-
A 910 9 83 HO
EPA
United States
Environmental Protection
Agency
Air
Region 10
1200 Sixth Avenue
Seattle WA 98101
November 1983
Carbon Monoxide Study
Anchorage, Alaska
November 22, 1982
to
February 11, 1983
-------
ANCHORAGE
CARBON MONOXIDE STUDY
November 22, 1982 - February 11, 1983
Prepared by
Jon W. Schweiss
U.S. Environmental Protection Agency, Region 10
With The Cooperation and Concurrence
of
George LaMore
Anchorage Air Pollution Control Authority
Thomas Chappie
Alaska Department of Environmental Conservation
US. EPA UBRAflY REGKVt 10 MATERIALS
lllllllllllil
RX0D00D20M2
-------
DISCLAIMER
This report has undergone the U. S. Environmental Protection Agency's (EPA)
peer review process and has been reviewed by both the Anchorage Air Pollution
Control Authority/Municipality of Anchorage (AAPCA/MOA) and the Alaska
Department of Environmental Conservation (ADEC) and is approved for
publication. Approval does not signify that the contents necessarily reflect
the views and policies of the EPA, AAPCA/MOA, or ADEC, nor does mention of
trade names or commercial products constitute endorsement or recommendation
for use.
-------
PREFACE
As prescribed in the Clean Air Act (CAA) of 1970, the U.S. EPA established
National Ambient Air Quality Standards (NAAQS) for protection of the public's
health from carbon monoxide in air external to buildings to which the public
has access. In a number of cities nationwide, including Anchorage, these
standards have not yet been attained. Plans to achieve the standards are
required under the CAA Amendments of 1977. It is hoped that the material
presented herein will assist in achieving progress towards the protection of
the public's health through the attainment of these standards.
i11
-------
ACKNOWLEDGEMENTS
I gratefully acknowledge the invaluable assistance of members of the
Anchorage Air Pollution Control Authority/Municipality of Anchorage
(AAPCA/MOA), Alaska Department of Environmental Conservation (ADEC) and Alaska
Department of Transportation (ADOT). Without their continuing cooperation and
unfailing efforts the successful completion of this study would not have been
realized.
The study was accomplished with the following division of labor. The EPA was
primarily responsible for study design and funding, quality assurance
development, some field training, data processing and analyses, and report
preparation. The AAPCA was primarily responsible for budget and contractor
management, sampling initiation and maintenance, data collection and
reduction, and quality assurance functions. These efforts were coordinated by
George LaMore, Director of AAPCA/MOA and supported by a staff of Stephen
Morris, Wes Tindall, Brenda Horn, and Ron King of MOA's Planning Department.
Tom Chappie and Leonard Verrelli of ADEC were primarily responsible for
providing State input to most study functions and coordinating the
implementation of the traffic count program with ADOT and the MOA's Traffic
Engineering Department.
The contributions of the National Weather Service/Anchorage International
Airport, the control tower crew at the Merrill Field airport, and the
meteorological staff at Elmendorf Air Force Base were also greatly appreciated.
Finally, a great debt is owed to the many members of the EPA Regional staff
who provided guidance, encouragement, and assistance to the task at hand.
Special gratitude is due to both Kenneth Carson and Laurie Fiske for their
endeavors at the computer keyboard and Cathy Chavez for her enduring patience
in many hours with the word processor.
1 v
-------
ABSTRACT
Typically, levels of ambient carbon monoxide (CO) vary widely among the four
existing permanent monitoring sites distributed throughout the city of
Anchorage. An ambient air sampling program was designed and implemented to
clarify and define, if possible, the relationship of carbon monoxide (CO)
levels reported from these permanent sites and levels occurring elsewhere in
the city. Integrated bag sampling was conducted on weekdays at approximately
50 sites during the interval spanning November 22, 1982 and February 11,
1983. Samples collected from each site were analyzed by the non-dispersive
infrared (NDIR) method. Comparisons were then made between data arising from
the study sites and the four permanent monitoring sites. A comprehensive
quality assurance program was developed and ordered to the study to ensure the
collection of data that were of known and appropriate accuracy, precision,
representativeness, comparability and completeness.
In largely fulfilling the purpose of the study, the primary conclusions
arising from analysis of the study data were twofold: 1) The permanent
monitoring network does not adequately characterize either the absolute
magnitude of CO levels or the frequency of standards exceedances encountered
at an array of locations elsewhere in the study area, and 2) The basic or
immediate representativeness of each permanent monitoring site has largely
been established.
v
-------
TABLE OF CONTENTS
Page
PREFACE
ACKNOWLEDGEMENTS Jv
ABSTRACT v
TABLE OF CONTENTS vii
LIST OF FIGURES viii
LIST OF TABLES ix
INTRODUCTION 1
STUDY PURPOSE AND OBJECTIVES 1
STUDY DESIGN 2
Siting Methodology 2
Hot-Spot Screening Technique 2
GridDesign Technique 12
Sampling Methodology 13
Selection of Study and Sampling Intervals 13
Selection of Sampling Methods 13
Quality Assurance 14
DATA ANALYSES 14
LIMITATIONS 15
CONCLUSIONS 17
RESULTS AND DISCUSSION 18
Microscale Study Network 18
Study-Long Network Statistics 19
Dally Network Statistics 31
Correlation and Regression Analysis 36
Comparison of CBD and Corridor Sites 41
Relationship of "AM" to "PM" Four-Hour Averages 42
GRID NETWORKS 45
Garden Network 45
Sand Lake Network 48
Spenard and Benson Network 48
INTER-NETWORK COMPARISONS 50
QUALITY ASSURANCE 53
Sampling Quality Assurance 53
Analytical Quality Assurance 54
Data Handling Quality Assurance 55
Data Completeness 55
v11
-------
LIST OF FIGURES
FIGURE PAGE
1 Location of Sampling Networks 3
2 Location of Sampling Sites Central Business District Network . . 4
3 Location of Sampling Sites Corridor Network 6
4 Location of Sampling Sites Garden Network 8
5 Location of Sampling Sites Sand Lake Network 10
6 Characteristics of Weekday Carbon Monoxide Average Concentrate-
tion for an 8-Hour Period (11:00 a.m. to 7:00 p.m.) at Each
Si te 20
7 Characteristics of Weekday Carbon Monoxide Average Concentrate-
tlon for an 8-hour Period (11:00 a.m. to 7:00 p.m.) at Each
Site Central Business District Network 21
8 Characteristies of Weekday Carbon Monoxide Average Concentrate-
tion for an 8-Hour Period (11:00 a.m. to 7:00 p.m.) at Each
Site Corridor Network 22
9 Characteristics of Weekday Carbon Monoxide Composite of Daily
8-Hour (11:00 a.m. to 7:00 p.m.) Ratios of Each Study Site to
the 7th and C Study Site 29
10 Characteristics of Weekday Carbon Monoxide Composite of Daily
8-Hour (11:00 a.m. to 7:00 p.m.) Ratios of Each Study Site to
the Spenard and Benson Study Site 30
11 Characteristics of Weekday Carbon Monoxide Composite Range of
CO concentration for all Sites During the 8-Hour Period (11:00
a.m. to 7:00 p.m.) November and December 1982 32
12 Characteristies of Weekday Carbon Monoxide Composite Range of
CO concentration for all Sites During the 8-Hour Period (11:00
a.m. to 7:00 p.m.) January and February 1982 33
13 Characteristics of Weekday Carbon Monoxide Average Concentrate-
tlons for the "AM" 4-Hour Period (11:00 a.m. to 3:00 p.m.) at
Each Site 43
14 Characteristics of Weekday Carbon Monoxide Average Concentrate-
tions for the "PM" 4-Hour Period (3:00 p.m. to 7:00 p.m.) at
Each Site 44
15 Characteristics of Weekday Carbon Monoxide Garden Network ... 47
16 Characteristies of Weekday Carbon Monoxide Average Concentrate-
tlons for an 8-Hour Period (8:00 a.m. to 4:00 p.m.) at Each
Site Sand Lake Network 49
17 Characteristics of Weekday Carbon Monoxide Spenard and
Benson Network 51
vi 11
-------
LIST OF TABLES
TABLE PAGE
1 Site Identification Central Business District Network 5
2 Site Identification Corridor Network 7
3 Site Identification Garden Network 9
4 Site Identification Sand Lake Network 11
5 Results of Pre-Study Data Analyses 13
6 Distribution of Maximum 8-hour CO Concentrations from the
Study Site 23
7 Comparison of Maximum 8-Hour CO Concentration from Study Sites
and Permanent Sites 24
8 Comparison of Second-Highest 8-Hour CO Concentration From
Study Sites and Permanent Sites 26
9 Comparison of Daily Study Network Maximum 8-Hour CO Concentrate-
tion to Permanent Site Values 34
10 Correlation/Regression Results for Group 1 Sites 38
11 95% Confidence Limits About 'r' as a Function of V 40
12 Correlation/Regression Results for Permanent and Study Sites . . 46
13 Correlation/Regression Results for Permanent Sites 52
14 Valid Data Recovery Rates by Site 56
ix
-------
INTRODUCTION
Since the onset of ambient air monitQring in 1974, a carbon monoxide (CO)
problem has been identified with the city of Anchorage, Alaska. Violations of
the standard* established by EPA for the protection of the public's health
from ambient CO levels have been routinely recorded at each of the four
permanent monitoring sites currently operated in Anchorage. It is estimated
that some 90 percent of all the emissions of this colorless, odorless, and
tasteless pollutant in Anchorage are directly attributable to motor vehicle
exhaust. The persistence and severity of this problem have aroused and
garnered the active concern of the general public, automobile industry and
virtually all levels of government, local to federal.
Additional information relative to the magnitude and spatial distribution of
this problem was sought to define the relationships between CO levels measured
at the permanent sites and concentrations elsewhere throughout Anchorage.
Accurate knowledge of this kind is critical in preparing an effective and
comprehensive abatement strategy insofar as the interpretation of the ambient
record bears heavily on the nature, scope and degree of control required.
The three entities with jurisdictional interest in the issue, the Municipality
of Anchorage(MOA), the Alaska Department of Environmental Conservation (ADEC)
and the U. S. Environmental Protection Agency (EPA) conceived and conducted a
sampling study towards resolving the representativeness of the permanent
monitoring network. This report presents the major results and conclusions
from that study.
STUDY PURPOSE AND OBJECTIVES
The express purpose of the study was to examine and establish, if possible,
the representativeness of each site in Anchorage's permanent CO monitoring
network in characterizing the magnitude, spatial, and temporal aspects of the
city's CO problem. The immediate utility of the information arising from the
study would be twofold. It would assist in establishing a credible technical
basis for the derivation of a design value for the city. This is the value to
which the ultimate control strategy would be targeted for reduction of ambient
CO to levels in compliance with EPA's standard. And it would serve in the
selection of the permanent monitoring s1te(s) against which the effectiveness
of the ultimate control strategy would be subsequently Indexed.
Explicit objectives were developed to ensure that this purpose was fulfilled
within the context of Intervening time and resource constraints. Particular
emphasis was given to the representativeness of the "7th and C", "Spenard",
and "Garden" oermanent CO sites 1n recognition of their relative Importance In
completing study alms. Another primary study objective was to provide for the
retrieval of data possessing both high and demonstrable quality and
statistically adequate quantity through application of a comprehensive and
rigorous quality assurance program.
* The National Ambient Air Quality Standard (NAAQS) for CO is "...10
milligrams per cubic meter (9 p.p.m.) - maximum 8-hour concentration not to be
exceeded more than once per year." (40 CFR Part 50)
- 1 -
-------
STUDY DESIGN
Prior to the onset of actual sampling, a monitoring plan was developed to
integrate and implement the various study objectives. The plan was designed
to encompass three largely distinct functional components: siting, sampling
and data analyses. These respective functions represented the three basic
phases through which the study progressed. What follows is a brief
description of each of these phases characterizing the study. It should be
noted that a more exhaustive treatment of the siting and sampling functions
may be found in two support documents: "Anchorage Carbon Monoxide Monitoring
Plan 1982-1983" and the "Quality Assurance Plan for 1982-1983 Anchorage CO
Study".
SITING METHODOLOGY
This particular study was unique in that it incorporated intensive and
simultaneous sampling from the three spatial scales of representativeness most
often emphasized in comprehensive CO monitoring programs: micro- (up to 100
meters), middle- (100 to 500 meters) and neighborhood- (.5 to 4 kilometers)
spatial scales. Concurrent monitoring in each of these spatial scales
provided a profile of CO Impacts experienced in the urban core, along major
traffic facilities, and residential neighborhoods.
Two terms are used here to discuss the concept of representativeness:
1) "Homogenous representativeness" is used in reference to the airmass
over which the concentration of a pollutant is considered uniform.
2) "Analagous representativeness " is used in reference to two or more
non-adjoining areas of homogenous representativeness sharing essentially
identical pollutant concentration characteristics
There were two principal methods employed in designing the bulk of the study
network located 1n Figure 1. The hot-spot approach, applied primarily to the
design of the central business district (CBD) portion of the network and
depicted in Figure 2 and Table 1, focused on the issue of analogous
representativeness, while the grid technique used to configure the "Garden"
(Figure 4 and Table 3) and "Sand Lake" (Figure 5 and Table 4) portions of the
network emphasized homogenous representativeness. The "Corridor" network
portrayed in Figure 3 and Table 2 was designed using both techniques.
Corollary, but largely subordinate study interests also intervened in the
design exercise, and will be Identified and discussed throughout the narrative
that follows.
Hot-Spot Screening Technique
Generally, attempts to model absolute concentrations of CO over areas of small
dimensions (up to 100 meters) and high emission density have met with little
consistent success. Therefore, as in previous Region 10 CO studies, the
screening model found in EPA's Carbon Monoxide Hot-Spot Guidelines (EPA
450/3-78-035) was used to identify sites of potentially high, NAAQS
threatening CO concentrations and to subsequently configure an effective
microscale sampling network for measuring maximum CO concentrations.
- 2 -
-------
FIGURE 1
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
LOCATION OF SAMPLING NETWORKS
CENTRAL BUSINESS DISTRICT NETWORK
<*T
HQHOHlOf
=. GARDEN NETWORK
AVfNUt «-
FmfwttD
n n
CORRIDOR NETWORK
AMCHO'MGf
INTIA*
NO flOAO
C eiTH AVCNUt —
SAND LAKE NETWORK
CAmPBcLL
RINCAIO «OAO
100TH AVC.
OMALtfV
ROAD
ROAD
KIATT
HUf*MAN
- 3 -
-------
FIGURE
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
LOCATION OF SAMPLING SITES
CENTRAL BUSINESS DISTRICT NETWORK
f
N
20
$X
:£
4
17 IV
3 s I
r
35.
mimimi
I
4
PERMANENT SITE
7TH a C-7C
- 4 -
-------
Table 1
Anchorage CO Study
November 22, 1982 - February 11, 1983
Sits Identification
Central Business District Network
Si te
Group*
Adjacent
Cross
Si de
Type of
Spatial
Street
Street
of St
Samp! er
Seal e
1
1
5th Ave
E/0 L
N
Integrated
Micro
2
2
I St
S/0 5th Ave
W
ii
ll
3
2
5th Ave
W/0 G St
S
II
II
4
2
6th Ave
E/0 G St
N
ii
ll
5
1
6th Ave
W/0 E St
S
ll
ll
6
1
E St
N/0 6th Ave
W
M
ii
7
1
6th Ave
W/0 D St
N
II
ll
8
1
6th Ave
W/0 n St
S
II
ll
g**
1
C St
S/0 6th Ave
E
II
II
10**
N/A
C St
S/0 6th Ave
E
II
ll
11 ***
1
C St
S/0 6th Ave
E
II
Mi ddle
12
1
A St
S/0 5th Ave
W
II
Micro
13
1
5th Ave
E/0 C St
S
II
II
14
2
5th Ave
E/0 D St
s
II
ll
15
1
5th Ave
E/0 E St
s
II
ll
16
2
5th Ave
E/0 E St
N
II
il
17
2
4th Ave
W/0 C St
s
II
ll
18
1
3rd Ave
W/0 E St
s
II
ll
19
1
5th Ave
E/0 Gambell
St s
II
ii
20
1
F St
N/0 2nd Ave
E
II
Neighborhood
31
2
5th Ave
E/0 I St
s
II
Micro
34
2
4th Ave
W/0 C St
n
II
II
35
2
L St
N/0 9th St
w
II
ll
Permanent Si te
7th & C
N/A
C St
S/0 6th Ave
E
Conti nuous
Mi ddle
* - Group 1 sites were sampled a nominal 54 days.
Group 2 sites were sampled a nominal 30 days.
** - Collocated
*** - Collocated with permanent site
- 5 -
-------
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
LOCATION OF SAMPLING SITES
CORRIDOR NETWORK
203
: :s f V-iiii":-;f/m, <¦
¦i ,J '
: '
I
ov
rr
|lj204/s-V:
!
2?1b *201 26
202
. •: .
.1.
I
r
»
4
}
1
23
PERMANENT SITE
SPENARD & BENSON-SB
f
N
21
.si;®....
¦¦
22
~n
o
t—
m
-------
Table 2
Anchorage CO Study
November 22, 1982 - February 11, 1983
Site Identification
Corridor Network
Si te
Group*
Adjacent
Cross
Side of
Type of
Spatial
Street
Street
Street
Sampler
Seal e
21
1
Seward
N/0 Northern Lights
W
Integrated
Micro
22
2
Tudor
W/0 Lake Otis
N
ll
ll
23
2
Tudor
E/0 C St
N
II
II
24
1
Spenard
W/O Minnesota
S
ll
ll
25**
1
Benson
W/0 Spenard
S
li
ll
26
1
Benson
W/0 C St
S
ll
ll
27
1
Arctic
S/0 Northern Lights
W
ti
(l
201
N/A
Address:
1101 30th Ave
N/A
Sequential
Mid/Neigh
202
N/A
Address:
3002 Spenard Road
N/A
ll
ll
203
N/A
Address:
900 W 25th
N/A
ii
ll
204
N/A
Address:
1411 W 33rd
N/A
ll
ll
38
N/A
Address:
1807 Mckinley
N/A
ll
ll
Permanent Site
Spenard/Benson
Benson
W/0 Spenard
S
Conti nuous
Micro
*
Group 1
sites were sampled a nominal 54
days.
Group 2
sites were
sampled a nominal 30
days.
** - Collocated with permanent site
-------
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
LOCATION OF SAMPLING SITES
GARDEN NETWORK
101»
103
105
102»
104
PERMANENT SITE
GARDEN-G
- 8 -
-------
Table 3
Anchorage CO Study
November 22, 1982 - February 11, 1983
Site Identification
Garden Network
Site Adjacent Cross Side of Type of Spatial
Street Street Street Sampler Seal e
101
E 15th St
W/0 Alder
N/A
Sequential
Nei ghborhood
102
A1 der
S/0 E 20th St
W
II
II
103
Rosemary
S/0 E 15th St
W
ti
11
104
E 20th St
W/0 Nichols
S
II
II
105*
E 16th
E/0 Garden
S
ii
ll
Permanent Site
Garden 16th St
E/0 Garden
Conti nuous
* - Collocated with permanent site
- 9 -
-------
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
LOCATION OF SAMPLING SITES
SAND LAKE NETWORK
28
*
#
SL
#29
30
PERMANENT SITE
SAND LAKE-SL
- 10 -
f-IClkF 5
f
N
-------
Table 4
Anchorage CO Study
November 22, 1982 - February 11, 1983
Site Identification
Sand Lake Network
Site Adjacent Cross Side of Type of Spatial
Street Street Street Sampler Scale
28 W. 64th St W/0 Cranberry St N Integrated Neighborhood
29 Cranberry St N/0 W. 71st St E " "
30 Caravelle Or W/0 Crawford St S " "
Permanent Si te
SI* Raspberry Rd W/0 Cranberry St S Continuous Neighborhood
* - Sand Lake
- 11 -
-------
This screening model is predicated on an extensive array of relatively severe
underlying assumptions (low ambient temperature and wind speed, "ideal"
receptor location, vehicle composition, etc.). When considered collectively,
these assumptions compose a conservative or worst-case scenario for the
inducement of CO exceedance potential at subject intersections. Model output
is simply in terms of whether an intersection exhibits hot-spot potential.
The model does not characterize the nature of this potential with respect to
either absolute magnitude or projected frequency of standard exceedance.
The number of intersections identified by the model as potential CO hot-spots
far exceeded the number of samplers available to the study. Therefore, each
intersection identified to possess potential subsequently underwent a
second-tier evaluation towards ranking the entire candidate pool with respect
to the adjusted strength of potential. Heavily reliant on previous sampling
experience, several factors were subjectively weighted relative to their
aggregate contribution to CO concentrations at each intersection. Finally,
intersections from this ranked listing were considered against the logistical
limitations posed by both the sampling methodology and the spatial
distribution of candidate sites.
Each of these samplers was sited in conformance with EPA's siting criteria for
monitoring maximum concentrations of CO in a micro spatial scale'(40 CFR Part
58, Appendix E).
Grid Design Technique
The grid design technique is a relatively straightforward method of
establishing both the homogenous and analogous aspects of pollutant
concentrations throughout an airmass. This technique is particularly
applicable to the design of a sampling network emphasizing the middle- and
neighborhood- spatial scales of representativeness of existing permanent
monitoring sites.
The technique used here involved designing a grid of samplers at sites both
equidistant from the permanent monitor and each other and in basically
comparable physical environments. The dimension of the circumscribing radius
was arbitrary. The dimension(s) selected for this study coincided as nearly
as possible with the increments EPA uses in defining middle- and neighborhood-
scales of representativeness: 100 meters to .5 kilometers and .5 kilometers
to 4 kilometers respectively (40 CFR 58, Appendix 0). Once the general
sampling location had been determined, other pertinent factors were considered
in the selection of individual prospective sites towards enhancing inter-site
comparability.
Each of these samplers was sited to conform to EPA's siting criteria for
sampling in a neighborhood spatial scale (40 CFR Part 58, Appendix E). Due to
design peculiarities, probe inlet height for the sequential samplers was
approximately two (2) meters lower than the lower limit prescribed by the
siting criteria. However, since these sites were sufficiently removed from
roadways, this slight probe inconsistency is not thought to have affected the
data to any discernible extent.
- 12 -
-------
SAMPLING METHODOLOGY
This function was itself composed of several individual elements including:
selection of study and sampling intervals, selection of sampling methods, and
quality assurance.
Selection of Study and Sampling Intervals
Ambient CO levels are typically cyclic in nature, revolving through several
temporal scales simultaneously, from diurnal to seasonal in duration. In
order to optimize the probability of sampling the phenomenon of interest, i.e.
high CO concentrations and thereby realize the most effective utilization of
resources, a pre-study analysis was performed to determine the seasonal and
daily intervals most frequently characterized by maximum CO potential. This
was accomplished through a historical review of the data reported from each of
the four permanent CO sites in Anchorage during the four previous winter
seasons. The results of the review are summarized in Table 5.
TABLE 5 Results of Pre-Study Data Analyses
Permanent CO Site
Months of Greatest 5 Consecutive Daily 8-hour or
Exceedance Fre- Days of Greatest 16-Hour Interval* of
quency (Decreasing Exceedance Greatest Exceedance
Order)
Frequency
Frequency
7th and C Dec, Jan, Nov, Feb Monday-Friday
Spenard and Benson Dec, Jan, Mov, Feb Monday-Friday
Garden Dec, Jan, Nov, Feb Monday-Friday
Sand Lake Dec, Jan, Nov, Feb Monday-Friday
11:00 a.m. - 7:00 p.m.
11:00 a.m. - 7:00 p.m.
9:00 a.m. - 1:00 a.m.
N/A
* - Duration of subject interval corresponds to type of sampler used in
conjunction with permanent site.
In retrospect, all sampling intervals selected for these sub-networks were
largely validated by subsequent sampling data.
Selection of Sampling Methods
Sampling methods were selected which retrieved the types of information that
most effectively responded to the study objectives. This, wtiile satisfying a
nix of other selection criteria including: direct and indirect resource
consumption per data unit, physical and performance specifications, etc.
Two basic types of samplers were employed to collect ambient CO samples:
single bag samplers and multiple, consecutive sequencing bag samplers,
hereinafter referred to as integrated and sequential samplers respectively.
Both samplers operate on the integrated principal where an ambient sample is
pumped at a constant rate over the time interval of interest. All bag samples
were analyzed via the NDIR (non-dispersive infra-red) method* to yield the
"average" ambient CO concentration over the subject interval.
* - EPA-designated reference method: Beckman Model 866 CO Analyzer
- 13 -
-------
The integrated samplers, deployed extensively in the CBD and Corridor portions
of the study network, were used to collect two consecutive four-hour "average"
samples each study day. The resultant concentrations were then averaged to
construct an eight-hour average concentration of CO for comparison against the
eight-hour NAAQS. Three of these samplers were modified to collect a single
eight-hour bag sample and deployed in the grid about the Sand Lake permanent
moni tor.
The sequential samplers were located throughout the study network, but
primarily in the Spenard and Garden grid networks where a discrete hourly
profile was desired for comparison with the focal permanent site and other
sites in the grid. These samplers collected 16 consecutive samples allowing
for the construction of up to as many as eiqht overlapping 8-hour intervals
each study day for comparison against each other and the standard.
Traffic and meteorological data were also collected over the term of the
study. While these data are of particularly critical significance to future
analyses of the data, they will not be included in this report.
Quality Assurance
A comprehensive and rigorous quality assurance (QA) program was developed
documented, and implemented to ensure that study data were of known and
aopropriate quality, completeness, comparability, and representativeness.
This program provided for routine measures of accuracy and precision for*
sampling, analytical, and data reduction functions.
The quality of all meteorological and traffic count data are largely unknown
due to the lack of direct control of the data generation operation. The
quality is believed to be sufficient for the ultimate intended purpose of the
data.
DATA ANALYSES
Several methods exist by which to analyze and compare data from the study
sites and the permanent monitors. Two basic approaches are presented in this
report to examine the representativeness of the permanent sites. The first
approach compares data on a day-to-day basis. Because the study sites were
sampled for a single eight- or sixteen-hour period each day, data from the
permanent monitors for the identical interval were chosen for purposes of
temporal congruity.
The second approach compares data from the entire study interval, regardless
of whether the compared data occurred on the same day. This method of
analysis lends itself to examining larger patterns and frequencies of CO
levels throughout the term of the study while smoothing the daily inter-site
variability which can occur especially as a result of meteorological impacts.
The results of the analyses presented here reflect the most significant
results and conclusions stemming from a more extensive treatment. Because of
time constraints, this expanded treatment of study data will not be compiled
into a single report document until some later date.
- 14 -
-------
LIMITATIONS
Even wel1-designed studies of this nature are subject to uncertainties of
which both researcher and reader alike should be cognizant. These
qualifications do not necessarily impair the validity of the study results,
but rather frames their present and future application and interpretation
within the context of appropriate caution. The following uncertainties
have been identified with this study:
1) The study spanned only a single CO "season". There is a possibility,
albeit remote, that the variety of conditions influencing CO levels
(traffic, economic, construction, meteorological, etc.) combined to create
a situation grossly anomalous with respect to both previous and succeeding
seasons.
Comment: In a general sense, this situation is not thought to have
occurred here. Cursory inspection of the two factors to which CO levels
are particularly sensitive, traffic volumes and certain meteorological
parameters, indicate basic conformity to conditions characterizing
previous seasons. However, future construction activities and traffic
revisions may impair the long-term utility of site-specific study data by
the degree of their cumulative effect on CO levels.
2) Study sampling data was collected daily for discrete eight- or
sixteen-hour periods. Just as inter-site relationships may exhibit some
degree of daily variability when data are compared for concurrent periods,
these relationships may also vary during periods within a day for which
comparable study data are largely lacking.
Comment: The pre-study analyses conducted to determine optimum sampling
intervals were validated by data actually collected during the study.
While these intervals, particularly the eight-hour, may not have wholly
accounted for any or all of this potential temporal variability, the study
data strongly reflect the intervals most frequently exhibiting the daily
maximum concentrations as measured by the focal permanent sites. The
glaring exception to this is found at the Sand Lake network which enjoyed
a more limited study treatment and where the pre-study analysis was
largely ignored for logistical reasons.
3) As referenced previously, the number of intersections identified as
possessing CO potential in the pre-study siting exercise far outstripped
the number of available samplers. Because of resource and logistical
considerations, many of the intersections rated at higher potential were
passed over for lower rated ones.
Comment: When reviewing the study results, the reader should note that
data for that portion of the network sited to retrieve maximum
concentrations in no way reflects all areas in Anchorage thought to
possess CO potential. Additionally, the reader should be cautioned not to
interpret the proportion of relatively higher impact sites to lower impact
sites in the data displays as necessarily characteristic of the severity
of CO levels occurring throughout the study area.
- 15 -
-------
4) There were some uncertainties associated with the siting of individual
microscale samplers for measuring maximum concentrations. Because CO can
be a highly localized phenomemon, especially when considered over micro
spatial scales, there is a relatively low theoretical probability of
selecting the particular leg and then the particular side of the leg of an
intersection at which the maximum concentration most frequently occurs.
Comment: It is possible there are other locations at or near subject
intersections that experience consistently higher CO levels than those
measured at the study site. Previous experience in evaluating site
specific features enhances the probability of proper selection. However,
this too is often counterbalanced by difficulties in siting opportunity
and/or logistics. On balance then, the data presented herein should not
be interpreted to necessarily represent the maximum CO concentrations
occurring at any particular intersection. Therefore, caution should be
exercised when drawing inter-site and MAAQS comparisons.
5) The study data were generated by ambient air quality sampling methods
which are not approved by EPA for use as the primary basis for either
NAAQS attainment/nonattainment detemunations or the definitive
demonstration of control strategy effectiveness.
Comment: There are uncertainties associated with virtually all methods
employed to monitor ambient pollutants, EPA-approved or not. While the
methods chosen for this study are subject to relatively greater
variability in nrecision and accuracy than the EPA-approved methods
located at the permanent sites in Anchorage, special measures were taken
towards defining and minimizing it. As discussed later, these measures
were really quite successful in yielding a data base of roughly comparable
quality to that generated by the permanent network.
6) Gaps in the data record for each study site can impair inter-site
comparisons in that data from certain sites may not reflect ohenomena of
interest which were measured successfully at other sites.
Comment: This is a real problem which we hoped to minimize by samnling
overa long interval. What is particularly troublesome in relatively
short-term studies of this kind is that the inter-site relationships that
are generally well-described by regression analysis for instance may not
be so well defined for some relatively isolated but nevertheless important
features of interest, such as maximum concentrations. Because of this,
qaps in the data record can be critical. Unfortunately, the occurrence of
at least some gaps are unavoidable (refer to Table 13 for data capture
rates for each site). Every effort was made to minimize the number of
these gaps whil6 preserving the fundamental integrity of the data base.
Additionally, analysis involving direct comparisons between sites were
performed, where possible, using data bases reflecting only concurrently
sampled data.
- 16 -
-------
CONCLUSIONS
The major conclusions relative to the primary study ourpose are as follows:
1) Carbon monoxide levels at a number of sites throughout the study area
exceeded the standard with greater frequency and were of consistently
higher magnitude than the sites in the permanent monitoring network for
the period of study. The general consensus among the study principals
(MOA, ADEC.and EPA) is that this situation is duplicated within a range at
an array of other locations throughout the Municipality.
2) When considered in aggregate, the permanent monitoring network
frequently exhibited sub-exceedance values when one or more study sites
elsewhere in the study area reported standards exceedances.
3) The most severe CO impacts in terms of both magnitude and frequency,
were exhibited by microscale sites on larger traffic facilities or
corri dors.
4) There was typically wide variability in the CO levels between some
locations throughout the study area for corresponding intervals. On a
number of occasions when one or more microscale study sites measured
concentrations exceeding the standard, sub-exceedance values were being
measured at other microscale and neighborhood sites.
5) While only certain combinations of study sites from the microscale
network were well-correlated when considered on a date-paired
(simultaneous) basis, all of these sites, including those collocated with
the permanent sites were extremely well-correlated with each other on a
rank-paired (season-long) basis.
6) The 7th & C permanent site was relatively representative of the lower
level microscale sites and may be at or below levels measured at the
neighborhood sites.
7) While the Spenard & Benson permanent site was often representative of
study sites reporting CO levels in the mid to upper range (but not the
highest range) on a study-long basis, it was not very successful in
characterizing levels at other individual study sites on a daily basis.
This site is also properly designated as a microscale site, although it
may have definite utility in characterizing levels in adjoining
(homogenous representativeness) and nearby but non-adjoining (analogous
representativeness) neighborhoods with an appropriate correction factor.
8) The Garden permanent site was not unduly influenced by a single and/or
immediate CO source, and generally characterized CO levels throughout the
adjoining Garden neighborhood grid (homogenous representativeness). It
may experience CO levels somewhat elevated over other areas in the Garden
grid by virtue of its central location in the emission grid.
9) The Sand Lake permanent site generally characterized CO levels in both
the adjoining (homogenous representativeness) and nearbv but non-adjoining
(analogous representativeness) neighborhoods.
- 17 -
-------
RESULTS AND DISCUSSION
Samples collected during virtually identical periods over a large array of
sites affords a characterization of CO distribution over a relatively wide
area. Data were analyzed for (1) the relative magnitude of CO concentrations
reported at various sites throughout Anchorage and (2) any suggested patterns
of ambient levels. Direct comparisons were made with data arising from the
permanent monitors. Comparisons were also made in the form of ratios and
regression/correlation analyses. Since at most study sites either two 4-hour
or 16 hourly samples were collected daily, some measure of temporal
variability was also subject to comparative evaluation.
Summary statistics are primarily depicted in the form of tables and box-plots
or box-plot/base map combinations which enable a visualization of the spatial
and temporal distribution of values for the statistics of interest. The
box-plots portray the distribution of subject data as follows: maximum value,
9th decile, 3rd quartile, mean, median (2nd quartile), 1st quartile, 1st
decile, minimum value. The treatment of eight-hour average data also includes
the number of instances when the standard was exceeded and, when individual
sites are considered, the second highest value to which the standard is
indexed.
The tables and box-plots do not depict data from all sites at which sampling
was conducted. Sites which were sampled either over a very short term or to
fulfill relatively minor study objectives were not included.
Finally, the various major analysis sections may not contain identical slates
of parameters that underwent analytical consideration. Parameters were chosen
that best reflected the critical emphasis of the study.
MICROSCALE STUDY NETWORK; CBD AND CORRIDOR SITES
This first section discusses the results from that portion of the study
network sited primarily to retrieve maximum CO concentrations in a micro
spatial scale. The microscale network was composed of two rather distinct
sub-networks, one located within the general boundaries of the Anchorage's CBD
and the other adjacent to outlying (the CBD) traffic corridors. While
virtually all of these microscale sites were identical with respect to
physical probe siting character!sties (i.e. distance to: nearest traffic lane
nearest intersection, obstructions, ground, etc.), there are several basic
features that distinguish the CBD network from the Corridor network: 1) The
CBD generally has a higher density of streets with 'significant' traffic
volumes, 2) Several CBD sites were located on streets bordered on one or both
sides by one-story or higher buildings whereas all of the Corridor sites were
located in relatively open, well-ventilated areas, and 3) Streets adjacent to
and nearby the Corridor sites typically carried higher traffic volumes than
those in the CBD.
As a quality assurance measure of inter-method comparability, study sites 11
and 25 were collocated with the 7th & C and Spenard permanent sites
respectively. In an effort to bolster the comparability of data actually
undergoing analyses (with respect to variability, completeness, etc.),
comparisons of data from study sites to the 7th A C and Spenard & Benson
- 18 -
-------
permanent sites were actually referenced to the data record for these
integrated sites, hereinafter referred to as site ll/7th&C and site 25/Spenard
respectively.
It was recognized from the study's outset that in order to ensure that
sampling objectives were effectively and efficiently realized, the study
network as originally configured would be subject to periodic revision as a
function of ongoing data analysis and resource considerations. Study design
prescribed the magnitude and schedule of these network revisions by striking a
balance between the statistical integrity (number of cases) and ultimate
utility of the study data.
As a result, of the 29 individual study sites evaluated in this exercise, 18
(hereinafter referred to as Group 1 sites) were sampled a nominal 50 days.
Group 1 consists of sites 1, 5, 6, 7, 8, 9, 11, 12, 13, 15, 18, 19, 20, 21,
24, 25, 26, and 27. Site 10 was collocated with site 9 and was therefore
excused from the exercise. Site 20 was the "background" CBD site and is
included here for purposes of contrast and comparison.
The remaining 10 sites (hereinafter referred to as Group 2 sites) were sampled
a nominal 30 days. Group 2 consists of sites 2, 3, 4, 14, 16, 17, 22, 23, 34,
31, and 35. Eight sites, 2, 3, 4, 14, 16, 17, 22, and 23 were sampled during
roughly the first half of the study. With the exception of site 17, these
sites were discontinued because they were redundant with other sites in the
study network and were re-sited one or more times in order to fulfill other
short-term study objectives. Site 17 was the target of chronic vandalism and
was subsequently moved directly across the street at mid-study and re-numbered
to 34. Sites 31 and 35 were the only other sites samnled during the second
half of the study for which data are considered in this exercise.
Because of the disparate size and contribution of the these two groups, Group
2 data receives minimal treatment in this narrative.
Study-Long Network Statistics
The statistics that follow primarily reflect the characteristics of CO
measured at each study site.
Maximum Eight-Hour Averages -
Elevated concentrations of CO were measured not only in the immediate vicinity
of the 7th & C and Spenard permanent monitors, but at other sites throughout
Anchorage (see Limitations 3 and 4). Figures 6, 7, and 8 and Tables 6 and 7
exhibit the maximum eight-hour averages reported from each site during the
study.
Group 1 Si tes
(General Discussion)
A. Eight-hour maximums for Group 1 sites ranged from 6.2 ppm (site
20) to 27.4 ppm (site 24).
B. Eight-hour maximums for site 11 /7thAC and site 25/Spenard were
8.1 ppm and 15.1 ppm respectively.
- 19 -
-------
FIGURE f
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
CHARACTERISTICS OF WEEKDAY CARBON
MONOXIDE
AVERAGE CONCENTRATIONS FOR AN 8-HOUR
PERIOD
(11:00 A.M. to 7:00 P.M.) AT EACH SITE
BOX HOT LIGINO
X
^ on A»OVf » 2 PPM
— MAXIMUM
| ¦¦ 2ND HIGH
n
*TH OICILI
rU
— 3NO QUAKTtLI
—
— MtOlAN
X
— MCA*
TT
— 1ST QUAHTIUI
U
l*T OtCIK
MINIMUM
• 1-HOUR CO «AAQ« It 2 MM1
¦
-COLlOCATtO WITH 7TM AMO C
SI -COllOCATIQ WITH SPINAAQ
O— QAROIN
SL-SANO LAXI
40'
GROUP 1 SITES
PERMANENT
SITES
O. 30
M
SITE NUMBER
- 20 -
-------
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
FIGURE 7
CHARACTERISTICS OF WEEKDAY CARBON
MONOXIDE
AVERAGE CONCENTRATIONS FOR AN 8-HOUR
PERIOD
(11:00 A.M. to 7:00 P.M.) AT EACH SITE
CENTRAL BUSINESS DISTRICT NETWORK
UJ
O
z
o
o
o
o
cc
D
0
X
1
CO
40
5
Q.
Ol
Z
~ 30
z
o
H
<
CC
H
20
10
GROUP 1 SITES
•as HOT U0IN9
Jfc
on amvi it f
MAXIMUM
M4M
«TM OICLt
sue auiimu
• MUN
• ifT au*«m«
. ht oiau
..... Moua ea namimmm
• -SOUflUrit WITH TTM AMO C
.¦•.couocAns w»n»
a—4a«»* v*-
¦:v> - ; c&. \
iSSWi ? £ ?£'i S
¦34
• ^
• ^
2'
k .. ¦!>-
16
is Y4
'4;
iitaBBiSwatew:
: V • 1 : : ¦•
':i:i ::
! ;i
r. j-MSvisM v^^s's? fc-JiSiXW:' iSs-'itss: •'<'
':fv :
'3: :
- 21 -
-------
ANCHORAGE CARBON MONOXIDE STUDY z
11/22/82 TO 2/11/83 O
rf 30
CHARACTERISTICS OF WEEKDAY CARBON <
MONOXIDE H-
AVERAGE CONCENTRATIONS FOR AN 8-HOUR Z
PERIOD o_
(11:00 A.M. to 7:00 P.M.) AT EACH SITE 2 g 0.
CORRIDOR NETWORK
OS'
O CL
O Z
o —
DC
0 10f
1
I
00
27«
25^B
- »
26
24
GROUP 1 SITES
j
. F
1
1
!
si 3 3 il
ms not mm
S==
...... MMMMiMtUmi
1 -COUOCATCS WfTM TTW AH* c
as - eOUOCATtS WITH IMUIt
lu— •**« OKI
i r 1 1 1
SITE NUMBER x
21*
t
CO
CM
I
G~5
cr
TO
00
22
-------
(Study/Permanent Site Comparisons)
Eight-hour maximums at 15 of the 17 Group 1 sites were 10% to
238% higher than that for site 11/7th&C (9 of which were over
50% higher).
Eight-hour maximums at five Group 1 sites were greater than or
equal to that for site 25/Spenard (four of which were between
26% and 81% higher).
E. Eight-hour maximums at 11
that at site 25/Spenard.
Group 1 sites were within + 30% of
Table 6 Distribution of Maximum 8-Hour CO Concentration From the Study Sites
Range of Maximum
8-MR CO (In ppm)
0 -
3.0 -
6.0 -
9.0 -
12.0 -
15.0 -
18.0 -
20.9 -
2.9
5.9
8.9
11.9
14.9
17.9
20.9
Percentage
of Group 1
Within Range
0%
0%
16.7%
33.3%
16.7%
11.1%
16.7%
5.6%
Percentage
of Group 2
Within Range
0%
0%
45.5%
27.3%
18.2%
9.1%
0%
0%
Percentage
of Group 1
and Group 2 Within
Range
0%
0%
27.6%
31.0%
17.2%
10.3%
10.3%
3.4%
- 23 -
-------
Table 7 Comparison of Maximum 8-Hour CO Concentrations*
From Study Sites and Permanent Sites
Site No. Date of Maximum 8-HR Ratio of Study Site Ratio of Study
Group Max 8-HR CO Cone. to Site 11** Site to Site 25***
1 Sites CO Cone. (ppm) (8.1 ppm) (15.1 ppm)
1
01/05/82
6.9
0.9
0.5
5
01/05/83
10.7
1.3
0.7
6
02/09/83
11.6
1.4
0.8
7
12/20/82
15.1
1.9
1.0
8
01/05/83
12.6
1.6
0.8
9
01/05/83
9.0
1.1
0.6
11**
12/3/82
8.1
1.0
0.5
12
01/05/83
10.1
1.3
0.7
13
12/20/82
12.8
1.6
0.9
15
12/03/82
11.7
1.4
0.8
18
12/03/82
10.6
1.3
0.7
19
12/20/82
13.2
1.6
0.9
20
01/05/83
6.2
0.8
0.4
21
12/03/82
19.4
2.4
1.3
24
12/03/82
27.4
3.3
1.8
25***
12/20/82
15.1
1.9
1.0
26
12/03/82
19.0
2.4
1.3
27
12/03/82
19.5
2.4
1.3
Group 2
Sites
2
01/04/83
6.0
0.7
0.4
3
01/05/83
6.4
0.8
0.4
4
12/03/82
7.3
0.9
0.5
14
01/05/83
11.5
1.4
0.8
16
12/03/82
10.1
1.3
0.7
17
12/03/82
12.5
1.5
0.8
22
12/30/82
13.1
1.6
0.9
23
12/03/82
16.0
2.0
1.1
31
01/21/83
7.5
0.9
0.5
34
02/03/83
10.5
1.3
0.7
35
01/14/83
8.2
1.0
0.5
* - Measured during the period 11:00 a.m. to 7:00 p.m. on study sampling
days only.
** - Collocated with the 7th & C permanent site
*** - Collocated with the Spenard & Benson permanent site.
- 24 -
-------
Second Highest Eight-Hour Averages -
The eight-hour NAAQS for CO is indexed to the second highest eight-hour
average concentration of CO measured at a given site in a calendar year.
Similar to maximum averages, the levels of second high CO averages were also
elevated throughout the study network. Figures 6, 7, and 8 exhibit the second
highest eight-hour average measured at each site during the study.
Group 1 Sites
(General Discussion)
A. Eight-hour second highs for Group 1 sites range from 4.6 ppm
(site 20) to 22.5 ppm (site 24).
B. Eight-hour second highs for site 11/7thAC and site 25/Spenard
were 7.2 ppm and 14.6 ppm respectively.
(Study/Perrnanent Site Comparison)
C. Eight-hour second highs for 15 of 17 Group 1 sites were between
13% and 213% higher than that for site 11/7thAC (11 of which
were over 50% higher).
D. Eight-hour second highs at 5 Group 1 sites equaled or exceeded
that for site 25/Spenard (3 of which were 34% to 70% higher).
E. Eight-hour second highs at 8 GrouD 1 sites were within + 30% of
that for site 25/Spenard.
- 25 -
-------
Table 8 - Comparison of 2nd Highest 8-Hour CO Concentrations*
From Study Sites and Permanent Sites
Site No.
Group
1 Sites
Date of
2nd Hi 8-HR
CO Cone.
2nd Hi
8-HR CO Cone,
(ppm)
Ratio of Study
to Si te 11**
(7.2 ppm)
1
12/22/82
6.1
0.85
5
02/09/83
8.1
1.13
6
01 /18/83
11.6
1.61
7
12/03/82
13.7
1.90
8
01/18/83
10.9
1.51
9
11/29/82
8.1
1.13
11**
02/01/83
7.2
1.00
12
12/03/82
9.6
1.33
13
12/21/82
11.9
1.65
15
12/13/32
11.4
1.58
18
01/05/83
10.2
1 .42
19
01/05/83
11.8
1.64
20
12/03/82
4.6
0.64
21
12/20/82
17.5
2.43
24
12/17/82
22.5
3.13
2 5* **
02/01/83
13.1
1.82
26
02/01/83
18.6
2.58
27
12/23/82
14.4
2.00
Site Ratio of Study .
Site to Site 25***
(13.1 ppm)
0.47
0.62
0.89
1.05
0.83
0.62
0.49
0.73
0.91
0.87
0.78
0.90
0.35
1.34
1.72
1.00
1.27
1.10
Group 2
2
12/13/82
5.9
0.82
0.45
3
01/04/83
6.0
0.83
0.46
4
12/1 6/82
6.4
0.89
0.49
14
12/03/82
11.1
1.54
0.85
16
01/04/83
9.7
1.35
0.74
17
12/13/82
9.3
1.29
0.71
22
12/09/82
11.7
1.63
0.89
23
01/05/83
9.3
1.29
0.71
31
01/31/83
6.8
0.94
0.52
34
02/01/83
10.4
1.44
0.79
35
01/18/83
5.7
0.79
0.44
* - Measured during the period 11:00 a.m. to 7:00 p.m. on study sampling
days only.
** - Collocated with the 7th & C permanent site
*** - Collocated with the Spenard & Benson permanent site.
- 26 -
-------
Eight-Hour Means and Medians -
Measures of central tendency such as means (arithmetic averages) and medians
(the mid-point value of data ranked by magnitude) were examined to evaluate
aspects of the chronic nature of concentrations reported at each site. As
there is little significant difference between the mean and median for each
site, only the mean will be referenced here. For each site, all eight-hour
averages reported during the study were averaged to produce the mean
eight-hour concentration. Figures 6, 7, and 8 exhibit eight-hour means and
medians for each sampling site.
Group 1 Sites
(General Discussion)
A. Means for the Group 1 sites ranged from 2.0 ppm (site 20) to
14.5 ppm (site 24).
B. Means for sites ll/7th&C and 25/Spenard were 4.1 and 8.1
respectively.
(Study/Permanent Site Comparison)
C. Means for 16 Group 1 sites were 2% to 254% greater than that for
site H/7thSC (10 of which were more than 50% higher).
D. Means for 5 Group 1 sites were 6% to 79% greater than that for
site 25/Spenard (3 of which were 27% to 79% higher).
E. Means for 4 Group 1 sites were greater than the standard while
none of the permanent sites exhibited one.
F. Means for 8 Group 1 sites were within + 30% of that for site
25/Spenard.
Frequency of Eight-Hour NAAQS Exceedances -
There was wide variability in the number of eight-hour NAAQS exceedances
reported from sites in the study network. The statistic chosen for evaluation
here is the simple frequency of exceedance values to all values for each
site. Figures 6, 7, and 8 exhibit exceedance characteristics.
Group 1 Sites
(General Discussion)
A. The frequency of exceedances for the 14 Group 1 sites exhibiting
exceedances ranged from 2% (site 5) to 84$ (site 24).
B. The frequency of exceedances for site ll/7th&C and site
25/Spenard were 0% and 37% respectively.
- 27 -
-------
(Study/Permanent Site Comparison)
C. The frequency of exceedances for 5 Group 1 sites were 30'*, to
111% higher than that for site 25/Spenard (3 of which were 80%
to 228°!0 higher).
D. The frequency of exceedances at 12 Group 1 sites were 70'?, or
lower than that for site 25/Spenard (5 of which measured one or
no exceedances).
Minimum Eight-Hour Averages -
Minimum values are of interest insofar as they assist in characterizing
"background" types of concentrations. Figures 6, 7, and 8 exhibit the minimum
eight-hour averages for each site.
Group 1 Sites
(General Discussion)
A. Minimum eight-hour averages for Group 1 sites ranged from 0.2
ppm (site 20) to 3.7 ppm (site 7).
B. Minimum eight-hour averages for site 11/7th&C and site
25/Spenard were 1.5 ppm and 2.3 ppm respectively.
C. Minimum eight-hour averages for 2 Group 1 sites in the CBD
(sites 6 & 7) were at least 1.0 to 2.0 ppm higher than than any
other Group 1 sites.
Ratios of Study to Permanent Sites
An intuitively appealing way of expressing the relationship between a pair of
sites is computing a simple ratio of the sites' values for the same day. The
relationship between each of the microscale sites and site 11/7thflC and site
25/Spenard are exhibited in Figures 9 and 10 respectively.
Group 1 Si tes
A. All but one Group 1 sites (site 20, the CBD "background" site)
had a mean ratio with site 11/7th&C greater than 1.0.
B. Seven of the 18 Group 1 sites had mean ratios with site
25/Spenard greater than or equal to 1.0.
- 28 -
-------
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
FIGURE 9
CHARACTERISTICS OF WEEKDAY CARBON
MONOXIDE
COMPOSITE OF DAILY 8-HOUR (11:00 A.M. TO
7:00 P.M.) RATIOS OF EACH STUDY
SITE TO THE 7TH & C STUDY SITE
BOX PLOT LBGiND
n
I.
- MAXIMUM
- 2ND Ml OH
-»TM OCCILK
- 3RD QUANTILC
• MCOIAN
• Mf AN
. 1ST QUANmi
,1»T DICILC
.MINIMUM
* -COLLOCATED WITH 7TH AND C
«¦ -COLLOCATIO WITH SPCNAAO
I
1 2 J il iS i5 iJ il
SITE NUMBER
- 29
-------
FIGURE 1C
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
CHARACTERISTICS OF WEEKDAY CARBON
MONOXIDE
COMPOSITE OF DAILY 8-HOUR (11:00 A.M. TO
7:00 P.M.) RATIOS OF EACH STUDY
SITE TO THE SPENARD & BENSON STUDY SITE
•OX PLOT LSQCNO
R-
I
• MAXIMUM
JNO MIGM
*»TM OICILI
• :«0 QUANTIU
MIOIAN
MIAN
1ST QUANTILI
1ST OCClLf
MINIMUM
¦ -COLLOCATED WITH 7TH ANO C
si — COU.OCATIO WITH SPCNAAO
-*-14
r:
T
rrn
I « zl ^
SITE NUMBER
- 30 -
-------
Daily Network Characteristics
Another facet of the data analysis examined the day-to-day character!sties of
the study data. This kind of analysis not only provides a profile of
"simultaneous" impacts throughout the study area, but facilitates examining
patterns of CO concentrations measured at the permanent sites and elsewhere in
the study area. Figures 11 and 12 exhibit the daily composite of 8-hour
concentrations reported from Group 1 study sites.
Daily Maximum Eight-Hour Averages -
The daily maximum value reported from among all sites in the study network
illustrate the variable and chronic aspects of inter-site relationships on a
day-specific basis. Table 9 exhibits the relationship of the daily study
network maximum to correspond!'ng site 11 / 7th&C and site 25/Spenard daily
values.
Group 1 Si tes
(General Discussion)
A. Daily maximum eight-hour averages for Group 1 sites ranged from
6.1 ppm (12/27) to 27.4 ppm (12/3), both measured at site 24.
B. The daily maximum eight-hour average was most frequently
recorded at sites 24 and 26. Site 24 was the site of the daily
maximum on 69% of the sample days (with 5 "ties"*), while site
26 recorded the daily maximum on 24% of the sample days (with 9
"ties").
C. The daily maximum eight-hour average was recorded at only four
other Group 1 sites at an individual frequency of less than 7%
of all sample days.
0. Site ll/7th&C and site 25/Spenard reported the daily maximum on
0% and 2% ("tied" with 3 other sites) respectively over all
sample days.
* - Tied values are those within approximately + 0.5 ppm of each other.
- 31 -
-------
FIGURE
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
CHARACTERISTICS OF WEEKDAY CARBON
MONOXIDE
COMPOSITE RANGE OF CO CONCENTRATIONS
FOR ALL SITES DURING THE 8-HOUR PERIOD
(11:00 A.M. TO 7:00 P.M.)
NOVEMBER & DECEMBER 1982
(GROUP 1 SITES ONLY)
SOX PLOT LIGENO
HO. Of VALCliS AT
Oft AtOVI 1.2 PPM
MAXIMUM
ZNO HIQH
ITH OCCtlf
11*0 QUAftTIlK
1ST QUAftTILI
1ST DCCILI
X - DA* or 10 OR LESS
VALID SAMPLES
s-mouii co «a*os (S.2
Q.
Q.
2
O
y-
<
tr
b-
z
LU
o
z
o
o
o
o
cc
=>
o
z
oo
DECEMBER 1982
NOVEMBER 1982
V
f ::
iaiaJaiasaJasaisJ 1 H W I U i iJ ll ik li i« li iJ ti li ae al tk ei ei a£ 2^ 2! ai s£ 3J
l
OflTE
- 32 -
-------
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
FIGURE 12
CHARACTERISTICS OF WEEKDAY CARBON
MONOXIDE
COMPOSITE RANGE OF CO CONCENTRATIONS
FOR ALL SITES DURING THE 8-HOUR PERIOD
(11:00 A.M. TO 7:00 P.M.)
JANUARY & FEBRUARY 1983
(GROUP 1 SITES ONLY)
BOX PLOT LEQENO
X -
¥:
HO. Of VALUES AT
OH ABOVE • 2 MM
- MAXIMUM
• 2ND HIQH
- »th occae
- 3*0 QUA ATI IS
' MEDIAN
• MEAN
• 1ST QUAftTlU
_ 1ST DECILE
.MINIMUM
X - DAY Of 10 OR LESS
VALID SAMPLES
a*ouA eo naaqs <«.*
401—
JANUARY 1983
FEBRUARY 1983
2
I
t
¥
ii
X
I
4 ?
i
¦t
t
l-l
I
i
-i
¦i
i
i
m
r
m
1 ii i i H i £ li il i£ li ii li iJ li ti ai al ai al 24 al a£ aJ ai a£ 3J 3! i
DATE
- 33 -
-------
I <3 U I O
ja i I y ^orr^ur i son of Stud/ Network Maximum j-Houf CO concentration to Pcrrmun^nf Site Values
Dai 1 y
Site ll/7thdc
Rdtio of
Daily
site 23/Spenard
Kdtio or
5 ite w/
P-iax i mum
Correspond i ng
Max i mum
S i te
Correspond i ng
Odi 1y Max.
Dai 1 y
6-HR CO
6-HR CU
To Si te
6-HR CO
Site To Site
Max(*i
Date
Cone. (PPM)
Cone. (PPM)
1l/7+n 4
C
Cone. (PPM)
25tn/5penc;rd
14
1 1 / i.l/'o'l
12.3
3.1
4.0
6.6
1 . 9
24
1 l/'^3/o/
I 4 . J
-
-
7.6
1 .6
7
1 1/24/82
to.a
-
-
-
-
24
1 l/2o/b2
14.1
2.7
5.2
7.3
1 .9
24( <:6)
1 1/29/62
1 1.3
5.4
2. 1
6.0
1.9
24
12/01/b 2
6.0
2.3
3.5
3.3
2.4
24
12/02/62
16. 1
5.1
3.6
10.5
1.7
24
12/03/62
27.4
6.1
3.4
14.6
1 .9
24
12/06/6^:
14.0
5.3
2.6
6.6
2.2
15
12/06/62
o.o
1 .5
4.5
2.3
3.0
24
12/09/62
19.6
5
3.3
10.7
1.8
24
12/10/62
13.3
3.2
4.2
8.3
1 .o
24
12/13/82
13.2
4.7
2.a
6.0
2.2
24
1 2/' 1 4/d2
1 4.4
-
-
9.7
1 .5
24
12/13/62
lb.4
3.5
4.7
10.4
l.o
24
12/ I6/O2
1 o.5
6.2
2.7
6.2
2.0
24
12/ 1 7/6/
22.3
5.0
4.5
9. y
2.3
24
12/20/6/
21 .0
o.2
3.5
15.1
1 .4
z6(2 7)
12/21/62
14.4
3.9
3.7
12.4
1.2
24
12/22/62
19.9
4.6
4.3
11.8
1 . 7
24
12/23/62
20.5
5.0
4. 1
12.3
1.7
24
12/24/62
16.0
2.4
7.5
10.4
1.7
15
12/27/82
6. 1
1.9
3.2
2.3
2. 7
24
12/30/62
17.3
6.1
2.8
7.3
2.4
7
12/31/82
8.7
1.5
5.8
7.7
1 . 1
26
I/03/O3
7.4
1 .7
4.4
4.9
1 .5
26(21)
1/04/8J
12.2
5.1
2.4
6.2
2.0
24
1/05/6 3
16.6
6.2
3.0
-
-
it,(7)
I/U6/83
10.4
-
-
9.0
1 .2
c:4
I/07/O3
9.7
3.3
2.9
5.7
1 . 7
Z4
1 /1 9)
1/31/63
14.7
5.5
2.7
8.9
I . 7
24
2/0l / 6j>
2Z.0
7.2
3. t
13. 1
1 .7
/4
2/02/63
13.7
3.2
2.6
6.3
2.2
ZO
2/03/63
15.4
5.0
3. 1
8.9
1 .7
24
2/04/6j
12.3
4.6
2.6
5.6
2 . 1
39
2/07/63
14.6
5.0
2.9
u.2
2.4
^4
//Uo/63
13.3
3.1
4.3
3.9
3.4
24
2/o9/o3
12.4
3.7
3.4
3.6
3.3
24
2/10/6^
16.0
3.5
4.8
4.9
3.4
c6(3*)
2/11/63
1 1 .0
J.4
3.2
4.3
2.5
* Tiea + 0.5 ppm
- 34 -
-------
(Study/Permanent Site Comparison)
E. The daily maximum from one (or more) of the Group 1 sites
exceeded the correspond!ng value from site ll/7th&C and site
25/Spenard on all but one sample day (1/13 when site 25/Spenard
"tied" three other study sites at a value of 6.9 ppm).
F. The daily maximums, from one (or more) of the study sites was on
average, 285% higher than the correspond!"ng values at site
11/7th&C, ranging from 110% to 650% higher.
6. The daily maximum from one (or more) of the study sites was on
average 195% higher than corresponding values at site
25/Spenard, ranging from 0% to 240% higher.
Frequency of Daily Eight-Hour Exceedances From the Network -
The number of eight-hour exceedances reported by the study network on a daily
basis provides a measure of the spatial severity of the CO situation during
identical periods.
Group 1 Si tes
(General Discussion)
A. An exceedance of the standard was recorded at one or more Group
1 sites on 85% of all sample days.
B. An exceedance of the standard was reported at four or more Group
1 sites on 57% of all days sampled.
(Study/Permanent Site Comparison)
C. Site ll/7th&C did not report an exceedance on any day when one
or more Group 1 sites did.
D. Site 25/Spenard did not report an exceedance on 56% of the days
when one or more Group 1 sites did and on 36% of the days when
four or more Group 1 sites did.
Distribution of Daily Eight-Hour Averages
Quartiles and deciles were derived to describe the distribution of values
reported from the study network during a sample day.
Group 1 Sites
(Study/Permanent Site Comparison)
A. On a daily basis, site 25/Spenard was at or below the median
(2nd quartile) of Group 1 sites on 28% of the days and at or
below the 3rd quartile on 74% of the days for which
corresponding data are available.
- 35 -
-------
B. On a daily basis, site 11 /7th&C was at or below the 1st quartile
of all Group 1 sites on 87% of the days for which correspond!- ng
data are available.
Range of Daily Eight-Hour Averages
The range of values reported from the study network during a particular day
can provide a valuable index of intra-network variability. In addition, it
can grossly imply that Dortion of the daily maximum which may be attributable
to "background" levels of CO. This information can be important insofar as
the calculation of the design value incorporates a "background" component.
For instance, a high "background" level might establish that less
site-specific control is required to bring levels at an 'offending' site into
compliance. The reverse is also the case for low "background" levels.
Group 1 Sites
A. The daily range of eight-hour averages ranged from 4.0
ppm(11/24) to 22.8 ppm (12/3).
B. The daily range of eight-hour averages averaged 11.9 ppm.
C. The daily range of eight-hour averages exceeded 18.0 ppm on
seven or 13% of the sample days, and 20.0 ppm on one day.
D. On the 22 days when the daily range of eight-hour averages was
less than 10.0 ppm, the daily maximum did not exceed 12.6 pom.
Correlation and Regression Analysis
One way of mathematically expressing the relationship between data from two
sites is by fitting a line that best minimizes the distance of all data points
to that line. One such line is the linear regression line. Straight by
definition, a simple equation describes the origin and rate of change or slope
of this line. So, by knowing the value of what is called the "independent"
variable, one can predict what could be described as the "average" value of
the "dependant" variable.
This "average" value is typically subject to some error due to the fact that
not all actual data points are situated precisely on the regression line.
Therefore, statistics are needed to describe the variability associated with
this prediction. One such statistic is the coefficient of correlation, when
squared, it becomes the coefficient of determination which is the proportion
of variation in the dependant variable explained by variation in the
independent variable. The higher the value of this coefficient, the more
variation is explained and the stronger the relationship is between two data
sets.
A particularly nettlesome problem with the traditional regression line is that
it assigns all the variability due to sampling error to the dependant variable
ignoring similar errors introduced by the independent variable. Since this
- 36 -
-------
would not accurately reflect the realities of errors in study sampling, a
variation of this approach called two-way regression has been devised by
statisticians to distribute this error term symmetrically to both variables.
The regression parameters for Group 1 sites, including both one-way and
two-way regression lines are presented in Table 10 (it should be noted that
one cannot actually "solve" for 1x' in the one-way line). It should be noted
that all references to regression parameters such as slope and intercept
reflect the two-way line.
Typically, the greater the number of cases one can inspect relative to a
particular phenomenon, the greater the confidence one has in 'understanding'
it. This is exhibited in regression/correlation statistics where the
confidence interval about some statistic (correlation coefficient for
instance) shrinks, or confidence increases as more cases are considered. For
reference in conjunction with the regression/coefficient tables, Table 11 has
been prepared to illustrate the effect the number of cases has on confidence
limits about various V values.
Two basic kinds of correlation/regression analyses were performed on the
data. The first type of analysis compared data for a particular interval that
were paired by date to examine relative levels of CO experienced concurrently
at a pair of sites. The second analytical approach used data sets ranked by
magnitude and paired by rank. This second approach recognizes the effects of
especially meteorology on the temporal variability of CO levels between sites,
in particular microscale sites, by not 'requiring' any pair of sites to
'perform' in some characteristic fashion simultaneously (as the first approach
does). This typically lends itself to examining larger (seasonal) natterns
and frequencies of CO levels throughout some study area.
Date-Paired Correlation/Regression
Group 1 Sites
A. All eighteen sites had coefficients of correlation greater than
or equal to 0.80 with at least one other site, ranging from one
site (site 25) to twelve sites (site 12).
B. Sixteen sites had coefficients of correlation greater than or
equal to 0.85 with at least one other site, ranging from one
site (sites 15 and 25) to six sites (sites 12 and 27).
C. Nine sites had coefficients of correlation greater than or equal
to 0.90 with at least one other site, ranging from one site
(sites 1,8,9, and 18) to three sites (sites 12 and 19).
Rank-Paired Correlation/Regression
Each Group 1 site correlated with all other Group 1 sites at 0.93 or better.
- 37 -
-------
TABLE 10 CORRELATION/REGRESSION RESULTS FOR GROUP 1 SITES
1 \ f ¦ . ^ l I (- I P . .S I 1 F O M ~. I A I 1 U'4 Nf . I P M PS 2*"»T PtOHfSSl'N ~ O ,
s
<"> . Mb9
0. 7 3fl
3H.U
6
u , b 7 7
0.4b a
JR .0
r
i', 7 "j 5
u,5 70
33.0
M
il. 7 M H
0,620
3H ,0
o.»1 5
o. b e 4
40, n
1 t
u,7m
0, © 1 2
40.0
» 2
0 . H99
». 80-4
40 .0
! J
0(«26
U.bH 3
42.0
I b
O.605
0. 366
34.0
1 9
o.dO J
0,644
4 J .0
1 O
o.Hib
u.oy7
35.0
2u
0 . 1 9 J
0.629
43,0
2 1
0.6 W
0.6e7
41 .0
2 4
9.795
0.6)2
35.0
?b
0,610
0. 372
41 .0
«? *>
".750
0,56^
4(1 ,0
27
o.m 7
0,66fi
39,0
b
0.794
0,630
39.0
1
0 ,6H6
0.471
3b,0
It
0.932
U.gfiH
J9,o
9
0.HU2
0.643
39.0
1 1
o.792
0,629
3H .1)
t 2
0.8 3#
0. 102
40.0
1 j
o , 7 H J
0.6 I 3
4 3,0
1 b
O.HO?
(l. b 44
3h • o
1 «
o.93o
0.A75
42.li
1 9
o,7b5
tj , 5 H 6
34,0
I "
O.H J2
o. h 9 j
47.0
I \
o. 725
0, b 25
41 .0
I I
o. 3 7H
37.0
7s
0.34Q
0, 1 15
3 S . 0
2*
0.570
0.332
41 ,0
2 »
0 ,6H0
0 . 4b 3
3 H. 0
>
U . J ? «
0.143
3b. 0
M
o,H J*
0. 702
37,0
9
0.79*
0,6 J4
30.0
1 1
0.744
0.5S4
40.0
1 2
O.oil
0.401
41.0
1 )
" .b* 1
0. 3b1
4 1.0
1 S
0.7H2
o.bll
3b,0
If
0,77/
n,hf)4
40.0
i y
o.b»4
0. J4l
34,0
20
o .h i»j
i). 40 3
4 3.0
?»
'i .boh
o. 2bH
41.0
2 \
3*S
0 . 1 3 i
3^.0
~ s
H ,b 5 o
0.423
3 7.0
) 9
0.7 9 /
0.*3b
2H.0
2<>
ft , 4«9
0.249
3^.0
21
U.KJ1
O.o94
'~•,0
24
>.*14
0.H62
3b,o
75
»».7H7
0.620
3b, n
7 f*
U . 75 3
0,5b7
39.0
2 1
0 ,rt M 1
0,77h
3 7,0
9
o,74b
0,bb4
38.0
\ t
0./25
0.S2K
3H.0
1 2
0.7U7
o.soo
*0.0
1 \
0 , f» ® 1
0.403
41,0
t b
il.HPI
0. 7 7^
36.0
1 M
o •M*b
0.749
40,0
1 *
o,t>52
0. »2h
34,0
20
0,*28
O.hRb
42.0
2 t
o. t> 2o
0. JHS
40.0
A 4
o, is *
0.20b
34,0
?N
'>,7hi
U.ObS
JH.O
/ f?
-l. »2°
t). I M
3«,0
7 >
o.»7 3
U . 2 2 i
?«.n
i i
0 . 4 J 7
u«H 4 I
40,0
l 2
'».*?7
4 t ,0
13
' 16
O.bl )
4 2,0
IS
o.S /7
0.33J
35.0
1 n
o.Ml
o. / cm
42.0
1 *
o . H 7 1
0.759
35.0
2"
o.fcSO
o. to;
4 3.0
/ 1
u. 7oO
0.490
40,0
74
o.b-*9
U.341
>7.0
7S
n. ton
0.J67
42.0
o . fth 1
0.43 1
42.0
77
0.54S
0,300
4 0,0
I . 7 1 *
-7.2^ 1
i *
1 . t«4x *
-r>. t? 1 l
2.009*
-1 . Jb^
I >
1 , 1 OM ~
2.40|
2.&H9A
- 1 . ^ 2 «
1 -
1.614* ~
2.2'H
2 ,ho7X
-4.5b5
T =
1 . 7 ~
-0.H61
1 .54SX
-1 .347
V *
1.1bb* ~
0.734
I.?"3X
.349
Y x
0.957K ~
0 . 0 7 }
1 .775X
-I .901
i -
1 .51 it *
-0.H 1 B
2.260X
•2.425
Y S
1 ,6 b H * ~
0 .1 *0
3.430X
-R,0J3
Y s
1 .409)( ~
0.17b
2.'22
Y «
1.036X ~
0,017
l.tliX
-1 .ObO
Y »
1.035X ~
•0,627
1.495X
-l.o«J
Y *
1.045X ~
1 .39b
0.571X
•1.134
Y •
0 . 5 1 9 X ~
-0.H50
1 • 7 2 0 X
o,3/o
* «
1 .OH 1J ~
3 ,
3.7 ft9X
-6.1b 2
Y a
1 ,SS4* ~
3 . 0 4 4 X
-0.4QH
i a
O.bJOX ~
4 . y 0 2
2.463X
•1 , /bd
Y *
1,012X ~
5 .890
2 • 111 X
•3,0/4
X s
l.lbHX ~
1 .09)
\ ,«3bX
-3.421
i 9
0.4HU ~
5.91 4
1 ,J^7X
•3,103
Y •
1 . 09 1 X ~
•1.173
0. M 71
-0 ,{.9 1
Y «
0,674* ~
0.279
0.7IWX
• 0 ,9b3
Y s
O.btfOX »
0.01 2
1,14PX
-2.3H \
Y »
0.691* ~
o./ss
1.502X
-3,boH
Y »
o,7b«X ~
1 .*24
1 .343X
-3.62H
* *
0 , 9M©I ~
-1,110
1.253*
• 3.«'20
Y *
0,9261 ~
•1 . 32*
1 .H 1 X
-S . J»>4
Y »
0.b3H* ~
1.346
o.5lbX
•1 .506
Y *
0.409J «
-0 . / HO
2 . 7i' 1 %
•4.U10
Y a
0 ,H90> ~
1 .O/'ft
7 . 0 7 7 X
-34,4h /
Y «
1 .0 IHX *
7.311
WS . 7 30 1
y s
n.iiPO «
7 ,B Jh
J .5MM
-I
( «
0 ,62H 1 ~
7./71
- 1 « . b H 0
i 2
0.74<; ~
3,7b5)
0 , if 2 4 X
Y *
0 » 4 5 1 X ~
2.147
0.40 1 X
0 , b4 7
Y a
0.347X ~
1 ,7H3
0.527X
•0.6H2
Y *
0.4 25 X ~
0.261
0.763*
-1 ,3H0
Y a
0.6H7X ~
•0.b76
1 , " 1 6 X
-2.20/
Y *
0,957* ~
-0.715
1 .075*
-*.919
V »
0,401 * ~
2 . 3b 7
O.S39X
-3.344
Y *
0,6241 ~
•0.4M9
1.2 75 X
-3, /67
* «
0.937J ~
•1.1<4
0.253a
•0,36 /
Y »
0.2|Bx *
• 0 ,04/
1 , 4b4X
-3.1^5
/ ¦
1 . 1 47 x ~
-0,742
2.579X
* ¦
1 .B16X »
-I .Nil
1 .J22A
•3, '63
t a
0.9H2X ~
-0 .*>h 7
I ,h*9X
- 3. M
Y a
1.126X ~
t .45 3
I .*5*X
•5.615
Y =
1,304* f
•1.779
0.5HOX
I.Mb
V a
0 . 4 9 2 X ~
1 .910
0.5"») *
0.9 JH
i a
0.43SX ~
1 .394
0,697*
1,7**
Y «
0.546* ~
2 , 1 9M
0,917*
1 .417
i a
0 , f. 4 t * ~
3.101
1 ,0?ox
-0,403
i a
6.H96X ~
0, JM
0. /99X
0,123
Y a
0,713* ~
0,669
1 .1131
0 . 0
1 a
0 ,i»2 ' X ~
H.llOh
1 .M40 *
• 4.011
I »
0 , 6 3 9 X ~
4,492
<1 . H67X
-0,330
Y a
o.HOIX ~
— 0 , '1 1 6
\ ,27bX
•o.mo
t a
1 ,011X ~
0.1 n /
j ,66 4 X
-I .341
Y a
1.0J9X ~
I .NM4
1 ,924X
•3.732
1 a
0.H57X ~
I >53
1 . 305a
•1.979
Y a
t . tosi ~
-0.b«9
1 .750X
•I.692
Y a
1,427)1 ~
-<>,069
0.64IX
•1 .2/2
1 a
0.500X »
•0 . b 70
2, 37 4X
• 2.1/79
Y a
1.327X ~
3,3 39
4,H7<»X
•10,\76
Y a
i.Hin* ~
b.llh
3,b 77 X
•9,991
Y a
0.769X ~
4.049
3.05 3*
- 3,Ml 1
Y a
1 . 4 H 6 * ~
«, 994
3. J52X
•«.102
Y a
1.171* ~
2,544
- 38 -
-------
(TABLE 10 CONT.)
1M . Silt hm'. Silt CUHJ-t LATKjN C'i^r.Lt1f\H. Ml.uf PAIhS 2-*AY RM;HFSM!I1 '"'3. l-^At K|> <;t-F I C N w,
11
2
" . 7 H 3
40,0
* 1.472X
Y a
I .211X ~
O.f.H^
11
4
it . UAh
o.bjy
4 i. u
* 1 ,7 ?0 X
0 . i> i -*
Y -
1 . 2S*X ~
2.0u7
11
5
0 .6o9
0,4 1H
37.li
s 1.953X
•2,111
t s
1 .On 5> ~
1 .M I
11
H
O.mH
0.714
42.0
s 1 , 55 7 X
-1.V72
Y s
J J)1K ~
'»,055
11
9
u. v 2 n
0.H47
3*>.U
* 1,9 49*
-0.794
Y *
1.7WX ~
0. 1 Kb
1 1
2 ^
O./Oi
(J.4Q3
4 3.0
* 0.7 0SX
-0.910
Y •
0.54H* ~
—0, /"5 J
11
J
0, 7 Ifl
43.0
* 2,b69X
-0.973
i s
1.592* ~
3.6H
; i
2 1
»). /Ol
0.49^
Jfl. u
* 4 , 5fl 7 X
-4.MB
1 *
2. IV»X ~
4.NH9
11
2S
0.41/
U . 2 0 0
4 1 .41
* 3.it4«vX
-h , 44h
Y s
o.jxu * ~
4.444
11
?b
0 . 744
U.5n t
42.0
« 2,yyix
-0,270
9
0.w2h
3b, 0
¦ 1 , 1b5 X
-1,10 3
V a
1.044X *
-0,2 3^
1 1
2<>
0.6 IS
0.37H
45,0
» 0,37fX
-0.747
Y 3
0,320* ~
-0, J35
1 1
2 I
0 . xn 7
0,ht>2
45,0
s 1.S34X
-0.911
Y s
1.I44X ~
1 ,°44
1 1
2 4
0 . a J1
41 .0
¦ 2.595X
•4.I1S
Y 3
1.90 2* ~
0.H74
! 1
?*¦
«¦» . 6 4 R
l), 120
4V.0
= 1.43FX
-2.35H
X s
O.M22X ~
1 \
7*
II # HO 3
o.M!>
4b.0
« 1 .bCbX
• 0 , 4 i 3
Y a
1 .2)0* ~
2.P2H
1 1
7 >
0 . *50
0.723
44,0
• J.609X
-2.9<«i
Y *
l,2Aox ~
-n.hj7
1 5
«
II. 7b5
0.5R5
3«,0
* 0,#4bA
0. ^OU
Y a
0,b7 3X ~
1 • ? 1 4
1 s
i
o.bt 1
0,2h|
30,0
• 1.307X
•o.bbl
Y a
0.5M7X ~
1.55*
I*
;<>
0.743
0.552
38,0
« 0.34bA
-0.141
Y a
0.320X ~
0.011
1 5
21
<>.551
0. 30#
3H.0
« 1 ,5b5X
0 . M 1 3
Y *
0,710* ~
5 . 9 4 H
1 "5
2 4
o. HO
0. 1 45
32.0
* 4,«b9X
-14.074
Y a
O.H45* ~
9.627
1 S
2 b
O.Q&H
0.00b
36,0
' 4.1871
-!•». 794
Y a
0 , 10 4J1 ~
7,*47
I 5
b
0.4u1
0 . 1 to t
37,0
a 2.?79X
-1 .502
Y a
0,575* ~
9 ,444
1 S
7
0.49 1
UW43
3b.0
a ] . 9M0 X
-2.^92
Y s
U.703X ~
4. 47H
1 *•
0.7-7
0.b03
3b . U
» 1.771*
~
0 , / u h
Y 3
U.93HX ~
t. S**
1 H
".«25
O.bMI
4b,0
¦ o.bovx
~
-0.5«9
Y s
0 , 4t> 7 A ~
-0.3 7b
1 ri
1
K,h7»
0.454
4 3,0
¦ 1 • b 1 2 X
~
1 .OHil
Y s
0 , 3 J b X ~
5.? 35
1 M
*
" .ft t 3
0,40 I
4 1,0
» J.2b2*
~
-I . 909
V a
1.45H* ~
7,20b
I *
? s
¦».2b2
U.UbH
4 2.0
« 2 • b 0 7 X
~
-5 ,10*
Y *
0. 34 T X ~
b.i'>40
\ n
20
0 .ft 12
0. J74
4 5,0
s ?.<»52X
t
I ,2 3b
i 3
U.972X ~
*>,n a
I H
27
o.bW
0.419
43,0
a 2,023*
~
-1 ,4H3
Y a
1.042X ~
3. ?^2
?n
U . h toO
0.4 lb
3H.0
a 0,339V
~
-0.394
Y a
0.307* ~
-0.133
t 9
I
U . to H 7
47,0
a ) , ? 3 2 A
~
I .304
I a
0.9Hb A «
3,093
1 *
? l
•>.*?!
0 . h 7 4
3 0,0
a / , 0 H 7 X
*
-0.292
Y s
1.52bX »
3.9?H
1 V
2*>
0.h71
'>.451
3 7.(i
* 1.224X
~
-O.hH 1
Y 3
0,7b9 x ~
1.5 30
1 9
in
0.H4/
709
34,0
a 1,3«7X
~
1 ,9b 3
Y *
1.1HX ~
3,^bfl
1 »
2 7
ii. Mtiq
O.U55
35,0
a 1.J03X
*
-I .0J«
Y 3
1 ,003 A ~
1 .105
2"
;i
o .h 7fc
0,45tl
44,0
a 3,*b7X
~
2.HHH
Y a
I.H07X ~
*.h9fl
20
1
o.boo
0.3bO
40,0
a b.414*
~
1 ,OHO
i *
2.5J9* ~
9.759
;>•
25
", 3h3
0.1)2
44,0
• b,bb 3X
~
-4,55^
Y a
0.99 IX ~
b.152
2*
0.52 «
0.274
45,0
a 5.5HDX
~
0,9b4
Y a
1 .651 A +
9,629
M
7
n .«0n
0. )to7
43.0
a 5.093X
~
•1 .OHb
i' *
1.9H2X »
4 , H6«
?1
2 4
*,*??
0,675
3M.0
• 1 ,b49X
~
-1,991
j «
1 ,?4I»X ~
2,1 *2
i l
*
u . 7* h
0.587
4 2,0
9 0,974*
~
-t,«304 X ~
•0,14b
/ :>
i f>
•», 7h*
t», h I b
4 J , 0
a 1.177X
~
2.2U9
Y a
0.B93X ~
4,573
,>;
".^&3
y, 7?f»
4^.U
a 1,109*
~
-0,bb2
Y =
0.932X ~
n, 7
;*
? 7
0.HS4!
0.7 2^
44,0
V a 0.971*
~
• 2. 7b 1
1 a
0.R31X ~
-1,127
- 39 -
-------
Table 11 95% Confidence Limits About
as a Function of 'n
\ h
B \
C.IO
G. 10
.61/-.57
0.10
.92Z-.36
0.10
.44/-.27
0.10
.4Q/-.22
C.IO
.37/-.18
0.10
.35/-.16
0.10
.33/-.14
0.10
• 31/-. 12
0.10
.30/-.11
C.IO
.29/-.10
Q.2U
0.2C
.74/- ,*9
0.20
.59/-.27
0.20
.52/-.17
0.20
.4 0/—•12
C .20
.45/-.08
0.2C
.43/-.06
0.20
.42/-.04
0.20
.40/-.02
0.20
.39/-.01
C.20
.30/0.00
C.JO
0.30
.78/-.41
0.30
.66/-.16
0.30
.60/-.07
C.30
.56/-.01
C.30
.53/0.02
0.10
.51/0.05
0.30
.50/0.C7
0.30
.49/0.09
0. 30
.49/0.lg
C.30
.47/0.11
0.40
0 .40
.a?/-. 31
0.40
.72/-.05
0.40
.66/0.0!
C * 40
.63/0.10
C.40
.61/0.14
0. 4C
.59/0.16
0.40
.50/0.10
0.4C
.57/0.20
0 . 40
.56/0.21
C.40
.55/0.22
C.50
0.50
¦ 86/- ¦ 19
0.50
.77/0.07
0.50
.73/0.17
0.50
.70/0.22
C.30
•60/0.26
0. 50
.67/0.28
0.50
.66/0.30
0.50
.65/0.31
0.50
.64/0.33
C.30
.6 3/0.34
0.60
0.60
.99/-.03
0.60
•92/0.21
0.60
.79/0.31
0.6C
.77/0.35
C.60
.75/0.39
0.60
.74/0 .41
0.60
.73/0.42
0.60
.72/0.44
0.60
.72/0.45
C .60
• 7)/0.46
0.70
0.70
.92/0.13
0.70
.07/0.37
0.70
.99/0.45
C.70
•83/0.90
C.70
•82/0.92
0.70
.81/0.54
0.70
.00/0.96
0.70
•80/0.57
0.70
.79/0.50
C.70
•7S/0.58
o.au
o.eo
.95/C.34
o.ao
.92/0.59
0.90
.90/0.62
C.90
.89/0.65
€.90
.80/0.67
0.90
.08/C.69
0.90
.07/0.70
0.90
.87/0.70
0.90
.86/0.71
C.90
.at/0.72
0.90
0.90
.90/0.62
0.90
.St/0.76
0.90
.95/0.80
0.90
.99/0.02
0.90
.94/0.83
0 . 9 0
.94/C .04
0.90
.94/0*84
0.90
.93/0.95
0.90
.91/0.85
C.90
.93/0.09
- 40 -
-------
Comparison of CBD and Corridor Sites
Both the magnitude of CO levels and the frequency of NAAQS exceedances
recorded at most Corridor sites were consistently and markedly higher than
those recorded at most of the CBD sites.
Group 1 Sites
A. Maximum eight-hour averages for CBD sites ranged from 6.2 ppm
(sites 1 and 20) to 15.1 ppm (site 7), while those for the
Corridor sites ranged from 15.1 ppn (site 25/Spenard) to 27.4
(site 24).
B. The second highest eight-hour averages for CBD sites ranged from
4.6 ppm (site 20) to 13.7 ppm (site 7), while those for the
Corridor sites ranged from 13.1 ppm (site 25/Spenard) to 22.6
ppm (site 24).
C. The frequency of NAAQS exceedances for CBD sites ranged from 0%
(sites 1, 9, 11, and 20) to 51% (site 7), while those for
Corridor sites ranged from 37% (site 25) to QA% (site 24).
D. The means and ranges (minimum to maximum) of values at all-
Corridor sites were markedly greater than those for all CBD
sites, except that CBD site 7's mean was greater than two of the
five Corridor sites (sites 25 and 27).
E. Site 1 had a coefficient of correlation greater than or equal to
0.75 with fourteen of the seventeen other Group 1 sites (and
0.80 for nine) in both CBD and Corridor locations.
F. While ten sites in the CBD (sites 1,5,6,8,9,11,12,13,18, and 19)
had coefficients of correlation greater than approximately 0.75
with between five (site 6) to eleven (site 5) other CBD sites,
they did not correlate with any Corridor sites at 0.75 or better.
G. Three CBD sites (sites 7,15, and 20) had coefficients of
correlation greater than approximately 0.75 with between four
(site 7) and ten (site 1) other CBD sites and with between two
(site 11) and four (sites 1,7,12,13, and 19) of the five
Corridor sites.
H. All five corridor sites had coefficients of correlation greater
than 0.77 with each other.
I. Site 25/Spenard did not correlate with any CBD sites at 0.80 or
better. However, when the continuous record is used in lieu of
the integrated data, the Spenard & Benson permanent site
correlated at 0.74, 0.77, and 0.81 with sites 7,12, and 13
respectively.
- 41 -
-------
Relationship of "AM" to "PM" Four-Hour Averages
The analysis of both intra-site and inter-site relationships of "AM" (11:00
A.M. to 3:00 P.M.) and "PM" (3:00 P.M. to 7:00 P.M.) four-hour values can
provide a characterization of the temporal variability of CO levels at
individual sites and over the entire study area. Figures 13 and 14 present
the composite statistics from each site for the "AM" and "PM" periods
respectively. Due to time constraints, this analysis received a relatively
limited treatment in this report.
Group 1 Sites Only
A. For parameters such as range, maximums, 2nd highs, minimums,
means, and medians, each site's "PM" values were consistently
greater than or equal to the "AM" values when all sample days
are considered. The notable exceptions to this were sites 13
and 20 where the maximum and 2nd highest "AM" values exceeded
those "PM" measures, with site 131s "AM" mean also exceeding
its' "PM" mean.
B. The maximum "AM" averages ranged from 6.4 ppm (site 1) to 25.1
(site 24). The maximum "PM" averages ranged from 5.8 ppm (site
20) to 30.1 ppm (site 24).
C. The minimum values for both "AM" and "PM" periods were
essentially equal (within + 1.0 ppm) at 15 of the 18 Group 1
sites.
D. The median of each site's daily "AM"/"PM" ratio ranged from 0.8
(sites 1,5,7,9,11,18, and 20) to 1.2 (site 13).
E. Some 16 of the Group 1 sites exhibited median "AM"/"PM" ratios
less than or equal to 1.0, demonstrating general dominance of
"PM" over "AM" values when considered on a daily basis.
F. The minimum "AM"/"PM" ratios ranged from 0.0* (sites 20) to 0.5
(sites 13,15, and 24), with an average minimum ratio over all
sites of 0.3.
G. The maximum "AM"/"PM" ratios ranged from 1.5 (sites 1 and 18) to
17.8 (site 26) with an average maximum ratio over all Group 1
sites of 3.0.
Groups 1 and 2 Combined
H. Considering all sites (Groups 1 and 2), the daily "PM" maximum
was greater than the "AM" maximum on 59% of all sample days (not
including 10 days that were within +0.5 ppm) by an average of
4.6 ppm, and ranging from 0.6 ppm (T2/1) to 16.3 ppm (2/1)
greater. Conversely, the daily "AM" maximum was greater than
the "PM" maximum on 22% of all sample days, not including those
days within 0.5 ppm, by an average of 3.2 ppm, and ranqinq from
0.7 ppm (1/3) to 10.7 ppm (12/10) greater.
* - Reflects rounding of a ratio value less than 0.05.
- 42 -
-------
FIGURE 13
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
CHARACTERISTICS OF WEEKDAY CARBON
MONOXIDE
AVERAGE CONCENTRATIONS FOR THE "AM"
4-HOUR PERIOD
(11:00 A.M. TO 3:00 P.M.) AT EACH SITE
BOX PLOT ICGENO
A-
MAXIMUM
2NO HIGH
STH OfiCllf
3MO QUAHTU1
MIOIAN
MIAN
1ST QUAMTILi
1ST OCCILC
MINIMUM
•^ou* co naam t» i
* -COUOCAT1D WITH 7TM AND C
ik .COLLOCATED WtTH SPCNARO
Q — <3 ANOCN
SL-SANO CAKC
A h
T 1 -
I
GROUP 1 SITES
~
j... ...
I1
i
:: k
.....
I I
1 £ 2 i i s n n 11 ii j
PERMANENT .
SITES
at ai
^5al aT
X
X
SITE NUMBER
- 43 -
-------
FIGURE 14
ANCHORAGE CARBOIM MONOXIDE STUDY
11/22/82 TO 2/11/83
CHARACTERISTICS OF WEEKDAY CARBON
MONOXIDE
AVERAGE CONCENTRATIONS FOR THE "PM"
4-HOUR PERIOD
(3:00 A.M. TO 7:00 P.M.) AT EACH SITE
40
SOX PLOT L.CQENO
MAXIMUM
2NO MIQH
9TH OECH.C
3*0 QUAATILI
Y:
— MIOIAN
— MIAN
1ST QUAKTttl
1ST OIClLf
, MINIMUM
S»«OlM CO NAAQS (t.2
« -COUOCATED WITH 7TH AND C
is —COLLOCATED WITH SPCNARO
Q— GARDEN
St— SANO LAKE
2
a.
30
z
o
H
<
CC
I-
Z 20
UJ
O
z
o
o
o
o
DC
D io
0
X
1
GROUP 1 SITES
I
I
. PERMANENT
1 SITES
T f
t I 1
I T I I { I
.... 1 l«
II !l
1 i I \ I 1 ,1 ,i .1 .1 .1 J si >i a .>
X
X
SITE NUMBER
- 44 -
-------
I. On 57" of all sample days, the daily composite median of all
sites1(again Groups 1 and 2) "AM"/"PM" ratios was less than 1.0.
J. The average number of daily exceedances reported from the study
network (Groups 1 and 2) was essentially identical regardless of
whether the "AM"/"PM" ratio was greater or less than 1.0.
However, on the day when the highest proportion of sites
exceeded the standard (78% on 12/3), "AM" averages were clearly
greater than "PM" at over 55% of the study sites. Conversely,
on the day when the second highest nroportion of exceedances
were recorded (68% on 1/5), "PM" averages were clearly greater
than "AM" averages at about 63% of the study sites.
GRID NETWORKS
The primary emphasis of the grid monitoring networks was to establish the
representativeness of the Spenard & Benson, Garden, and Sand Lake permanent
sites over middle and neighborhood spatial scales. The results and discussion
that follow are given on a network-specific basis.
Garden Network
The objective for the Garden study network was to examine and establish, if
possible, the homogenous representativeness of the permanent Garden site. A
total of five sequential samplers were dedicated to this study network. The
locations of these samplers are plotted in Figure 4. Note that site 105 was
collocated with the permanent probe for purposes of method comparison and
quality assurance. Inspection of the data reveals that a systematic
difference between the sequential and continuous samDling methods may be
indicated (refer to Quality Assurance section). Therefore, site 105 was
considered the surrogate of the permanent site for the purpose of maintaining
congruity among the data bases considered during the regression portion of the
analysis.
Even though the samplers employed in this network collected hourly data,
four-hour block data were used in these inter-site comparisons to overcome
potential biases due to autocorrelation and to dampen potential temporal
shifts or offsets associated with impacts at one sampler (or samplers) not
simultaneously, but eventually experienced at the other grid sites. If these
temporal variations were profound (on the order of two or more hours) and/or
intermittent, then the determination of homogeneity would be made immensely
more difficult. This is not thought to have occurred here to any appreciable
extent.
Data arising from the Garden network are displayed in Table 12 and Figure 15.
A. The Garden permanent site had a coefficient of correlation
between 0.94 (site 101) and 0.96 (sites 102 and 103) with each
of the five sites in the Garden study network (0.95 with
collocated site 105).
B. The regression lines for each of the Garden site/study site
pairings had slopes of between 0.33 and 0.37 with intercepts at
or below 0.6 ppm.
- 45 -
-------
Table 12 Correlation/Regression Results* for Permanent and Study Sites
AND. SITE Dtp. SITt COnNtLATION COt'F,<EH« NG.OF PAIH5 2-WAY PEGRC5SION EQ.
1-8A* REGRESSION EG.
7 AC
7 AC
7 AC
7AC
7aC
7AC
7*C
7 AC
7 AC
7AC
7 AC
7 AC
SPBE
spat
SPBE
SPBE
SPBE
sphe
SPBE
SPBE
SPBE
SPBE
SPBE
GAPD
CARD
CARD
CARD
GAWD
GARO
GANG
CARD
GARO
GARO
SDLK
SDLK
SOLK
SDLK
SUtK
SDLK
SDLK
SULK
SOU
1019
1019
1019
10 19
1019
1019
1019
1019
1029
1029
1029
1029
1029
1029
1039
1039
10 39
10 39
1039
1049
10 49
1049
1049
1049
1059
1059
10S9
1059
2019
2019
2019
2029
2029
SPHE
GAkO
SDLK
1019
10 29
1039
1049
1059
2019
2029
2039
3019
GARO
SDLK
1019
1029
10 39
1049
1059
2019
2029
2039
3019
SDLK
1019
1029
10 39
1049
1059
4019
2029
2039
3019
1019
1029
10 39
1049
1 0 5>9
2019
20^9
2039
3019
1029
10 J*
1049
1059
2019
2029
2039
3019
1039
1049
1059
2019
2029
3019
1049
1051
2019
2029
3019
1 059
201V
2029
2039
3019
2019
2029
20 39
3019
2029
2039
3019
2039
3019
3019
0.702
0.734
0 . o5 ^
0.b40
Q.t>«9
0.563
0.596
0.627
0.7*5
0,751
0. 788
0.690
0.694
0.7*3
0.596
0.632
0 .o47
0.546
0.590
0.687
O.R60
u.804
0.749
0.68b
0.943
0.958
0.95*
0.945
0.953
0.723
0.754
0.60b
0.470
0.58b
0.725
0.6 70
0. 5o5
0.SS7
0.588
u.701
0.667
0.569
O.W75
0.918
0.920
0.907
0.712
0.710
0.630
0.4S9
0.895
0.924
0.940
0.6S0
0. M9
0.410
0.890
0.963
0,779
0.838
0.50 3
0.887
0.712
0. 729
O.600
0.419
o.etib
0.704
0.619
0.459
0.891
0.832
0 . 602
0.862
0.651
0.828
0.492
0.5 38
0.425
0.410
0.347
0.317
0.355
0.393
0,616
0.564
0.621
0.477
0.4»2
0.613
0.344
0.400
0.419
0.298
0.3 48
0.471
0 . 7 40
0.646
0.561
0.469
0.H90
0.918
0.917
0.893
0.909
0.523
0.569
0.367
0.221
0.343
0.526
0.449
0.320
0.310
0.345
0.492
0.445
0.324
0.766
0.843
0.846
0.823
0.507
o .505
0.396
0.210
0.801
0.853
0.883
0.423
0.561
0.168
0.792
0.927
0.607
0.702
0.253
0.787
0.507
0.531
0. 3oO
0.176
0.471
0.496
0.183
0.21 1
0.791
0.692
0.362
0.743
0.424
0.685
211.0
Y a
2.118X ~
-0.971
Y ¦
1.215x
218.0
Y «
I.752* ~
-I.548
r ¦
1.118*
207 .0
1.224X ~
•I.748
* ¦
0.744X
10 7.0
*
I , 5*^5 * ~
-1.563
Y s
0.858X
51 .0
Y *
I.H72X ~
•2.552
i «
0.866X
44.0
Y *
1.568X ~
•1.558
r a
0.730X
98.0
Y *
1.602X t
-1.539
r «
0.794*
H6.0
Y a
1.509X ~
-1.462
'{ m
0.815X
97.0
Y •
0.956X ~
0.292
y ¦
0.7b8X
90.0
Y S
1.034X ~
•0.197
Y ¦
0.770X
56.0
¥ a
1.101X ~
-0.9U1
Y ¦
0.850X
75.0
Y a
2.05bA ~
0.417
Y a
1. 160X
213.0
V a
0.831X ~
•0.761
r *
0.610X
202.0
Y a
0.595X ~
-1.252
y a
0.518X
107.0
Y »
0.640X ~
0.086
i ¦
0.449X
54.0
Y «
0.680X ~
•0.635
y «
0.494*
46.0
Y *
0.575* ~
0.199
Y •
0.448X
99.0
Y »
0.6361 ~
0.246
Y a
0.424X
86.0
Y «
0.588X ~
0.500
Y a
0.428X
48.0
y »
0.498X ~
0.746
i •
0.417X
91.0
Y a
0.592X ~
0.075
1 a
0.546*
57.0
Y ¦
l> . 7 1 6 X ~
-0.7tob
1 a
0.613X
76.0
Y »
0.797X ~
2.384
Y a
0, 63 1X
209.0
Y «
0.685X ~
•0.540
1 a
0.527*
108.0
Y a
0.831X ~
0.234
Y a
0.79?*
54.0
Y •
0.875X ~
0. 1 33
y a
0.843X
47.0
Y «
0.835X ~
0.580
Y a
0.805*
100.0
Y »
Q.844X ~
0.317
Y *
0.805X
88.0
Y ¦
0.H47X ~
0.356
r »
0.8I4X
100.0
Y »
0.544X ~
1.308
* a
0.460X
93,0
Y a
0.591X ~
0.815
Y a
0.503X
58.0
Y •
0.653* ~
0.174
Y ¦
0.466X
76.0
Y «
1.078X ~
3.089
Y a
0.487X
103.0
1 «
l.bMI t
0.418
Y a
0.75 1 X
48.0
Y a
1.623* ~
•0.135
Y a
1.037*
47.0
Y «
1 .648* ~
0.171
Y a
0.944X
94.0
Y »
2.190X ~
-0.581
y a
0.907X
Y ¦
1.4to5X ~
0.536
y a
0.691X
94,0
Y «
1 . u 1 8 X ~
1.378
Y a
0.594X
93.0
Y •
1 ,035X ~
1.192
Y a
0.719X
bfl .0
Y «
1.089X ~
0. 761
y a
0.706X
71.0
i •
2.526* ~
1.827
Y ¦
1.011*
44.0
Y a
1 .179* ~
•0.456
Y a
1.012*
37.0
Y a
1 .081X ~
0.O63
Y a
0.9861
90.0
\ »
1.023X ~
0.089
Y a
0,940*
86.0
* a
1 .036X ~
0.058
Y a
0.937X
94.0
Y «
0.617X ~
1.198
t a
0.501*
89.0
Y •
0.670X ~
0. Hi
Y a
0,5)3*
58.0
Y ¦
0.H9X ~
-0.195
Y a
0,55 5*
63.0
* a
1.832X ~
0.647
Y «
0.613X
45.0
Y •
1.02RX ~
0.161
1 a
0.917*
46.0
Y a
0.917* ~
0.417
Y a
0.870*
28.0
Y ¦
1 .080X ~
•0.174
1 a
1,010*
40.0
Y ¦
0.433X ~
1.791
Y a
0.365X
35.0
t a
0.677X ~
0.669
Y a
0,558*
52.0
Y »
1.254X ~
2.864
1 a
0,450*
37.0
Y »
0.955X ~
0.068
Y a
0.854*
27.0
Y «
1.040X t
•0.105
Y a
1 .000*
30.0
1 a
0.784* ~
0,600
Y a
0.644*
34.0
Y a
0.767X ~
0.364
Y a
0.6 101
43.0
Y ¦
1.629* ~
1.873
* a
0.648*
67.0
Y «
1.047* ~
-0.U76
Y a
0,924*
83.0
Y «
0.648X ~
1 .064
Y a
0.52 IX
71.0
Y «
0.735* ~
0.3>6
1 a
0.581*
52.0
Y ¦
U.801* ~
•0.169
Y a
0,525*
58,0
Y «
1.669* ~
0.987
Y a
0.524*
/5.0
Y «
0.663* ~
0.931
Y a
0.515X
77.0
Y «
0.651* ~
0.688
V a
0.51 8 X
54.0
Y a
0,787* ~
0.022
Y a
0.533*
42.0
Y «
1.585* ~
0.726
Y a
0.571X
83.0
* •
t.uou ~
-0,245
X a
0,891*
53.0
Y »
1 . X U 1 X ~
•0.907
Y a
0,901X
57.0
Y a
2.392* ~
•1.177
Y a
1,058*
51.0
Y •
1.123* ~
-0.698
Y a
0.953X
47.0
Y ¦
2.031* ~
0.562
Y a
1 •055*
t 3.0
Y •
1.966* ~
-1.036
Y a
1.466*
* -
Reflects Data for Three 4-Hour Blocks per Day:
11:00 a.m.
3:00 p.m.
7:00 p.m.
• 3:00 p.m.
¦ 7:00 p.m.
-11:00 p.m.
- 46 -
-------
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
CHARACTERISTICS OF WEEKDAY CARBON
MONOXIDE
GARDEN NETWORK
ms hot litmi
rr quaivtvi
FIGURE 15
•COUaCAflO WTTW «««««¦
40r
a.
a.
z
o
p
<
cc
y-
Z
UJ
O
Z
o
o
o
o
DC
D
0
1
i
oo
30
20
10
4-HGUP BLOCK DATA
n= APPROX. 68
1 1 r
101 104 105
x
4-HOUR BLOCK DATA
n= APPROX. 25
i i i
101 102 103 104 105
SITE NUMBER x
4-
G
102» 104
- 47 -
-------
C. While the correlation coefficients for the pairings of site 105
and the other Garden network study sites were somewhat lower
than those for the Garden permanent site pairings, ranging from
0.89 (site 104) to .96 (site 103), the slopes were closer to
unity (1.0) ranging from 0.93 (site 102) to 0.97 (site 101),
with intercepts less than 0.6ppm.
Sand Lake Network
The objective for the Sand Lake study network was to examine and establish, if
possible, the homogenous and analogous representativeness of the Sand Lake
permanent site. Three study sites were deployed about the permanent site.
Their locations are plotted on Figure 5. Study site 29 was sited to examine
the homogenous representativeness of the Sand Lake site. Because sites 28 and
30 were located across one or more wel 1-traveled arterials from the Sand Lake
site, they yielded data relative to analogous representativeness.
As a result of the use of 8-hour integrated samplers in this network, daily
eight-hour block data were used in these inter-site comparisons. In addition,
sampling was not conducted during the period of most frequent maximum daily
concentrations for logistical reasons. Therefore, the relationships described
by Figure 16 and the correlation/regression results appearing in Table 12 may
not adequately characterize the circumstances of other periods within a day.
A. The Sand Lake permanent site had a coefficient of correlation of
0.88 v/ith site 28, 0.97 with site 29, and 0.92 with site 30,
with slopes indicating that study sites 29 and 30 were 2% and
19$ higher on average than the Sand Lake site, while site 28 was
6I lower on average.
B. Site 30, across Jewell Lake Road from the other sites,
correlated at 0.79 and 0.73 with sites 28 and 29 respectively,
with slopes indicating that it is 26% lower than site 28 and 4%
higher than site 29 on average.
Spenard & Benson Network
The objective for the Spenard A Benson network was to examine and establish
the homogenous and analogous representativeness of the Spenard permanent
site. A total of four sites were eventually deployed in the Spenard network,
consisting of three sequential and one integrated samplers. The location of'
these sites are plotted on Figure 3. Sites 201 and 203 were located across
one or more major traffic facilities and therefore retrieved data relative to
analogous representativeness. Sites 202 and 204 were located in the
neighborhood contiguous with the Spenard permanent site and examined
homogenous representativeness.
Again, though hourly data were available from three of these study sites,
four- hour block data were utilized in the correlation/regression analysis.
As site 204 retrieved 8-hour block data, comparisons with that site were made
on that basis.
In addition to the Spenard grid sites, site 38, located in the neighborhood SW
of the intersection of Spenard and Minnesota was included for comparison with
the Spenard grid network.
- 48 -
-------
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
FIGURE 16
CHARACTERISTICS OF WEEKDAY CARBON
MONOXIDE
AVERAGE CONCENTRATIONS FOR AN 8-HOUR
PERIOD (8:00 A.M. to4:00 P.M.)
AT EACH SITE
SAND LAKE NETWORK
Q.
0.
H
<
0C
H
Z
LU
O
z
o
o
o
o
cc
=>
0
1
I
oo
40
30
20
10
•oi nor uaiMQ
• sno win
-m» oiou
- HD au**nu
• MCStAII
• DIM
• 1ST aUAKTlU
irr eiau
a— 4aho«n sirt
»L- UNO UK
IT
—r
fi
SITE NUMBER
30
?•> •••. •• y V?: •
... SL
V
:^vi
...J :-;•••
¦ 0 : ^ jj .f^Wv
¦ 'S'v '*
- 49 -
-------
Data arising from this network are displayed in Figure 17 and Table 1?.
A. The Spenard permanent site had coefficients of correlation of
0.69, 0.86, and 0.80 with sites 201, 202, and 203 respectively,
at slopes of 0.50, 0.59, and 0.72,
B. On an 8-hour block basis, the Spenard permanent site correlated
at 0.72 with study site 204.
C. Site 202 correlated at 0.89 and 0.86 with sites 201 and 203
respectively, and with slopes of 1.00 and 1.12.
D. Site 201 correlated with site 203 at 0.83 with a slope of 1.10.
E. On an 8-hour block basis, site 204 correlated with sites 201,
202, and 203 at 0.80 or better with slopes indicating that it is
on average between 15» to 28® higher.
F. On an 8-hour block basis, site 204 correlated at 0.85 with site
38, running 4^ higher on average.
INTER-NETWORK COMPARISONS
Relationships of CO levels measured in areas spatially removed from each other
were examined in order to characterize their distribution throughout the
city. This kind of analysis can be difficult in that impacts may not be
simultaneous over the entire breadth of the study area (city). A more
exhaustive analysis towards accounting for any of these potential temporal
shifts will be attempted when time allows.
Four-hour block data was used throughout the bulk of this analysis.
Correlation/regression results are displayed in Table 12. Instances where it
was necessary to use 8-hour block data (as with the Sand Lake samplers) are
identified.
Because there was no integrated sampler collocated with the Garden permanent
site, Garden data were used for comparisons with the integrated sites.
However, it should be noted when reviewing these comparisons that there may be
a small systematic difference in the sampling methodologies between the
integrated and continuous sites withthe latter perhaps 3% to 5% higher than
the former on average (refer to Quality Assurance section).
Finally, all available data (including weekends) from the permanent sites were
used in comparisons with each other.
Inter-Permanent Site Comparisons
A. Between the permanent sites, correlations ranged from 0.69 to
0.80.
B. As indicated by the slope of the regression line, Spenard Ft
Benson was 98% and 31% higher on average than the 7th & C and
Sand Lake sites respectively, with intercepts of -0.62 and 1.10.
C. Spenard & Benson was 11% higher on average than the Garden site.
- 50 -
-------
ANCHORAGE CARBON MONOXIDE STUDY
11/22/82 TO 2/11/83
CHARACTERISTICS OF WEEKDAY CARBON
MONOXIDE
SPENARD & BENSON NETWORK
8-HOUR BLOCK
DATA
4-HOUR BLOCK
DATA
4-HOUR BLOCK :
30 ^ DATA
n= APPROX. 20
n= APPROX. 85 : n= APPROX. 55
i 1 1 r™—i 1 1 1 t 1
SB 201 202 SB 201 202 203 SB 201 202 203 204 38
SITE NUMBER
If >: L<
•OX HOT lEOCNfl
•• *
"£• vv:
; •• • • - '
NO. OP VALUES AT
on asovi » s rpu
MAXIMUM
2NO HI OH
»TM OICIll
]no ouAirnii
I
,L
MiOlAN
MIAN
1ST QUAIITill
nST OCCIlt
SI - SPENARD fc BENSON
^v-m:vfe38
yii;, ¦ '5V,;;V.;^ ?'¦/> |
•f i
-------
0. Garden was 87% and 73% higher on average than the 7th X C and
Sand Lake sites respectively, with intercepts of -1.45 and 0.11.
E. Sand Lake was 9% higher on average than the 7th ft C site, with
an intercept of -1.00.
Permanent/Group 1 Site Comparisons (8-Hour Block Data)
F. The 7th & C and Spenard & Benson sites have been compared to
Group 1 sites in a previous section.
G. The Garden site correlated at 0.80 or better with four other
Group 1 sites (sites 19,21,24, and 26).
H. The Sand Lake site did not correlate at 0.80 or better with anv
Group 1 sites.
Permanent/Garden Network Comparison
1. The 7th & C, Spenard & Benson and Sand Lake permanent sites did
not correlate at 0.75 or better with any Garden study site
Permanent/Sand Lake Network Comparison (8-Hour Block Data)
J. The 7th & C, Spenard & Benson, and Garden sites correlated at
0.74 to 0.80 with site 29 in the Sand Lake network. In
addition, Spenard & Benson correlated at 0.87 with site 28 in
the Sand Lake Network.
Permanent/Spenard & Benson Network Comparison
K. The 7th & C and Garden permanent sites correlated at 0.75 or
better with five (sites 201 ,202, and 203 on a 4-hour block
basis, and sites 204 and 38 on an 8-hour block basis), and one
(site 202) sites respectively in the Spenard & Benson network
The Sand Lake site did not correlate with any Spenard network*
site at 0.75 or better.
Other Network to Network Comparisons
L. Sites 201 and 202 of the Spenard grid network correlated at O 75
or better with one (site 103) and two (sites 102 and 103) sites
respectively in the Garden network.
M. Twelve of eighteen Group 1 sites and no Group 2 sites correlated
at 0.75 or better with site 38 (in the neighborhood adjoininq
Spenard and Minnesota). a
Table 13 Correlation/Regression Results* for Permanent Sites
IMO. SITE DEP. SITE COHHfLAT1QN COF.r.DETeR, NO,OF FAIRS 2-NAX RCCRLSSION CO. 1-WAT REGRESSION
7 AC
SPBt
O.7R0
7AC
CARD
0,151
7 AC
SULK
0.692
SPBC
CARD
0.737
SPBE
SULK
0.796
CARD
5DLK
0.70)
0.609
0,565
0.479
0,343
0.633
0.494
417.0
Y
a
1 .9B3X
~
•0.616
Y
¦
1.35 IX
434.0
1
a
1 .11662
~
•1.449
t
a
1.2I6X
410.0
*
¦
1 .0951
~
•1.002
If
a
o.m*
*
422.0
y
a
0.904X
~
•0.677
*
a
0.6MX
~
39R.0
r
a
0.551X
~
-0.609
f
a
0.491*
~
416.0
a
0.S7RX
~
-0.063
T
a
0.474X
~
* - Reflects Data for Six 4-Hour Blocks per Day
- 52 -
1.23?
0.404
0.039
0.460
•0.292
0.34*
-------
QUALITY ASSURANCE
The number and diversity of study sampling regimes necessitated a
comprehensive and highly coordinated approach to yield data of appropriate
precision, accuracy, and completeness. As referenced previously, a rigorous
quality assurance (QA) program was developed and applied to the study. An
enormous body of quality assurance documentation was amassed during the study
in support of datai quality. This QA program was composed of three basic
elements: sampling QA, analytical QA, and data handling QA to preserve both
the integrity and completeness of the data.
Sampl ing QA
Explicit and routine field QA protocols were designed and implemented to
ensure that samples being collected were both representative of ambient CO
concentrations at the individual sampling sites and comparable in terms of
quality to samples collected elsewhere in the study network. Measures of
sampling performance are described below.
Sampling Precision
A pair of integrated samplers was collocated (within 2 meters) to quantify the
extent of variability associated with the sampling method. The results in
terms of 8-hour averages are described below:
A. Collocated samplers 9 and 10 correlated at 0.97 with a slope of
nearly 1.0.
B. The mean difference between sites 9 and 10 was 0.0, with
individual differences ranging from -0.9 to 1.6 ppm.
C. Eighty percent of the differences between sites 9 and 10 were
within + 0.6 ppm (leaving only two differences greater than _+
0.6 ppm: -0.9 ppm and 1.6 ppm).
Sampling Accuracy
Pairs of integrated samplers/permanent monitors and sequential
sampler/permanent monitor were deployed in the study network. While in the
traditional sense this was not a true audit of sampler accuracy since an
absolute standard was not directly employed, it was assumed that data
retrieved from the permanent monitors were of sufficiently higher or at least
less variable quality that they were considered a "quasi" audit source. This
also provided a measure of method comparability. The results are described
below:
A. Collocated sites 11 and 7th & C correlated at 0.95 with the
permanent site running about 3% higher on average than site 11.
B. Absolute differences between site 11 and 7th C ranged from
-1.1 ppm to 1.1 ppm.
C. Collocated sites 25 and Spenard correlated at 0.90 with the
Spenard & Benson permanent site running about 5% higher on
average (at an intercept of -1.52) than site 25.
- 53 -
-------
D. Of the 43 total pairs of data resulting from the comparison of
site 25 and the Spenard <5 Benson permanent site, differences
ranged from -1.8 ppm to 4.8 ppm. Of these, 22 were within +_ 1.3
ppm and 38 were within + 2.6 ppm. Seventeen of the 21
differences greater than +1.3 ppm occurred when one or both
members of the pair were Tess than 9.0 ppm.
(Because the variability exhibited by this particular pair of
sites was greater than that of other collocated pairs, it became
the object of further investigation towards accounting for the
source of the variability. It is now thought that both sampling
devices were operating within their normal respective limits and
that the source of most of the few but large differences is
attributable to site-specific peculiarities.)
E. Using 4-hour block data, collocated sites 105 and Garden
correlated at 0.95 with the permanent site running 18% higher on
average than 105 (a systematic dilution problem associated with
the sequential method is suspected as the source of this offset).
Study Sampling Method Comparison
An integrated sampler and a sequential sampler were collocated to provide an
index of the comparability of these two methods. Collocated sites 7 and 301
correlated at 0.90 with the integrated sampler (site 7) running about 12%
higher on average than the sequential sampler (the same sequential dilution
problem referenced in "C" above is suspected as the source of this difference).
Analytical QA
Analytical performance was continually monitored to ensure the integrity of
the study data. Two measures of this analytical performance are described
below.
Analytical Precision
The Beckman Model 366 CO analyzer used for analysis of bag samples was
challenged 5 to 10 times daily with a precision atmosphere traceable to the
National Bureau of Standards (NBS). Due to the limited availability of
precision materials, the level of these precision checks varied from 12% to
36% of analyzer range. It should be noted that the precision confidence
limits were computed using the percent difference of analyzer response to the
known concentration of the precision atmosphere. As the proportional
difference of a fixed absolute difference is greater for lower than higher
concentrations, results from each precision level were considered
individually. The results of these checks are described below.
A. For the 24 precision checks performed at the nominal 6 ppm
level, the absolute differences ranged from -0.7 ppm to 0.3 ppm,
with percent differences yielding an upper 95% confidence limit'
(CI) of 5.62 and a lower 95% CI of -5.48.
- 54 -
-------
B. For the 84 precision checks performed at the nominal 9 ppm
level, the absolute differences ranged from -0.7 ppm to 0.3 ppm,
with percent differences yielding an upper 95% CI of 6.42 and a
lower 95% CI of -2.20.
C. For the 102 precision checks performed at the 17 ppm level, the
absolute differences ranged from -0.7 ppm to 0.3 ppm, with
percent differences yielding an upper 95% CI of 2.31 and a lower
95% CI of -2.01.
Another routine measure of analytical precision was the re-analysis of a
number of samples to ascertain the cumulative variability associated with the
analytical protocol. The results of these checks are not presented here.
Analytical Accuracy
The CO analyzer was audited two times with test atmospheres traceable to MBS
to evaluate its response to known and absolute concentrations of CO. The
number of audits do not merit a statistical treatment of the resulting data.
The results of these audits are summarized below:
A. The first audit (performed on January 5, 1983) indicated that
the analyzer was reading between 2.0Z and 4.4% high.
B. The second audit (performed on February 18, 1983) indicated that
the analyzer was reading between 0.4^ and 2.6% high.
Data Hand! i ng QA
The abundant amount of data arising from the study and the numerous
manipulations it underwent created the potential for errors in transcription,
processing, and computation. A rigorous program of routine checks was
instituted towards identifying and correcting these errors. The results of
this program will not be presented here. However, this program was extremely
effective in minimizing and eliminating these errors.
Data Completeness
Data completeness is a function of two somewhat competing objectives: 1)
maximizing the proportion of successfully collected samples relative to all
sampling attempts while 2) preserving the fundamental and desired integrity of
the data base. The stated objective of this study was to successfully capture
valid data for 85% of the total attempts (validity criteria was defined in the
study QA plan). Data recovery rates for each of the subject study sites are
shown in Table 13.
- 55 -
-------
Table 14 Valid
Data Recovery Rates
3y Site
"m"* Oata
"PM"** Data
8-Hour Data
Group 1
Recovery Rate
Recovery Rate
Recovery Rate
Si tes
(i n percent)
(i n percent)
(in percent)
1
85.2
96.3
81.5
5
96.3
85.2
83.3
6
92.6
88.9
83.3
7
81.5
94.4
75.9
8
90.7
79.6
77.8
9
98.1
87.0
85.2
11
92.6
92.6
85.2
12
96.3
88.9
85.2
13
96.3
94.4
92.6
15
81.5
88.9
74.1
18
90.7
96.3
90.7
19
92.6
75.9
70.4
20
96.3
90.7
90.7
21
94.4
88.9
85.2
24
90.7
90.7
81.5
25
96.3
87.0
85.2
26
96.3
92.6
90.7
27
90.7
94.4
85.2
Sand Lake
Si tes***
28
85.7
29
85.7
30
82.1
* - 11:00
a.m. to 3:00 p.m.
** - 3:00 p.m. to 7:00 p.m.
*** - Reflects eight-hour interval of 8:00 a.m. to 4:00 p.m.
- 56 -
------- |