20/
1999
92.28
97.99
6.44
6.18
6.44
6.18
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
65.02
59.47
­
8.17
­
8.69
8.17
8.69
9/
14/
1999
62.41
58.04
­
6.58
­
7.04
6.80
7.27
9/
15/
1999
80.23
74.80
­
6.05
­
6.74
6.09
6.78
9/
16/
1999
76.74
73.58
­
3.75
­
3.93
4.06
4.23
9/
17/
1999
86.07
81.69
­
4.80
­
5.05
4.90
5.15
9/
18/
1999
84.53
81.34
­
3.81
­
3.97
4.37
4.52
9/
19/
1999
89.66
88.50
­
1.19
­
1.23
1.57
1.60
9/
20/
1999
92.28
92.30
0.03
0.03
0.07
0.07
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
65.02
56.66
­
12.37
­
13.50
12.37
13.50
9/
14/
1999
62.41
54.43
­
12.27
­
13.43
12.27
13.43
9/
15/
1999
80.23
68.70
­
13.72
­
15.15
13.72
15.15
9/
16/
1999
76.74
72.67
­
4.63
­
5.10
8.08
8.48
9/
17/
1999
86.07
76.09
­
10.79
­
11.81
12.62
13.61
9/
18/
1999
84.53
73.87
­
12.64
­
13.67
12.64
13.67
9/
19/
1999
89.66
82.71
­
7.53
­
8.02
7.53
8.02
9/
20/
1999
92.28
91.82
­
0.16
­
0.27
3.72
3.77
E­
61
Table
E­
49.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678),
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
62.13
58.51
­
5.19
­
5.66
8.20
8.58
9/
14/
1999
64.46
65.12
1.66
1.03
8.35
8.12
9/
15/
1999
82.74
83.12
0.82
0.30
6.90
7.07
9/
16/
1999
81.99
80.44
­
0.87
­
1.34
8.60
8.76
9/
17/
1999
89.41
90.38
1.71
1.34
8.06
7.99
9/
18/
1999
95.64
93.61
­
1.81
­
2.19
6.55
6.78
9/
19/
1999
94.96
101.22
6.77
6.01
9.37
8.68
9/
20/
1999
95.02
98.41
3.45
3.17
6.50
6.41
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
62.13
57.67
­
6.72
­
7.14
6.72
7.14
9/
14/
1999
64.46
62.22
­
3.35
­
3.55
3.45
3.65
9/
15/
1999
82.74
79.95
­
3.11
­
3.45
3.19
3.54
9/
16/
1999
81.99
77.68
­
4.64
­
4.96
4.83
5.14
9/
17/
1999
89.41
86.33
­
3.21
­
3.36
3.35
3.50
9/
18/
1999
95.64
91.52
­
3.96
­
4.26
4.68
4.98
9/
19/
1999
94.96
93.74
­
1.16
­
1.20
1.54
1.57
9/
20/
1999
95.02
93.85
­
1.42
­
1.52
1.77
1.86
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
62.13
55.28
­
10.48
­
11.35
10.48
11.35
9/
14/
1999
64.46
58.54
­
8.70
­
9.44
9.90
10.61
9/
15/
1999
82.74
73.36
­
10.95
­
11.97
11.47
12.48
9/
16/
1999
81.99
74.65
­
8.04
­
8.78
9.77
10.47
9/
17/
1999
89.41
80.78
­
9.18
­
9.85
9.41
10.09
9/
18/
1999
95.64
84.15
­
11.73
­
12.92
12.66
13.84
9/
19/
1999
94.96
91.66
­
3.42
­
3.98
8.47
8.57
9/
20/
1999
95.02
87.22
­
8.04
­
9.15
9.81
10.90
E­
62
Table
E­
50.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
(%)
for
Corpus
Christi
Monitors
­
CAMS
4
&
21,
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
79.00
62.34
­
19.60
­
22.29
19.60
22.29
9/
14/
1999
83.00
69.91
­
15.78
­
17.13
15.78
17.13
9/
15/
1999
88.50
79.92
­
9.58
­
10.32
9.58
10.32
9/
16/
1999
84.00
79.98
­
4.77
­
4.91
4.77
4.91
9/
17/
1999
83.50
79.11
­
5.27
­
5.41
5.27
5.41
9/
18/
1999
92.00
99.94
8.64
8.28
8.64
8.28
9/
19/
1999
93.00
104.72
13.05
12.10
13.05
12.10
9/
20/
1999
84.00
86.88
3.58
3.49
3.58
3.49
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
79.00
62.34
­
19.60
­
22.29
19.60
22.29
9/
14/
1999
83.00
69.91
­
15.78
­
17.13
15.78
17.13
9/
15/
1999
88.50
79.92
­
9.58
­
10.32
9.58
10.32
9/
16/
1999
84.00
79.98
­
4.77
­
4.91
4.77
4.91
9/
17/
1999
83.50
79.11
­
5.27
­
5.41
5.27
5.41
9/
18/
1999
92.00
92.10
0.11
0.11
0.11
0.11
9/
19/
1999
93.00
92.95
­
0.06
­
0.06
0.06
0.06
9/
20/
1999
84.00
84.13
0.15
0.15
0.15
0.15
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
79.00
52.75
­
31.68
­
38.39
31.68
38.39
9/
14/
1999
83.00
62.39
­
24.76
­
28.33
24.76
28.33
9/
15/
1999
88.50
66.43
­
24.94
­
28.49
24.94
28.49
9/
16/
1999
84.00
68.27
­
18.75
­
20.72
18.75
20.72
9/
17/
1999
83.50
71.04
­
14.92
­
16.13
14.92
16.13
9/
18/
1999
92.00
91.27
­
0.84
­
0.95
4.59
4.63
9/
19/
1999
93.00
97.16
5.07
4.65
7.92
7.54
9/
20/
1999
84.00
72.78
­
13.28
­
14.24
13.28
14.24
E­
63
Table
E­
51.
Peak
Predicted/
observed
8­
hour
Average
Concentrations
and
Ozone
Metrics
for
CAMS
87,
September
13
 
20,
1999.
Methodology
1:
Date
MAXOBS1HR
MAXPRD1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
86.00
51.09
­
40.59
­
50.93
40.59
50.93
9/
14/
1999
76.00
67.43
­
11.28
­
11.95
11.28
11.95
9/
15/
1999
92.00
88.02
­
4.33
­
4.42
4.33
4.42
9/
16/
1999
87.00
84.09
­
3.34
­
3.40
3.34
3.40
9/
17/
1999
93.00
90.29
­
2.91
­
2.96
2.91
2.96
9/
18/
1999
94.00
85.76
­
8.77
­
9.17
8.77
9.17
9/
19/
1999
92.00
91.37
­
0.68
­
0.69
0.68
0.69
9/
20/
1999
110.00
91.67
­
16.66
­
18.18
16.66
18.18
Methodology
2:
Date
MAXOBS1HR
NEAR1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
86.00
51.09
­
40.59
­
50.93
40.59
50.93
9/
14/
1999
76.00
67.43
­
11.28
­
11.95
11.28
11.95
9/
15/
1999
92.00
88.02
­
4.33
­
4.42
4.33
4.42
9/
16/
1999
87.00
84.09
­
3.34
­
3.40
3.34
3.40
9/
17/
1999
93.00
90.29
­
2.91
­
2.96
2.91
2.96
9/
18/
1999
94.00
85.76
­
8.77
­
9.17
8.77
9.17
9/
19/
1999
92.00
91.37
­
0.68
­
0.69
0.68
0.69
9/
20/
1999
110.00
91.67
­
16.66
­
18.18
16.66
18.18
Methodology
3:
Date
MAXOBS1HR
CAMXPS1HR
Norm
Bias
Fractional
Bias
Norm
Error
Fractional
Error
9/
13/
1999
86.00
47.76
­
44.47
­
57.18
44.47
57.18
9/
14/
1999
76.00
61.97
­
18.46
­
20.34
18.46
20.34
9/
15/
1999
92.00
76.21
­
17.16
­
18.77
17.16
18.77
9/
16/
1999
87.00
81.47
­
6.36
­
6.56
6.36
6.56
9/
17/
1999
93.00
84.99
­
8.61
­
9.00
8.61
9.00
9/
18/
1999
94.00
76.46
­
18.66
­
20.58
18.66
20.58
9/
19/
1999
92.00
88.35
­
3.97
­
4.05
3.97
4.05
9/
20/
1999
110.00
87.70
­
20.27
­
22.56
20.27
22.56
E­
64
Figure
E­
33.
Normalized
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
1,
September
13­
20,
1999.
Downwind
Monitors:

Upwind
Monitors:

Coastal
Monitors:
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
3
CAMS
23
CAMS
38
CAMS
58
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
59
CAMS
62
CAMS
601
CAMS
678
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
4
CAMS
21
CAMS
87
E­
65
Figure
E­
34.
Normalized
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
2,
September
13­
20,
1999.
Downwind
Monitors:

Upwind
Monitors:

Coastal
Monitors:
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
3
CAMS
23
CAMS
38
CAMS
58
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
59
CAMS
62
CAMS
601
CAMS
678
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
4
CAMS
21
CAMS
87
E­
66
Figure
E­
35.
Normalized
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
3,
September
13­
20,
1999.
Downwind
Monitors:

Upwind
Monitors:

Coastal
Monitors:
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
3
CAMS
23
CAMS
38
CAMS
58
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
59
CAMS
62
CAMS
601
CAMS
678
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
4
CAMS
21
CAMS
87
E­
67
Figure
E­
36.
Fractional
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
1,
September
13­
20,
1999.
Downwind
Monitors:

Upwind
Monitors:

Coastal
Monitors:
­
60
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
59
CAMS
62
CAMS
601
CAMS
678
­
60
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
3
CAMS
23
CAMS
38
CAMS
58
­
60
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
4
CAMS
21
CAMS
87
E­
68
Figure
E­
37.
Fractional
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
2,
September
13­
20,
1999.
Downwind
Monitors:

Upwind
Monitors:

Coastal
Monitors:
­
60
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
3
CAMS
23
CAMS
38
CAMS
58
­
60
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
4
CAMS
21
CAMS
87
­
60
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
59
CAMS
62
CAMS
601
CAMS
678
E­
69
Figure
E­
38.
Fractional
Bias
for
Downwind,
Upwind,
and
Coastal
Monitors
Calculated
using
Method
3,
September
13­
20,
1999.
Downwind
Monitors:

Upwind
Monitors:

Coastal
Monitors:
­
60
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
3
CAMS
23
CAMS
38
CAMS
58
­
60
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
59
CAMS
62
CAMS
601
CAMS
678
­
60
­
50
­
40
­
30
­
20
­
10
0
10
20
30
1
2
3
4
5
6
7
8
CAMS
4
CAMS
21
CAMS
87
E­
70
Graphic
Analyses
EPA
also
recommends
the
use
of
graphics
to
evaluate
model
performance
(
EPA
1999).
These
recommendations
include
the
use
of
time
series
plots,
tile
plots,
scatter
plots
and
Q­
Q
plots.
Results
of
all
four
graphics
analyses
are
provided
in
this
section.

Scatter/
Q­
Q
Plots
Scatter
and
Q­
Q
plots
were
developed
using
the
ENVIRON
software
program
described
in
the
ozone
metrics
section.
Scatter
and
Q­
Q
data
are
combined
into
single
plots
for
each
methodology.
Results
of
scatter
and
Q­
Q
analyses
using
the
three
methodologies
for
data
at
CAMS
23,
CAMS
58,
and
the
group
of
Central
Texas
monitors
are
provided
in
figures
E­
39
through
E­
41.
Eight­
hour
average
observed/
predicted
data
points
are
designated
by
blue
"+"
signs.
Q­
Q
points
are
designated
by
magenta
circles.
Eight­
hour
scatter/
Q­
Q
plots
for
other
monitors
and
groups
of
monitors,
as
well
as
one­
hour
plots
will
be
made
available
to
the
TCEQ
and
EPA
on
compact
disc.

Each
scatter
plot
demonstrates
moderate
to
large
correlation
coefficients,
although
there
are
a
few
observed/
predicted
data
points
that
fall
outside
the
±
20%
indicator
lines.
Quantile
points
on
the
Q­
Q
plots
follow
the
1:
1
reference
line
fairly
well
for
each
monitor
and
monitor
group,
with
no
points
falling
outside
the
±
20%
indicator
lines.
The
results
from
these
analyses
indicate
a
high
degree
of
correlation
between
the
paired
predicted
and
observed
data.
E­
71
Figure
E­
39.
Scatter
/
Q­
Q
Plots
for
CAMS
23
Calculated
using
Three
Methodologies.

Daily
maximum
8­
hour
ozone
near
a
monitor
All
sites
and
all
days.
Subregion
=
AACOG
CAMS
23
Nearest
daily
maximum
8­
hour
ozone.

All
sites
and
all
days.
Subregion
=
AACOG
CAMS
23
Daily
maximum
8­
our
ozone
at
monitor.

All
sites
and
all
days.
Subregion
=
AACOG
CAMS
23
E­
72
Figure
E­
40.
Scatter
/
Q­
Q
Plots
for
CAMS
58
Calculated
using
Three
Methodologies.

Daily
maximum
8­
hour
ozone
near
a
monitor.

All
sites
and
all
days.
Subregion
=
AACOG
CAMS
58
Nearest
daily
maximum
8­
hour
ozone.

All
sites
and
all
days.
Subregion
=
AACOG
CAMS
58
Daily
maximum
8­
hour
ozone
at
monitor.

All
sites
and
all
days.
Subregion
=
AACOG
CAMS
58
E­
73
Figure
E­
41.
Scatter
/
Q­
Q
Plots
for
the
Eight
Central
Texas
Monitors
(
CAMS
3,
23,
38,
58,
59,
62,
601,
678)
Calculated
using
Three
Methodologies.

Daily
maximum
8­
hour
ozone
near
a
monitor.

All
sites
and
all
days.
Subregion
=
AACOG
Central
TX
Nearest
daily
maximum
8­
hour
ozone.

All
sites
and
all
days.
Subregion
=
AACOG
Cent
TX
Daily
maximum
8­
hour
ozone
at
monitor.

All
sites
and
all
days.
Subregion
=
AACOG
Cent
TX
E­
74
Tile
Plots
Tile
plots
provide
an
indication
of
where
the
model
is
or
isn't
performing
correctly.
These
plots
are
visual
representations
of
the
model's
predictions
and
provide
such
information
as
when
and
where
the
model
predicts
urban
plumes.
The
following
tile
plots
(
figures
E­
42
through
E­
47)
represent
the
8­
hour
daily
maximum
concentrations
within
the
modeling
domain
for
each
day
of
the
primary
episode.

As
demonstrated
by
these
plots,
urban
plumes
are
replicated
well,
both
in
terms
of
intensity
and
spatial
allocation.
Peak
ozone
concentrations
are
predicted
downwind
of
city
centers
and
major
point
sources
in
these
tile
plots.
E­
75
Figure
E­
42.
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Wednesday,
September
15,
1999.
Figure
E­
43.
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Thursday,
September
16,
1999.
E­
76
Figure
E­
44.
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Friday,

September
17,
1999.
Figure
E­
45.
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Saturday,
September
18,
1999.
E­
77
Figure
E­
46.
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Sunday,

September
19,
1999.
Figure
E­
47.
Predicted
Daily
Maximum
8­
hour
Ozone
Concentrations
within
the
4­
km
Subdomain
for
Monday,

September
20,
1999.
E­
78
Diagnostic
Evaluations
Diagnostic
tests
are
conducted
throughout
model
development
to
assist
with
identifying
and
troubleshooting
performance
problems.
An
important
method
of
conducting
diagnostic
evaluations
involves
the
use
of
sensitivity
tests.
These
tests
are
conducted
by
perturbing
various
types
of
model
input.
For
example,
sensitivity
tests
were
conducted
on
the
1999
episode
by
altering
the
model's
boundary
conditions.
As
a
result
of
these
tests,
some
boundary
conditions
were
modified
from
default
values
to
measurements
collected
during
studies
of
background
concentrations
(
described
in
appendix
B).

Sensitivity
tests
may
also
be
used
to
evaluate
the
base
and
future
case
runs
to
ensure
the
model
responds
appropriately
to
changes
in
emissions
inputs
and
to
estimate
the
impact
of
control
strategies.
This
section
presents
the
results
of
conducting
sensitivity
runs
on
the
1999
base
case
as
a
means
of
assessing
the
responsiveness
of
the
model
and
providing
an
indication
of
which
type
of
emissions
reductions
may
be
most
useful
for
improving
air
quality
in
the
San
Antonio
region.

Zero­
out
Runs:
Urban
Areas
One
sensitivity
evaluation
involved
removing
the
anthropogenic
VOC
and
NOx
emissions
for
the
8­
county
Houston
area
from
the
photochemical
model.
Similarly,
two
other
sensitivity
runs
were
conducted
by
removing
anthropogenic
precursor
emissions
for
the
5­
county
Austin
area
and
2­
county
Corpus
Christi
area
from
the
model.
All
three
of
these
urban
areas
were
upwind
of
the
SAER
on
at
least
one
September
1999
episode
day.

The
results
of
removing
anthropogenic
emissions
were
averaged
over
the
6­
day
primary
episode
period
(
September
15
 
20th)
for
each
of
the
three
sensitivity
runs.
Figure
E­
48
provides
the
estimated
average
reductions
in
ozone
concentrations
within
the
SAER
after
removing
the
anthropogenic
emissions
for
Austin,
Corpus
Christi,
and
Houston
from
the
model.
As
indicated,
removing
Houston's
precursor
emissions
had
the
greatest
impact
on
estimated
ozone
concentrations
in
the
San
Antonio
region
(
2.72
ppb),
followed
by
Corpus
Christi
(
0.64
ppb),
and
Austin
(
0.27
ppb).

These
outcomes
are
consistent
with
the
meteorological
conditions
that
existed
during
the
episode.
Although
Austin
is
closer
in
proximity
to
San
Antonio,
back
trajectory
information
indicates
that
wind
parcels
traveled
through
Houston
and
Corpus
Christi
more
often
during
the
September
1999
episode
than
through
Austin.
In
addition,
Houston
is
a
much
larger
source
of
precursor
emissions
than
either
Corpus
Christi
or
Austin.
E­
79
Figure
E­
48.
Predicted
Reduction
in
Ozone
Concentrations
(%)
in
the
SAER
after
Removing
Austin,
Corpus
Christi,
and
Houston
Anthropogenic
NOx
and
VOC
Emissions
from
the
Photochemical
Model.
(
Average
reductions
for
September
15
 
20,
1999).

Incremental
Removal
of
VOC
and
NOx
Precursors
Across­
the­
board
sensitivity
runs
were
conducted
by
removing
25%,
50%,
75%,
and
100%
of
the
local
(
4­
county
SAER)
NOx
emissions,
VOC
emissions,
and
combinations
of
the
two,
from
the
CAMx
Run
17b
model.
Figures
E­
49
through
E­
54
provide
the
results
of
the
across­
the­
board
reduction
runs
for
each
day
of
the
primary
episode.
These
graphs
provide
the
model's
ozone
concentration
predictions
at
CAMS
23,
the
controlling
monitor,
as
the
result
of
reducing
local
precursor
emissions,
compared
to
the
monitor's
design
value.

Due
to
time
constraints,
VOC/
NOx
reduction
analyses
were
not
conducted
on
the
final
run,
CAMx
Run
18.
However,
several
precursor
sensitivity
runs
were
conducted
on
a
prior
version
of
the
September
1999
model,
CAMx
Run
17b.
The
primary
difference
between
CAMx
Run
17b
and
CAMx
Run
18,
for
the
1999
base
cases,
4
was
the
use
of
a
refined
MOBILE6
on­
road
EI
in
the
latter
model,
as
described
in
section
3.4
of
the
Executive
Summary.
Rerunning
all
the
sensitivity
tests
again
on
Run
18
would
have
added
an
enormous
amount
of
work.
Based
upon
experience,
it
was
assumed
that
the
4
The
2007
projection
developed
from
CAMx
Run
18a
incorporates
a
refined
regional
EI
(
described
in
Section
3.6
of
the
Executive
Summary);
however,
the
regional
EIs
for
Runs
17b
and
18a
1999
base
cases
are
identical.
2.72
parts
per
billion
0.64
parts
per
billion
0.27
parts
per
billion
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
3.5%
Austin
Removed
Corpus
Removed
Houston
Removed
Percent
Reduction
in
Ozone
E­
80
general
findings
and
directional
guidance
determined
from
previous
runs
would
remain
stable
with
relatively
small
emissions
adjustments.

As
indicated
by
the
VOC/
NOx
sensitivity
runs,
reductions
of
NOx,
VOC,
or
a
combination
of
NOx
and
VOC
effectively
reduced
ozone
concentrations
at
the
25%
reduction
level
on
most
episode
days.
An
exception
occurs
on
September
17th,
when
the
model
predicts
a
NOx
reduction
disbenefit
as
the
result
of
removing
25%
and
50%
of
the
anthropogenic
SAER
emissions.
Because
of
the
NOx
reduction
disbenefit
predicted
on
the
17th
and
20th,
reducing
VOC
emissions
alone,
not
a
combination
of
VOC
and
NOx
reductions,
was
the
most
effective
method
of
reducing
ozone
concentrations
on
those
episode
days.
Another
point
that
is
evident
from
these
analyses
is
that,
in
general,
VOC
reductions
were
more
effective
than
NOx
reductions
over
the
range
of
controls
required
to
demonstrate
attainment.

Figure
E­
49.
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
(
4­
county
SAER)
NOx
and
VOC
Emissions
from
Simulation
17b,
September
15,
1999.

55
70
85
100
0
25%
50%
75%
100%
Anthropogenic
Emission
Reductions
(
percentage)
Ozone
Concentration
(
ppb)

NOx
Reduction
VOC
Reduction
NOx
&
VOC
Reduction
8
hour
standard
­
85
ppb
E­
81
Figure
E­
50.
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
(
4­
county
SAER)
NOx
and
VOC
Emissions
from
Simulation
17b,
September
16,
1999.

Figure
E­
51.
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
(
4­
county
SAER)
NOx
and
VOC
Emissions
from
Simulation
17b,
September
17,
1999.
55
70
85
100
0
25%
50%
75%
100%
Anthropogenic
Emission
Reductions
(
percentage)
Ozone
Concentration
(
ppb)

NOx
Reduction
VOC
Reduction
NOx
&
VOC
Reduction
8
hour
standard
­
85
ppb
55
70
85
100
0
25%
50%
75%
100%
Anthropogenic
Emission
Reductions
(
percentage)
Ozone
Concentration
(
ppb)

NOx
Reduction
VOC
Reduction
NOx
&
VOC
Reduction
8
hour
standard
­
85
ppb
E­
82
Figure
E­
52.
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
NOx
and
VOC
Emissions
from
Simulation
17b,
September
18,
1999.

Figure
E­
53.
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
NOx
and
VOC
Emissions
from
Simulation
17b,
September
19,
1999.
55
70
85
100
0
25%
50%
75%
100%
Anthropogenic
Emission
Reductions
(
percentage)
Ozone
Concentration
(
ppb)

NOx
Reduction
VOC
Reduction
NOx
&
VOC
Reduction
8
hour
standard
­
85
ppb
55
70
85
100
0
25%
50%
75%
100%
Anthropogenic
Emission
Reductions
(
percentage)
Ozone
Concentration
(
ppb)

NOx
Reduction
VOC
Reduction
NOx
&
VOC
Reduction
8
hour
standard
­
85
ppb
E­
83
Figure
E­
54.
Predicted
Ozone
Concentrations
at
CAMS
23
after
Removing
Local
NOx
and
VOC
Emissions
from
Simulation
17b,
September
20,
1999.

COMPARISONS
BETWEEN
UT
AUSTIN
AND
AACGO
1999
BASE
CASES
Although
the
original
September
1999
model
was
developed
by
ENVIRON
and
refined
through
a
collaboration
between
ENVIRON
and
UT
Austin
(
meteorological
model
and
air
quality
input
refinements),
the
model
was
eventually
provided
to
the
NNA
partners
(
or
their
contractors)
for
further
modifications.
These
modifications
included
refinement
of
the
emissions
inventory
inputs,
development
of
the
future
case,
and
clean
air
strategy
analyses.
Because
the
model
was
modified
by
more
than
one
agency
during
this
process,
there
was
a
concern
that
the
various
agencies'
models
would
become
dissimilar
and
provide
different
predictions
for
the
base
case,
future
case,
and
control
strategy
runs.

A
great
amount
of
effort
was
spent
ensuring
that
the
Austin
and
San
Antonio
base
and
future
cases
contained
identical
input.
Often
this
involved
discussions
between
the
two
agencies,
as
well
as
TCEQ,
regarding
the
most
appropriate
EI
data
for
local
and
regional
areas.
Discrepancies
in
emissions
inputs
were
corrected
prior
to
the
final
AACOG
run
and
the
final
UT
run.

An
analysis
was
conducted
by
AACOG
staff
to
determine
any
differences
between
the
final
1999
run
refined
by
UT
Austin
and
the
final
run
refined
by
AACOG,
based
on
ozone
predictions
near
two
Austin
monitors.
The
results
of
these
analyses
are
provided
in
table
E­
52.
The
table
lists
peak
predictions
within
the
7x7
array
of
cells
near
the
Murchison
and
Audubon
monitors
for
the
AACOG
(
highlighted
in
yellow)
and
UT
1999
base
cases.

As
shown,
the
differences
between
predictions
by
AACOG's
final
run
(
labeled
1999_
sos.
f)
and
UT's
final
run
(
labeled
1999_
v3)
are
insignificant.
With
regards
to
the
Murchison
monitor,
the
average
difference
(
six
episode
days)
in
ozone
concentrations
between
the
two
1999
base
cases
was
0.00
ppb,
while
the
average
difference
at
the
55
70
85
100
0
25%
50%
75%
100%
Anthropogenic
Emission
Reductions
(
percentage)
Ozone
Concentration
(
ppb)

NOx
Reduction
VOC
Reduction
NOx
&
VOC
Reduction
8
hour
standard
­
85
ppb
E­
84
Audubon
monitor
was
0.05
ppb.
The
2007
future
cases
developed
by
AACOG
and
UT
Austin
were
similarly
compared.
These
comparisons
are
provided
in
appendix
G.

The
performance
of
the
September
1999
photochemical
model
was
thoroughly
analyzed
and
tested
by
AACOG
staff,
both
in
terms
of
1­
hour
and
8­
hour
predictions,
using
a
variety
of
EPA­
recommended
performance
evaluations.
Because
the
model
is
being
used
by
other
Texas
NNA
regions,
performance
analyses
have
been
conducted
by
other
agencies
as
well.
Each
has
concluded
that
the
September
1999
model
meets
EPAacceptance
criteria
for
attainment
demonstration
modeling.
Furthermore,
comparisons
between
the
1999
base
case
refined
by
AACOG
and
the
1999
base
case
refined
by
UT
Austin
provides
additional,
independent
verification
of
the
quality
of
the
model
in
terms
of
performance.

Table
E­
52.
Comparison
of
Predicted
Peak
8­
hour
Concentrations
for
Final
UT
and
AACOG
Base
Case
Runs.

Monitor
UT
1999_
v3
AACOG
1999_
sos.
f
Average
Difference
Days
Date
MURCHISON
84.6
84.6
0.00
6
9/
15
 
9/
20
AUDU
81
80.9
0.05
6
9/
15
 
9/
20
Monitor
UT
1999_
v3
AACOG
1999_
sos.
f
Daily
Difference
Days
Date
MURC
77.8
77.8
0.0
1
9/
15
MURC
75.5
75.4
0.1
1
9/
16
MURC
86.8
86.7
0.1
1
9/
17
MURC
84.5
84.4
0.1
1
9/
18
MURC
89.6
89.7
­
0.1
1
9/
19
MURC
93.6
93.6
0.0
1
9/
20
AUDU
76.2
76.1
0.1
1
9/
15
AUDU
78.2
78.2
0.0
1
9/
16
AUDU
87.4
87.4
0.0
1
9/
17
AUDU
84.5
84.4
0.1
1
9/
18
AUDU
89.4
89.5
­
0.1
1
9/
19
AUDU
70.1
70.2
­
0.1
1
9/
20
E­
85
REFERENCES
Emery,
C.
A.,
Tai,
E.,
Wilson,
G.
M.,
Yarwood,
G.
(
August
6,
2002).
Development
of
a
Joint
CAMx
Photochemial
Modeling
Database
for
the
Four
Southern
Texas
.
Near
Non­
Attainment
Areas.
Novato,
CA:
prepared
for
the
Texas
Near
Non­
Attainment
Areas
through
the
Alamo
Area
Council
of
Governments
by
ENVIRON
International
Corporation.

Emery,
Chris,
Tai,
Edward,
McGaughey,
Gary,
Allen,
David
T.
(
March
31,
2003).
Revised
Meteorological
Modeling
of
the
September
13­
20,
1999
Texas
Ozone
Episode.
Novato,
CA:
ENVIRON
International
Corporation
and
Austin,
TX:
Center
for
Energy
and
Environmental
Resources
at
The
University
of
Texas
at
Austin.

Goldan,
P.
d.,
Kuster,
W.
C.,
Fehsenfeld,
F.
C.
(
1995).
"
Hydrocarbon
Measurements
in
the
Southeastern
United
States:
The
Rural
Oxidants
in
the
Southern
Environment
(
ROSE)
Program
1990."
J.
Geophysical
Research,
100,
D12,
25945­
25963.

Jimenez,
M.,
Wilson,
G.,
Ganesh,
U,
Coulter­
Burke,
S.,
(
February
26,
2002).
Emissions
Processing
for
the
Joint
CAMx
Photochemical
Modeling
of
Four
Southern
Texas
Near
Non­
Attainment
Areas.
Novato,
CA:
prepared
for
the
Texas
Near
Non­
Attainment
Areas
through
the
Alamo
Area
Council
of
Governments
by
ENVIRON
International
Corporation.

Minerals
Management
Service
(
1995).
Gulf
of
Mexico
Air
Quality
Study,
Final
Report,
Volume
1:
Summary
of
Data
Analysis
and
Modeling.
U.
S.
Department
of
the
Interior,
Minerals
management
Service
Gulf
of
Mexico
OCS
Region
(
MMS
95­
0038).

Texas
Commission
on
Environmental
Quality
(
January
8,
2003).
Available
FTP:
//
ftp.
tceq.
state.
tx.
us/
pub/
OEPAA/
TAD/
Modeling/
HGAQSE/
Modeling/
EI/
PointEI_
2000AUG_
20030108.
tar.
gz
The
University
of
Texas
at
Austin
and
ENVIRON
International
Corporation
(
September
17,
2003).
Development
of
the
September
13­
20,
1999
Base
Case
Photochemical
Model
for
Austin's
Early
Action
Compact.
Draft
Report.
Austin,
TX:
prepared
by
the
Capital
Area
Planning
Council
with
contractors
The
University
of
Texas
at
Austin
and
ENVIRON
International
Corporation.

U.
S.
Environmental
Protection
Agency
(
July
1991).
Guideline
for
Regulatory
Application
of
the
Urban
Airshed
Model.
Research
Triangle
Park,
NC:
Office
of
Air
Quality
Planning
and
Standards.

U.
S.
Environmental
Protection
Agency
(
May
1999).
Draft
Guidance
on
the
Use
of
Models
and
Other
Analyses
in
Attainment
Demonstrations
for
the
8­
hour
Ozone
NAAQS.
Research
Triangle
Park,
NC:
Office
of
Air
Quality
Planning
and
Standards.

U.
S.
Environmental
Protection
Agency
(
On­
line,
no
date).
Summary
Descriptions
of
Alternative
Air
Quality
Models.
Available:
http://
www.
epa.
gov/
scram001/
models/
other/
altmodel.
pdf
E­
86
Watkins,
B.
A.,
Parrish,
D.
D.,
Buhr,
S.,
Norton,
R.
B.,
Trainer,
M.,
Yee,
J.
E.,
and
Fehsenfeld
F.
C.
(
1995).
Factors
Influencing
the
Concentration
of
Gas
Phase
Hydrogen
Peroxide
During
the
Summer
at
Kinterbish,
AL.
J.
Geophysical
Research,
100,
No.
311,
22841­
22851.

Wesely,
M.
L.
(
1989).
Parameterization
of
Surface
Resistances
to
Gaseous
Dry
Deposition
in
Regional­
Scale
Numerical
Models.
Atmospheric
Environment,
23,
p.
1293.

Yarwood,
G.
(
September
22,
2003).
Revised
boundary
conditions
for
regional
modeling
of
the
August
1999
ozone
episode.
Draft
Memorandum.
Novato,
CA:
prepared
for
the
Texas
Commission
on
Environmental
Quality
and
the
U.
S.
Environmental
Protection
Agency,
Region
6
by
ENVIRON
International
Corporation.

Yocke,
M.
A.,
Yarwood,
G.,
Emery,
C.
A.,
Heiken,
J.
G.,
Stoeckenius,
T.
E.,
Chinkin,
L.,
Roberts,
P.,
Tremback,
C.,
Hertenstein,
R.
(
1996).
Future­
Year
Boundary
Conditions
for
Urban
Airshed
Modeling
for
the
State
of
Texas.
Prepared
for
the
Texas
Natural
Resources
Conservation
Commission,
Austin,
TX.
