-------
volumes. The first volume describes the conclusions of the
program, provides a brief history, the quality assurance
performed on the data collected, a summary of the data results,
the interpretation of the data set in terms of how the
concentrations measured during the monitoring program represent
ambient air quality conditions in the metropolitan area, and the
health risk associated with these concentrations. Volume two
contains the appendices.
Work will continue on an informal bases on the
interpretation of the data set. An area of interest regarding
future work involves additional quality control and assurance of
the data. Comparison of the data collected during the program
with similar data collected concurrently by the recently
completed Denver Metropolitan Brown Cloud study and with previous
studies on Denver's ambient air quality will also be a quality
control measure of the data base.
The following list of names of individuals has been
developed for the review of the attached report. Please review
the list for your name and the area of the report you are
requested to provide comments.
General Review
EPA Headquarters-Wash. D.C.
Art Koines ( 3 Copies )
State of Colorado
EPA Region VIII
Steve Arnold, APCD
Frank Rogers, APCD
Alan Dunhill, APCD
Gordon Pierce, APCD
Larry Svoboda, ESD
Ken Lloyd, IEMP
Wm. Basbagill, ESD
Gordon Macrae, ESD
EPA Research Triangle Park-( RTP )
Gerald Akland, RTP
Specific Review
Annular Denuder and Particulate Data
Robert Stevens, RTP
Thomas Dzubay, RTP
-------
Aldehyde Data
Silvestre Tejada, RTP
Roy Zweidinger, RTP
Volatile Organic Compound Data
Gerald Akland, RTP
William McClenny, RTP
William Laxton, OAQPS, ( 2 Copies )
PUF Data
Nancy Wilson, RTP
Exposure Assessment
Richard Moraski, ORD-Region VIII
Risk Assessment
Suzanne Wuerthele, Region VIII
Judy Graham, RTP
Charles Ris, HHAG
Upon completion of your review, you should provide me with a
memorandum which briefly summarizes your review and which either
approves, disapproves, or approves the report subject to
specified changes. Please also send the author, Mark Komp, a
copy of your memorandum at the above address. His mail code is
8AT - AP. If there are problems, I suggest that you resolve them
directly with Mark Komp. He may be reached at 303-293-1768 (FTS
564-1768). The review should be for technical and policy content
and not for style or rhetoric.
If you have any questions, call me at 303-236-5061 (FTS 776-
5061 ) .
Attachments:
-------
REPORT
ON
THE AIR MONITORING PROGRAM
FOR
THE DENVER METROPOLITAN AREA
INTEGRATED ENVIRONMENTAL MANAGEMENT PROJECT
PREPARED BY
MARK KOMP*
U.S. ENVIRONMENTAL PROTECTION AGENCY
REGION VIII
ENVIRONMENTAL SERVICES DIVISION
* NOW AFFILIATED WITH
U.S. EPA REGION VIII
AIR & TOXICS DIVISION
AIR PROGRAMS BRANCH
PLANNING SECTION
MR. DAVID SULLIVAN
SULLIVAN ENVIRONMENTAL CONSULTING, INC.
SUITE 28OC FORT HUNT CENTRE
1900 ELKINS STREET
ALEXANDRIA, VIRGINIA 22308
AND
INTEGRATED ENVIRONMENTAL MANAGEMENT PROJECT
VOLUME ONE
CONCLUSIONS
DATA SUMMARY
&
RISK ASSESSMENT
-------
Acknowledgements
The authors of this report wish to thank the following
individuals for their guidance, support, help and expertise that
they have provided in the development of this report. Mr. Jim
Lehr, Director of the Environmental Services Division-EPA Region
VIII; Mr. Irv Dickstein, Director Air and Toxics Division-EPA
Region VIII; Mr. Ken Lloyd, Director IEMP-EPA Region VIII. Mr.
Larry Svoboda, Chief, Environmental Monitoring and Assessment
Section-EPA Region VIII; Mr. William Basbagill, Environmental
Monitoring and Assessment Section-EPA Region VIII; Mr Gordon
MacRae, Environmental Monitoring and Assessment Section-EPA
Region VIII; Mr. Steve Frey, Environmental Enforcement Section-
EPA Region IX. Their help and support made the IEMP Air
Monitoring Program a success.
A special thank you goes to the many individuals at the
U. S. EPA 's facilities at Research Triangle Park, N.C.. The
individuals helped Region VIII secure the monitoring equipment
or assisted in the selection of the methods that enable the
Denver's IEMP Air Toxic Monitoring Program to collect data on
ambient air toxic concentrations in the Denver Metropolitan area.
These individuals and the monitoring areas that they assist in
are listed below.
Inorganics and 2.5 um Particulates Semi-Volatlles
Robert K. Stevens, ARSL Nancy K. Wilson, EMSL
Aldehydes and Canisters
Roy B. Zweidinger, ARSL
Silvestre B. Tejada, ASRL
William A. McClenny, EMSL
-------
TABLE OF CONTENTS
1.0 Conclusions 1
2.0 Introduction 6
3.0 Background 8
3.1 Objectives 8
3.2 Monitoring Approach 9
3.3 Monitoring Locations and Sampling Frequency .... 9
4.0 Data Validation 14
4.1 Routine Checks 14
4.2 Internal Consistency Checks 15
4.3 Historical Consistency Checks 21
4.3.1 Carbon Monoxide and Sulfur Dioxide Data
Comparisons 22
4.3.2 Particulate Data Comparisons 24
4.3.3 Volatile Organics Data Comparison 26
4.4 Consistency of Parallel Data Sets 27
4.4.1 Percent Relative Standard Deviation 28
4.4.2 Consistency of Parallel Data Sets of Related
Parameters 28
4.5 Future Data Validation 31
5.0 giifflwiafri on of Data Results 32
5.1 Carbon Monoxide and Particulate Data 32
5.1.1 Carbon Monoxide Data 32
5.1.2 PM-10 Particulate Data 35
5.2 2.5 um Particulate Data 37
5.3 Aldehyde Data 42
5.4 Denuder Data 4 5
5.5 Volatile Organic Compounds 53
5.6 Semi-Volatile Compounds 64
6.0 Interpretation of the Measured Air Toxics Data Set ... 66
6.1 Compilation of Data Bases 68
6.1.1 Measured Criteria Pollutant Data 68
6.1.2 Meteorological Data 69
6.1.3 Emissions Variability Data 70
6.2 Representativeness of Measured Air Toxics
Concentrations to Denver Air Quality 7 5
6.2.1 Regression Analyses Used to Support
Interpretation of air Toxics Data 7 6
6.2.2 Representativeness of Measured Air Toxics
Data to Estimate Long-term Averages at
Monitoring Sites 82
6.2.3 Review of the Influence on Control Measures
on the Representativeness of Measured
i
-------
Concentrations 85
6.2.4 Representativeness of Air Toxics Data
Collected at Monitoring Sites to Broader
Spatial Coverage within the Metropolitan
Area 90
6.3 Comparison of Concentrations With Other Metropolitan
Areas 91
7.0 Health Assessments 93
7.1 Exposure Assessments 93
7.1.1 Typical Exposure 93
7.1.2 MEI Exposures 97
7.2 Risk Assessment 102
7.2.1 Cancer "Cases" Over Metropolitan Area . . . 102
7.2.2 Noncancer Health Risks 103
7.2.3 MEI Risk 104
7.3 Limitations of Ambient-Based Exposure/Risk
Assessments 105
ii
-------
LIST OF FIGURES
Figure
4-1.
Carbon Tetrachloride Outliers in the Summer VOC
Data Base.••>••••••••••«•••«
17
Figure
4-2.
Decreasing 1,1,1,-Trichloroethane
Concentrations measured at Auraria Monitoring
Station
19
Figure
4-3.
Gradual Increase and Sudden Decrease in
Dichlorodifluoromethane Concentrations. . . .
19
Figure
4-4.
High Vinyl Chloride Concentrations Measured at
end of Sampling Period
20
Figure
5-1.
8hr Running Averages of CO Measured at the
Auraria Monitoring Station during the Winter
Period
33
Figure
5-2.
8hr Running Averages for CO Data Collected at
the Palmer Station during the Winter Monitoring
Period
33
Figure
5-3.
Slimmer PM-10 concentrations at Auraria. . . .
35
Figure
5-4.
Summer PM-10 Concentrations measured at
Arvada
35
Figure
5-5.
Winter PM-10 Concentrations Measured at
Auraria
39
Figure
5-6.
Winter PM-10 Concentrations measured at
Arvada
39
Figure
5-7.
Monthly Average Cadmium Concentrations for
Winter
40
Figure
5-8.
Monthly Average Chromium Concentrations for
Winter
41
Figure
5-9.
Monthly Average Lead Concentrations for
Winter
41
Figure
5-10
Monthly Average Formaldehyde Concentrations in
Summer
42
Figure
5-11.
Monthly Average Acetaldehyde Concentrations in
Summer
44
Figure
5-12
Monthly Average Propionaldehyde Concentrations
in Summer
44
Figure
5-13.
Monthly Average Formaldehyde Concentrations for
Winter
45
Figure
5-14.
Monthly Average Acetone Concentrations in
Winter
46
Figure
5-15.
Monthly Average Acetaldehyde Concentrations in
Winter
46
Figure
5-16.
Monthly Average Proionaldehyde Concentrations
in Winter
47
Figure
5-17.
Summer AM AND PM Nitrous Acid Levels in the
Denver Metropolitan Area
47
Figure
5-18.
Winter AM and PM Nitric Acid levels in the
Metropolitan Denver Area . ... .
48
Figure
5-19.
Winter PM Nitrous Acid Levels Measured in the
Metropolitan Denver Area
49
Figure
5-20.
Winter AM Nitric Acid Levels Measured in the
Metropolitan Denver Area.
51
iii
-------
Figure 5-21. Winter Nitrate Concentrations Measured at the
Auraria Monitoring Station 51
Figure 5-22. Winter Sulfate Concentrations Measured at the
Auraria Monitoring Station 52
Figure 5-23. Winter Benzene Concentrations 61
Figure 5-24. winter EthylBenzene Concentrations 61
Figure 5-25. Winter Toulene Concentrations 6 2
Figure 5-26. winter 4-Ethyltoulene Concentrations 62
Figure 5-27. Winter o-Xylene Concentrations 63
Figure 5-28. Winter m/p-Xylene Concentrations 63
iv
-------
LIST OF TABLES
Table 4-1. Data validation Parameters 14
Table 4-2. Historical Air Quality Data Collected by the
State of Colorado used in Comparison with IEMP
Data. 22
Table 4-3. Comparison of IEMP XRF Data to Previous
Study 2 5
Table 4-4. Comparison of IEMP VOC Data with VOC Presented
in the Literature 2 8
Table 4-5. 2.5 vs. PM-10 Comparisons 3 0
Table 5-2. PUF Composite Concentrations 64
Table 6-1. Relative Wood Consumption Rates 7 3
Table 6-2. Summary of Regression Analyses Based on
Reference Criteria Pollutants and Reference
Toxic Pollutants 77
Table 6-3 Equations for CO and Ozone as a functions of
Wind Flow Quadrant and AM/PM Periods 8 2
Table 6-4. Comparison of Observed to Predicted Relative
Concentrations for the Base Year and Five-Year
Data Set 84
Table 6-5. Concentrations Data Partitioned by Control
Options: High Oxygen Fuel Partitions 88
Table 6-6. Concentrations of CO and Heavy Metals by
Control Options coal/Gas Burn Days 8 8
Table 6-7. Comparison of Denver Air Toxics Concentrations
with Other Metropolitan Areas 91
Table 7-3 Summary of the Results of the MEI Risk
Analyses 103
v
-------
1.0 Conclusions
The Denver Integrated Environmental Management Project (IEMP)
Air Toxic Monitoring Program attempted to evaluate the level of
air toxic concentrations and their impact on the health of
residents in the Metropolitan Denver area. The evaluation was
achieve to a limited extent by the measurement of air toxic
concentrations during the summer and winter season of 1987/88, the
comparison of the measured concentrations to criteria pollutants
in order to establish temporal and spatial patterns, the
determination of the representativeness of the measured
concentrations for the exposure assessment, and the assessment of
health effects from the measured air toxic levels.
As a result of the success of the program an extensive data
base for air toxics has now been established for the metropolitan
Denver area. This data base has revealed that many of the toxic
pollutants measured during the IEMP Air Monitoring Program
followed the same temporal and spatial patterns that have been
exemplified by the monitoring of criteria pollutants in the Denver
area during the past several years. For example, the summer
sampling period indicated that toxic concentrations are generally
lower in concentration as compared to concentrations measured
during the winter period. The higher winter concentrations
occurred, the majority of the time, during periods when higher
concentrations of the criteria pollutants occurred. This would
suggest that, in general, high air toxic concentrations can be
expected during periods of poor air quality conditions.
Concentrations measured during the two monitoring periods by
the four monitoring stations Arvada, Auraria, National Jewish
Hospital, and Palmer Elementary school revealed that the downtown
Auraria monitoring station was subject to the highest
concentrations measured during the program. However, the
relatively small difference in the Auraria concentrations when
compared to the other three monitoring stations does not suggest
that unique air toxic sources contributed to the higher
concentrations at Auraria. Rather the location of the Auraria
Monitoring station near the central business district and the
inherent dispersion conditions of the area may have more of an
influence on the concentrations measured at the station than any
nearby sources of air toxics. The similar pattern of air toxic
concentrations measured at all of the monitoring stations also
suggests that the ambient air quality of the Denver metropolitan
area is not subject to a limited number of air toxic emission
sources located in one area but probably to many sources located
over the entire metropolitan area.
However, cautious interpretation of the IEMP Air Toxic data
base must be made due to the fact that validation procedures
involving the data revealed some problems. The most notable of
1
-------
these problems were within the Volatile Organic Compounds (VOC)
and 2.5 um particulate data. VOC measurements made during the
summer and winter demonstrated a wide variation in all 26 VOC
compounds measured. Many VOC compounds detected during the summer
were not detected during the winter. A review of some of these
compounds suggest that contamination, laboratory procedures, or a
combination of both resulted in some of the VOC concentrations to
be considered atypical for the area. Further review of the VOC
data suggests that despite some contamination problems the summer
data set appears to be well correlated with CO. This suggests
order within this data set. While questions were raised within
this report regarding the absolute values of several compounds
(Benzene/1,2, Dichloroethane, Vinyl Chloride, and 1,1,2,2,
Tetrachloroethane) , the summer data set appeared to be internally
consistent and compared well with national data sets. The winter
data set, on the other hand, was found to be of unsuitable quality
to support any interpretation. It appears that the modification
to the analytical procedure from Gas Chromatography-Electron
Capture/Flame Ionization Detection to Gas Chromatography/Mass
Spectrometer resulted in detection limits generally above
concentrations occurring in the ambient air.
Problems in interpreting data were also found in the 2.5 um
particulate data. Contamination from larger than 2.5 um
particulates being deposited on the filters resulted in the
organic and elemental carbon analyses of the data being biased and
prevented further review of these data from being performed.
However elemental analyses for arsenic, cadmium and chromium were
used. This was done because these metals were thought to be less
affected by the contamination problem because most of the mass for
these elements is expected to be present in the smaller particles
sizes (<2.5 um).
Additional data validation is needed for the entire data base
before it can be considered correct. Further comparison of
collocated samples, field blanks, and the correlation of
concentrations between the various monitoring locations may reveal
more outliers within the data base beyond those that have been
already been identified and removed.
With these concerns regarding the data base in mind, the
following compounds measured during the IEMP Air Toxic Monitoring
Program are compared with data collected across the nation.
Benzene/1,2 Dichloroethane (treated as Benzene) and Cadmium were
found to match the 90 percentile national value based only on
using summertime data. This should be interpreted to mean that
90% of the metropolitan areas around the nation having similar
data were found to have Benzene or Cadmium levels at or below the
levels measured in Denver. Chromium and Carbon Tetrachloride were
found to be fairly typical compared to the national data reviewed.
An average concentration developed from the Chromium and Cadmium
data taken at all of the monitoring stations were within the 50
2
-------
percentile of the national data. Average Formaldehyde
concentrations were found to be roughly a factor of 25-50% below
concentrations for typical metropolitan areas. Perchloroethylene
and Trichloroethylene average concentrations were also within the
50 percentile compared to the national average. Average
Benzo(a)pyrene concentrations were slightly higher than the 50
percentile. These compounds or elements are suspected human
carcinogens and due to their associated risk were highlighted in
the report. The remaining parameters measured during the
monitoring program were found at varying concentration levels but
their associated health risk, if any, was too low to be reviewed.
The atmospheric conditions that occurred during the
monitoring program were evaluated for their representativeness of
typical conditions. This was accomplished by comparing conditions
that occurred at the time of sampling with the previous five years
of meteorological data collected at Stapleton Airport. Although
meteorological data from Stapleton may not represent conditions
over the entire area, it does provide a first look of how
representative conditions were during the sampling program. Based
on this analysis the meteorological conditions which occurred
during the monitoring period appear to be consistent with average
conditions.
Qualitatively, it was inferred that the data from Auraria and
Arvada were reasonably representative of typical conditions within
the Central and Western Sections of the metropolitan area,
respectively. NJH was considered to represent impacts along
highly traveled traffic corridors, and Palmer used to best
represent typical exposures within the Eastern Section.
Based on the representativeness of monitoring data and the
qualitative assumption that the monitoring sites were
representative of the metropolitan area, an assessment of the
health risks from exposure to air toxic concentrations were made.
Each compound or element in which a health risk existed for that
compound was analyzed for its associated risk. The highest and
lowest concentrations for each compound or element measured during
the course of the two sampling periods were calculated for the
risk associated for both concentrations. Each compound or
element's high and low risk was added to determine the total
number of high and low "cases" of cancer that are anticipated to
develop from long term exposure to the concentrations. The total
cases over 70 years from exposure to the ambient concentrations of
the pollutants reviewed in this study ranged from roughly 500-900.
An estimate of approximately 7 00 "cases" was hypothesized to be
the best estimate. Based on a population for the Denver
Metropolitan area of 1.3 million people used for this report, 7 00
"cases" represents 0.05 percent of the population. It should be
strongly emphasized that this is an rough estimate of the number
of "cases" in which cancer develops from the exposure to air
toxics at levels measured during the IEMP Air Toxic monitoring
3
-------
program. The number of cases in which cancer develops as
presented in this report does not account for those individuals
who may later recover from their condition with the cancer in
remission. Nearly half of the cases were shown to be from
exposure to Benzene/1,2 Dichloroethane and 1,1,2,2
Tetrachloroethane. The following should be noted in interpreting
this data.
o Only a subset of the complex mixture of toxic air
pollutants within the Denver metropolitan area were
addressed in this study.
o Bias created by the limitations of the measured data set
(as noted in this report) should be considered when
interpreting the results.
o Actual risks may be substantially higher than shown in
this report for some pollutants where indoor exposures
may be much higher than ambient concentrations.
The IEMP Air Toxic Monitoring program provided a important
step in defining the level of air toxic concentrations in the
Denver area. It provided valuable experience on the operation of
an air toxic monitoring network, the analyses of the data, and the
interpretation of the data. The problems discovered in the data
base may be avoided in future monitoring based on the valuable
experience gained from the operation of the IEMP program. A
measure of the success of the program can be demonstrate in the
fact that the State of Colorado is now conducting aldehyde
monitoring at one of its monitoring stations. The aldehyde
monitoring is a result of the information gained from the IEMP
program.
However, more work involving air toxic monitoring, modeling
and risk assessment is needed. Risk assessments indicate that
Benzene levels in the Denver metropolitan area may be elevated
compared to the national level. This fact combined with the
problems found in the IEMP VOC data base suggest that additional
VOC monitoring be conducted in Denver to better assess VOC
concentrations in the metropolitan area. Difficultly in
interpreting < 2.5 um size particulate data due to bias in the
IEMP particulate data necessitates that additional sampling be
done. This would help to better define the elemental composition
of particulates found in Denver's ambient air. The modeling
performed to support the interpretation of the measured data set
may have general application for modeling air toxics or criteria
pollutants in future studies.
The exposure and risk assessment made several assumptions in
its interpretation of the data. Based on these assumptions the
estimate of risk to the residents of the metropolitan Denver must
4
-------
be considered a rough approximation. More analyses of the data
base and additional interpretation is needed before added
confidence to health risk interpretation for Denver's ambient air
can be made.
Finally, interpretation of the IEMP Air Toxic Monitoring data
must consider that the monitoring methods employed by the program
are evolving. Alternate monitoring methods for the same
parameters monitored by the IEMP program have been employed by
other studies in Denver and other metropolitan areas. Each method
is subject to its own amount of uncertainty. It is only through
longer term monitoring, than occurred during the IEMP program, can
these uncertainties become better defined and interpreted.
5
-------
2.0 Introduction
The Integrated Environmental Management Project (IEMP) is
part of the larger national demonstration program known as the
Environmental Strategies Project (ESP). The concern that national
regulations might not effectively handle problems unique to a
specific locale resulted in the establishment of the ESP program.
ESP involves communities across the country in defining,
evaluating and responding to local environmental problems. Pilot
projects (IEMP) were established as a vehicle in which local
communities could explore ways in which to improve environmental
management at the local level.
Currently, choices and decisions made in environmental
management are driven primarily by laws and regulatory
requirements which separate problems by media i.e. air, land, and
water. The magnitude of the impact of different problems is
sometimes not considered in setting priorities. Problems in each
media are usually not evaluated simultaneously to determine which
ones warrant immediate attention and funding. Local factors and
values that may play a significant role in efforts to reduce
pollution are often not taken into account.
The Denver IEMP program, like previous pilot projects,
responded to the above concerns by concentrating on two
components:
1. Providing technical information on ambient air quality
levels to be used in the development of risk
assessments. These assessments would, in turn, be used
in the decision making process for the development of
environmental priorities.
2. Involve local decision makers from many levels of
government and leaders from business, scientific,
citizen, and environmental communities in objectively
developing these environmental priorities.
It is the work performed in providing technical information to the
first component listed above that this report summarizes.
Specifically, this report summarizes information gather regarding
air toxic concentrations in the Denver metropolitan area. By
comparing the potential problem regarding air toxic concentrations
in the Denver area using risk assessment, a criteria for assessing
the current situation, interpreting the results, defining the
environmental problem, if any, found during the assessment, and
developing management strategies is established.
With these objectives in mind, this report documents the
work conducted to obtain technical information on air toxic
6
-------
concentrations and the associated risks for the Denver area. The
goals of this report are listed below.
1. Evaluate the level of concentrations of air toxics
measured in the Denver area.
2. Validate that the concentrations measured are correct.
3. Assess the health and welfare effects of the air toxic
concentrations measured using risk assessment and other
quantitative techniques, relying primarily upon
available environmental and health data.
4. Establish the representativeness of the data for the
ambient air quality conditions that exist in the Denver
area.
5. Demonstrate that the measured air toxic concentrations
can, in specific cases, be attributed to air emission
events.
6. Perform air dispersion modeling for selected areas where
significant concentrations of air toxics are
anticipated.
With the attainment of these goals, the information will be
peer reviewed and, in turn, be given to local community leaders
for their assessment in setting environmental priorities.
The portion of Denver's IEMP program described in this report
was extensive in its goals. Considerable time was spent preparing
and operating the program, and analyzing the results. Analyses of
the data is still being performed. Consequently this report is
considered to be a draft and describes only a portion of the
conclusions that have been drawn at the present time. A subsequent
final report will summarize all of the results and conclusions.
7
-------
3.0 Background
Previous studies conducted in the Denver area have
investigated the composition of air pollution in the area
(Russell, 1977 and Heisler, et. al., 1980) but these studies did
not examine the gaseous and particulate toxic compounds associated
with mobile and stationary sources. The work performed by Lewis,
et. al. (1986) examined mobile and stationary sources but was
restricted to the study of these sources' contribution to
particulate concentrations. In order to evaluate the human health
risks associated with breathing both gaseous and particulate toxic
compounds that may be present in metropolitan Denver's ambient
air, the IEMP field monitoring program for air toxics was
developed.
3.1 Objectives
One purpose of the IEMP air toxics monitoring program was to
quantify concentrations of air toxic compounds at four sites that
were evaluated and considered to be representative for the Denver
area. The Denver air toxic monitoring program had several other
objectives and these objectives are outlined below.
A. Toxic compound concentrations were to be collected
that were known or suspected to be present in
Denver's ambient atmosphere based on previous
ambient data collection efforts or knowledge of
existing emission sources.
B. Sampling methods and analyses techniques were to be
used that were capable of detecting and quantifying
the toxic compounds suspected to be present in
Denver's atmosphere at their anticipated
concentration levels.
C. Available toxicological and other health related
data were to be used in comparison with the
measured concentrations of toxic compounds from the
monitoring program to give a perspective on the
risk associated with breathing Denver's air.
D. Agency-accepted data for determining cancer potency
and Reference Doses (RFD) for the inhalation of
toxic compounds were to be used, where available,
as part of the health related data used in the
comparison.
For those compounds where no carcinogenic risks factors
existed, a qualitative evaluation of health risks, if any,
associated with the inhalation of these compounds were made using
United States Environmental Protection Agency (EPA) accepted
8
-------
procedures and guidelines. In addition to examining carcinogenic
and non-carcinogenic risks from individual compounds, the
cumulative risks from the simultaneous inhalation of a series air
toxics were estimated.
3.2 Monitoring Approach
The methods utilized to monitor for the air toxic compounds
of interest involved both reference method instrumentation and
instrumentation which were considered to be in the development
stages. The program was designed with the assistance of the Office
of Research and Development (ORD) within EPA to utilize experts on
these developing technologies, methods and procedures. Air toxic
monitoring included five pollutant classes:
1. Volatile and Semi-volatile Organics
2. Aldehydes
3. Organic and Inorganic Inhalable Particulates
4. Carbon Monoxide
5. Nitrates and Sulfates (Particulate Phase)
In determining the list of compounds to monitor, a
comprehensive list of compounds was reviewed and selection of the
above pollutants was made based on several factors. A discussion
of the selection process is given in the air monitoring plan for
this project (Komp et. al., 1988).
For the purpose of the IEMP study, a series of analyses were
conducted to determine whether sampling for less than a one year
period could produce ambient data which would adequately represent
annual averages. The results suggested that sampling during the
summer season (June through September) and the winter season
(November through February) would be sufficient to conservatively
estimate an annual average (Versar, 1987a).
The number and location of the monitoring sites were selected
on the basis of a series of statistical analyses of criteria
pollutant data for a three year period (Versar, 1987). The
analyses demonstrated that three monitoring sites could represent
ambient concentrations for the metropolitan Denver area. The use
of three sites as being representative of the metropolitan Denver
area was attributed to differences in topographical,
meteorological characteristics, population distribution and, the
distribution of sources of air toxic emissions in the Metropolitan
Denver area.
3.3 Monitoring Locations and Sampling Frequency
9
-------
The primary monitoring location was located at the Auraria
Community College Campus site near Speer Boulevard and Larimer
Street in the central business district. The Auraria site was
chosen over other existing stations in the area based on the need
to select a site representing average exposure to the Denver work
force who commute daily downtown and the permanent downtown
residential population. The second and third monitoring locations
were located at Arvada and National Jewish Hospital (NJH). The
Arvada site (57th and Garrison) represented the western
geographical area and meteorological conditions. The area
represented medium density residential neighborhoods and was
considered, based on historical air quality data, more heavily
influenced by residential woodburning than other areas consider
for monitoring locations by IEMP. Conversely, the NJH site (14th
and Albion) represented the eastern most metropolitan area. The
site was located near a major arterial intersection (Colfax Ave.
and Colorado Blvd.) and was selected for the IEMP project in an
attempt to quantify the risk associated with Denver residents who
are exposed to emissions from mobile sources that utilize these
arterial highways.
A fourth site was added to the monitoring program when some
concern was raised regarding whether the NJH site would truly be
representative or ambient air concentrations for the eastern
metropolitan residential area. Thus, a fourth site was added at
the Palmer Elementary School (995 Grape Street). Figure 3-1
depicts the location of all of the sites.
Each site utilized the same type of monitoring equipment,
however, the equipment configuration was unique to each site.
Table II of the monitoring plan (Komp, et. al. 1988) provides a
list of the equipment used at each site. The sampling frequency at
which the equipment operated was based on two priories. These
priorities were the development of high quality annual average
exposure data for risk assessment, and a statistically valid data
base for subsequent source apportionment analyses.
The sampling frequency considered the establishment of one of
the sites as a primary station site where samples are to be
collected more frequently than the other sites.
10
-------
Figure 3-1,
Location of Monitoring Stations
11
-------
For IEMP it was determined that the Auraria site, because of its
downtown location, would collect samples every third day while the
three remaining sites would collect samples on a less frequent
basis. This less frequent basis was determined to be a one in six
day schedule and emulated the State of Colorado's one in six day
sampling schedule during the two monitoring periods. Daylight and
nighttime sampling was also initiated at the Auraria site to
assess any differences in ambient air quality between daytime and
nighttime air toxic emissions in the downtown area.
The expectation that specific meteorological conditions would
contribute to high concentrations of ambient air toxic
concentrations necessitated different sampling times and duration
of sampling during the two monitoring periods. During the summer
period, the day and night sampling consisted of two 12 hour
sampling periods (7am through 7pm and 7pm through 7am). However
during the winter period with the shorter daylight periods, the
sampling times were modified to conform to the anticipated change
from daylight to nighttime dispersion regimes. Air sampling
equipment at Auraria operated from 9am through 4pm (7 hours of
sampling). Nighttime sampling occurred during the hours of 4pm
through 9am (17 hours of sampling). This sampling schedule was
also incorporated into the metro-Denver Brown Cloud Study and
allowed for a comparison of data between the two studies. During
part of the winter monitoring period the sampling schedule for
nitrates and sulfates (gas and particulate phases) being measured
at the Auraria site was changed from a one in three day sampling
schedule to an everyday sampling schedule. This variation in the
sampling schedule was performed to help support the 1987-88 Denver
Metro Brown Cloud Study, which was also being conducted during the
IEMP winter monitoring program.
The remaining three monitoring locations operated only on a
24 sampling schedule. Sampling began at 7am in the summer and 9 am
during the winter and ended 24 hours later. The specific days that
sampling occurred during the two monitoring periods are listed in
Table III of the monitoring plan.
Smaller studies conducted for varying lengths of time were
added to the winter monitoring period. Specifically, these studies
included the following:
1. Everyday sampling for acids, nitrates and sulfates
at the Arvada IEMP monitoring site.
2. Sampling for acids, nitrates and sulfates on the
roof of the downtown Federal Building (1929 Stout
Street).
The remaining sections of this report describe the data
validation employed in analyzing the data, a summary of the
results obtained from the data analyses, interpretation of the
12
-------
results, and the health and risk assessments determine from the
interpretation of the results.
13
-------
4.0 Data Validation
An essential element of the IEMP air toxics monitoring
program was the validation of the data collected during the two
monitoring periods. Data validation refers to the methods
performed after the data have been collected and serves as a
screening process to ensure that the data is correct before the
data were used in any decision making process. The data validation
attempted to prevent erroneous data from becoming a part of the
overall data set and provided for an overall review of the data.
Validation, therefore, became an integral factor in the successful
data analysis regime. The parameters measured at the monitoring
stations, the frequency of sampling, and the laboratory analyses
performed are listed in Table 4-1 . Data validation was performed
on the results of the laboratory analyses.
Data validation for IEMP air toxics monitoring program
followed the systematic process outlined in the EPA document
Validation of Air Monitoring Data (U.S. EPA, 1980). The procedures
outlined in this document were applied, where possible, to the
parameters measured at each of the four monitoring stations that
were a part of the air monitoring program. The procedures that
were appropriate for the IEMP air toxics monitoring program were
routine checks of the data, tests for internal consistency, tests
for historical consistency, and tests for consistency within
parallel data sets. Each of these procedures are described in the
remainder of this section.
4.1 Routine Checks
These checks consisted of examining flow rates, duration of
sampling, problems with the sampling equipment as noted by the
site technician, unusual sampling events, and performance checks
that were routinely conducted on the equipment. A checklist was
maintained for each of the parameters in which routine checks were
performed. This checklist provided a convenient method of tracing
the number of checks performed for each data set and provided
documentation that the checks were in fact performed.
Improper sampling times were assessed as to whether the data
could still be considered representative for the sample day. Flow
rates were carefully considered since along with the sample times
they directly affected the concentrations reported. A check of the
flow rates, and duration of sampling was accomplished by referring
to the spreadsheets which the contractor, assigned to the project,
had developed. These spreadsheets are contained in the project
report completed by the contractor (PEI, 1987 & 1988). These
reports also provide documentation of the performance checks, i.e.
calibrations and audits, performed on the equipment periodically
throughout the sampling program. Calibrations were used to
adjust, where necessary, flow rates to reflect any instrument
14
-------
Table 4-1. Data Validation Parameters
IEMP Data Collection and Analytical Methods
Parameter/
Method
Sampling
Frequency
Laboratory
Analysis
Carbon Monoxide/
EPA Reference Meth.
Particulates/
2.5 um
PM-10 um
Volatiles/
Canisters
Semi-Vglatiles
PUF
Inorganics
Annular Denudere
Aldehydes
Dnph
Continuous
1 in 3 or 6 day
1 in 3 or 6 day
1 in 3 or 6 day
1 in 3 or Every day
1 in 3 or 6 day
None
Mass,
Carbon
Mass
XRFa &
GC/FID/ECD
GC/MS
ICc
IC
HPLC
a. X-ray Fluorescence
b. Gas Chromatography/Flame Ionization Detection/Electron
Capture Detection for summer sampling period and Mass
Spectrometer during the winter sampling period.
c. Ion Chromatography
d. Polyurethane Foam Sampler
e. Sampling frequency was modified during the winter sampling
period to every day sampling.
f. Dinitrophenylhydrazine reagent coated cartridge.
deviation from its proper response. Audits were used to assess the
adequacy of the quality control of the program.
Unusual sampling events or problems with the equipment were
obtained from the site technician's log book and assessed by the
reviewers of the data for their impact on data results. Adjustments
made to the data as a result of errors discovered during the
routine checks were documented on the checklist and/or the project
report.
4.2 Internal Consistency Checks
15
-------
Internal consistency tests examine data set values which
appear, upon first review, to be atypical when compared to other
values within the particular data set being examined. Common
anomalies of this type include unusually high or low values known
as outliers which result in large differences in adjacent values
sampled before or after the outlier value. These outliers can
usually be attributed to incorrect operation of the sampler,
contamination of the sample, or the incorrect calculation of the
concentration from flowrate and laboratory analysis of the sample.
However, in some cases an explanation of the outlier can not be
determined. At this point it is the subjective decision of the
agency or individual reporting the results as to the determination
of the validity of the data point.
Plotting of the data set is one of the most effective means
of identifying possible data outliers. It was used exclusively in
performing internal consistency checks of the data collected by the
IEMP monitoring program because of its effectiveness in identifying
unusual data that would not ordinarily be identify using other
internal consistency tests.
For the Denver, IEMP data, plotting for internal consistency
checks involved plotting of data with time. Usually a plot was made
for the entire sampling period in order for the all the data points
to be compared to one another. The ability to examine all of the
data at once in order to establish patterns within the
concentrations measured aided greatly in establishing the existence
of possible outliers. It also provided a means of determining
whether the outliers were grouped within a specific time period.
If the outlier had occurred during one specific time period, the
cause of the outlier could be attributed to a possible event during
the sampling program that had been documented either by site
technician or reviewers of the program.
The most numerous examples of outliers in the IEMP air toxics
monitoring program data set can be found in the VOC data. Unusually
high concentrations were found in both the summer and winter data
when compared to adjacent values collected prior to or after the
outlier value. Many of these outliers can be explained by events
which occurred during sampling or by procedures utilized during the
laboratory analyses. However several occurrences of outliers could
not be documented adequately and were removed from the data base
on the subjective opinion of the reviewers of the data. A
discussion of the outliers and figures depicting the erroneous data
are presented in the following paragraphs.
A review of the summer VOC data revealed that laboratory
analyses of all 26 compounds demonstrated unusually high
concentrations during the first samples taken in June 1987. Figure
4-1 depicts the plot of Carbon Tetrachloride concentrations for the
summer monitoring period. The Carbon Tetrachloride example in the
figure is typical of the first sample taken for all 26 compounds.
16
-------
In the figure the first VOC sample taken at the three monitoring
stations were all unusually high compared to samples taken later
in the monitoring period. It has been suggested by several of the
reviewers that these high concentrations can be attributed to the
improper purging of air toxic compounds within the canisters and
the samplers prior to the receipt of the equipment and its use in
the IEMP air toxics sampling program. The first sample purged the
sampler of any "contaminated" air. Unfortunately the purged air was
collected by the canisters and sent in for analysis. Subsequent use
of the canisters samplers apparently alleviated the contamination
problem and it is believed that representative samples were
collected for the remainder of the period. Although there is no
strong evidence to support this opinion, the fact that all 2 6
compounds were unusually high for the first sample gives support
to this opinion. As a result of the outliers, all VOC data for the
2 June 1987 sampling period were removed from the data base.
2.2
2 0
1 B
1 6
r\
I
5 12
| 1 0
H 0 8
0 6
0 4
0.2
0.0
June 87 July 57 Aug, &? 5ept, 87
MONlTOQirc PS5100
o ARVAC* COLLOCATED 1 KPVtf* F^IM^RY i.OfilDe
• <-
AWAQA FRlMfcOT OUTlIBi
' <-
NJH Ol/TL fB?
, <-
AURARIA OUTLIER
)<-
AJVADA COLLOCATED CVTLIB5
4
6
0
0
6 a 4 4 4 I
4 6 8 *
I t I 4 »
A
+ 8
d.i
* $ *+ ..
t t A A 0 0 A (
i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—r
Data collected for the compound 1,1,1-Trichloroethane during
the summer monitoring period indicated that a decreasing
concentration of the compound with time occurred at the Auraria
monitoring station. It was suspected that this pattern of
concentrations at Auraria was erroneous. Figure 4-2 depicts the
pattern of concentrations at the monitoring stations that sampled
the compound. Concentrations of 1,1,1-Trichloroethane measured at
17
-------
Auraria decreased until mid-July axsd remained stable for the
remainder of the monitoring period. At the same time concentrations
of the compound at the Arvada and NJH monitoring stations remained
stable during the entire monitoring period. Two explanations for
the pattern of decreasing concentrations with time for this
compound measured at the Auraria monitoring station have been
suggested. First, a source of the compound near the Auraria station
may have released of the compound in decreasing amounts with time.
Second, 1,1,1-Trichloroethane may have been present in the Auraria
canister sampler prior to its installation in the field and several
sample runs were necessary to purge the sampler of the compound.
Given that concentrations measured at the other stations did not
decrease with time and that the concentrations measured at Auraria
remained stable after Jnid July, the second explanation of
contamination within the sampler appears to be more plausible.
Consequently, 1,1,1-Trichloroethane data collected at the Auraria
monitoring station from 2 June 1987 through 8 July 1987 were
removed from the data base.
18
-------
IEMP SUMMER VOC DATA
40 . Q
35. D
30. Q -
r\
a
& 25 0 "
w A<-AURARIA pre decreases with time
z
9 20.0 -
I-
s
O 15 0 -
8 *<-
10.0 -
• ^A
° +
5 a - . a a
4 A A A t 0
nn r*! D 9 i a e A a * I * « * • 8AB*iftA(i
D'D T i—:—:—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—
June July Aug Sept.
MONITORING PERIOO 19B7
~ ARVADA COLLOCATE) + ARVADA PRIMW 0 NJH a AURARIA
Figure 4-2. Decreasing 1,1,1,-Trichloroethane Concentrations
measured at Auraria Monitoring Station.
Figure 4-3 depicts concentrations of Dichlorodifluoromethane
measured during the summer at the VOC monitoring stations. An
increase with time in the concentration of the compound followed
by a sudden decrease at all the stations is seen in the figure.
This increase was considered unusual and further examination of the
laboratory results was initiated. During the review, conversations
between the contractor and their laboratory personnel (Zimmer,
1988a) indicated that Dichlorodifluoromethane had been a difficult
compound to analyze. Problems analyzing the compound were resolved
during the course of sampling. Although there was insufficient
evidence to conclusively indicate that all reported concentrations
of the compound measured through the middle of July were incorrect,
the tendency for the concentrations to increase through mid-July
followed by a sharp decrease made the results questionable. As a
result Dichlorodifluoromethane concentrations reported for all
stations through 14 July 1987 were removed from the data base.
Towards the end of the summer sampling period high values of
Vinyl Chloride were observed from the laboratory analyses of
1,1,1,-Tfl ICHLOROETVWNE
A<-AURARIA PPB DECREASES WITH TIME
A<-
A
A ^ A A ^ A
-f ^ A A 0
~ 94 9 e 4 ® A ¦ 6 i ft A
I 1 1 1—I 1 1—I 1 1 1 1—J 1—I 1 1—I—J J—J—I 1—I—J J—I !~
19
-------
A
8
13 T
12 -
11 -
10 -
9 -
0 -
7 -
6 -
5 -
4 -
3 -
J -
1 II
0 --
June
EMP SUMMER VOC DATA
DICHLORCOI FLUCPCfcETHANE
~ <-DE CREASE IN PP6
A A
0
~
o
1—i—i—i—i—r
4 * 9
M H A
i { ' «
A A • $ L (|
i i i i i r
July
n—I—i—i—r
l—i—i—r
Aug
i—j—i—i—i—i—i—r
i r
Sep-t
ARVADA COLLOCATED
MONITORING PER I CO 1987
+ ARVADA PRIMW
MJH
AURAR IA
Figure 4-3.
Gradual Increase and Sudden Decrease
Dichlorodifluorometbane Concentrations.
in
canister samples taken at the three monitoring stations. A plot of
the concentrations against time is provided in Figure 4-4
Conversations with laboratory personnel confirmed that the
laboratory analyses for Vinyl Chloride and Vinylidene Chloride were
difficult to obtain from all of the summer canister samples due to
interference from other compounds (Zimmer, 1988b). Although plots
of Vinylidene Chloride did not show the occurrence of outliers as
was the case with Vinyl Chloride, concerns regarding the ability
to analyze for these compounds have been expressed by others (Komp,
1988) . Vinyl Chloride samples collected after 22 August 1987 were
deleted from the data base for all the VOC monitoring stations
based on the outliers which occurred after that date. However there
was no evidence that Vinylidene Chloride concentrations were
incorrect and they remained in the data base.
The occurrence of outliers in the VOC data collected during
the summer period and the concerns expressed by laboratory
personnel prompted a change in the analytical method used to
determine VOC concentrations from canister samples. During the
winter period VOC samples were analyzed using mass spectroscopy.
20
-------
EMP SUMMER VOC DATA
VIMiX CHLORIDE
240
i
220 -
200 -
180 -
150 -
140 -
120 -
100 -
BO -
BO -
40
20
0
June
~ o
+ A
0
1 i i r
~i—i—i—i—r
9 jM4i
A—Q 1
July
a & • a
~i i i i i r i i i i r
Aajg.
ARVAOA COLLOCATED
MONITORING FERIOO 19B7
4 ARVAEA PRIMARY
I *? i—i—I—r
Sept
NJH
AURAH IA
Figure 4-4. High Vinyl Chloride Concentrations Measured at end
of Sampling Period.
This improved the ability to determine the presence of a particular
compound although at the cost of some sensitive in the detection
limit.
The majority of the outliers in the IEMP air toxics monitoring
program data set were found in the VOC data and the examples
presented above were typical of the entire VOC data set. However
outliers occurred for all of the parameters measured during the
program. The total amount of data determined to be outliers from
the parameters not addressed in this section varied for each
parameter. The overall amount of data removed from the entire data
base because of outliers was approximately 7 percent.
4.3 Historical Consistency Checks
Checks for historical consistency compared the data set being
examined with similar data collected in the same or nearby location
in the past. This process is useful in detecting data averages or
individual data points that are considered unlikely to occur under
21
-------
the atmospheric conditions expected for the sampling area during
a specific sampling period. Step one of examination involves the
establishment of upper and lower limits expected for an individual
parameter based on the historical data being used. Upper limits
were established for the IEMP air toxics data by the use of the
maximum value observed over the sampling period. Lower limits were
assumed to be the nondetectable limit depending on the parameter
being reviewed. Next, where it was appropriate, averages were
established by reviewing the historical data usually on a month by
month or seasonal bases. These averages were either hourly, daily,
monthly or by the year depending on the format of the historical
data base. Finally, a pattern was established for the data set
under examination in order to assess a pollutant behavior which has
never or rarely occurs. For example, unusually high CO levels (>
2 0 ppm) occurring during the summer that are more likely to occur
during poor dispersion conditions usually found during the winter
season.
For the purposes of the IEMP study, historical consistency
checks were limited to a select number of parameters. A
sufficiently large historical data base exists in the Denver area
for the measured parameters of Carbon Monoxide (CO), Sulfur Dioxide
(S02), PM-10 particulates, and to a lesser extent sulfate and
nitrate data. However, many of the air toxic concentrations (i.e.
Volatile Organic Compounds (VOC) and Semi-Volatiles (PUF)) measured
by IEMP air toxics monitoring program are unique to the program.
Air toxic data have been collected in the Denver metropolitan area
by previous monitoring programs but in many cases these data were
collected by a variety of monitoring equipment not used by the IEMP
air toxics monitoring program. The historical air toxic data
reviewed had also been collected in locations that were a
considerable distance from the IEMP monitoring locations. Since the
historical data did not conform to the monitoring methods used by
IEMP air toxics monitoring program or were collected at a location
determine not to be representative of the IEMP Monitoring
locations, less importance was placed on the historical data base
for air toxics.
4.3.1 Carbon Monoxide and Sulfur Dioxide Data Comparisons
CO data collected during the last five years by the state of
Colorado Air Pollution Control Division (APCD) at selected
locations in the Denver metropolitan area were used as an
approximation of the upper limit to be expected for the
metropolitan area. Locations used in the averages were Welby (E.
78th and Steele), Camp (21st and Broadway), Carriage (2325 Irving)
and National Jewish Hospital (14th and Albion). These locations
were chosen because of their proximity to the IEMP air toxics
monitoring locations. Table 4-3 depicts the comparison of maximum
and minimums developed from the five year data base.
22
-------
Table 4-2. Historical Air Quality Data Collected by the State
of Colorado used in Comparison with IEMP Data.
Selected Air Quality Data Collected by APCDa
APCD Monitoring
Parameter Station Location
Welby Camp Carriage
NJH
Carbon Monoxide (PPM)
5 yr. lhr Max.
17
44
26
33
5 yr. 8hr Max.
13
26
20
22
Sulfur Dioxide0 (PPM)
5 yr. 2nd 24hr Max.
0. 02
0.04
X
X
5 yr. Max. Ann. Ave.
0. 008
0. 014
X
X
Adams City
Arvada
Camp
Gates
PM-10 Particulate (ug/m3)
J
1987 Annual Arith. Ave
46
20
32
38
1987 2nd 24hr Max.
. 109
81
93
93
1986 Annual Arith. Ave
.d 64
27
36
60
1986 2nd 24hr Max.
142
47
61
98
Sulfate & Nitrate Part.c
(ug/m3)
Sulfate
5 yr. 2nd 24hr Max.
18
X
30
X
5 yr. Max. Ann. Ave.
7
X
7
X
Nitrate
5 yr. 2nd 24hr Max.
30
X
31
X
5 yr. Max. Ann. Ave.
6
X
6
X
a. Data taken from 1983 thru 1987 APCD annual air quality
reports.
b. CO maximum based on two years of data. Welby site began
CO sampling in 1986.
c. Data was collected only at the indicated sites.
d. Average based on incomplete data recovery.
23
-------
Any data outside the limits established by the five year data
base were highlighted and received further review to determine if
in fact it was correct. For data validation purposes all of the
IEMP air toxics data was within the upper and lower limits
established from the historical data base. Section 4 of this report
describes the CO data collected as part of the program.
The use of historical Sulfur Dioxide (S02) data collected by
APCD was limited to the Welby and Camp Stations and was not a
direct comparison with S02 data collected by the IEMP air toxics
monitoring program. IEMP S02 data were derived from the annular
denuder measurements at the Auraria monitoring station. S02 data
collected by APCD was performed using continuous monitors employing
the pulsed fluorescence method. The difference in the two
collection methods was taken into account in the comparison.
Despite the differences in the two unique methods used in obtaining
S02 data, the IEMP data was within the limits established by the
5 years of data collected by APCD.
4.3.2 Particulate Data Comparisons
PM-10 data have been collected for a short time period in the
Denver area and a five year comparison was not possible. Therefore,
the amount of data used in the comparison was based on PM-10 data
collected during the 1986 and 1987 monitoring year. IEMP air toxics
monitoring program PM-10 data were within the limits set by the
historical data set and these data are described in detailed in
Section 4 of the report.
Sulfate and nitrate concentrations are presented in Table 4-
3. These concentrations were measured by APCD by extracting the
water soluble fraction of these two compounds from total suspended
particulate (TSP) data collected on glass fiber filters exposed by
high volume samplers. Similar to the situation regarding SO^ data,
the. APCD method of collecting sulfate and nitrate data differed
from the IEMP monitoring method. Collection of sulfate and nitrate
data in the IEMP air toxics monitoring program was performed using
the annular denuder and was restricted to the sulfate and nitrate
concentrations associated with particulates < 2.5 um in size. TSP
sampling performed by APCD includes a much broader size range of
particles. However, the sulfate and nitrate data collected by the
state does provide a coarse approximation of the upper limit
expected for these parameters.
Three nitrate values collected during the winter period were
above the upper limit established for the data. The first value was
collected at the Auraria monitoring station on December 18, 1987.
Upon further review of the laboratory results for this value, it
was determine that the original value (178 ug/m ) was a
typographical error. It was adjusted to 17.8 ug/m and incorporated
into the data base. The two remaining values were collected on the
same day at two different monitoring locations. On January 8, 1988
24
-------
a value of 39.9 ug/m was reported for the Auraria monitoring
station. On the same day a value of 35.4 ug/m was reported for the
Arvada monitoring station. These values are higher than the 5 yr
maximum report by APCD (30 - 31 ug/mJ) for the TSP size range of
particulates. As a result of these high values, additional
examination of the nitrate data will be performed and will be
reported in a subsequent report. All sulfate data were within the
limits of the historical data. However, additional analyses of the
sulfate will be undertaken to confirm the concentrations measured
during the program.
Historical comparisons of the fine size fraction (<2.5um)
particulate sampling conducted by the IEMP air toxics monitoring
program consisted of comparing X-Ray Fluorescence (XRF) analyses
of the data with similar analyses performed on past fine
particulate studies conducted in Denver. The XRF analyses provided
a description of the elemental
Table 4-3. Comparison of IEMP composition of the particulates
XRF Data to collected. A review of the
Previous Study. elemental analyses performed on
data revealed that the elemental
composition of the particulates
was not what was expected for
the metropolitan Denver area.
High concentrations in
Aluminum(A1), Silicon(Si),
Calcium(Ca), Iron(Fe), and other
elements characteristic of
coarse size particulates
(>2.5um) from soils were found
from the analyses. These data
were inconsistent with a
previous study (Lewis et. al.,
1986) of the elemental
composition of < 2.5um
particulates conducted in Denver
which indicated that fine
particulates consist of smaller
concentrations of the above
elements. The fine particulate
analyses performed for this
study indicated that elemental
concentrations were a factor of 10-100 less than elemental
concentrations reported for the IEMP Air Toxics Monitoring study.
Table 4-3 presents the comparison of the IEMP XRF analyses with the
previous study in Denver. IEMP data presented in the table are an
average concentration for the compound for the entire winter
sampling period. The Lewis study concentrations are an average
based on a data set that was collected for a 19 day period. The
IEMP data were collected at the Auraria monitoring station between
the hours of 9am and 4pm. The Lewis study collected data in
northeast Denver between the hours of 6am and 6pm. Despite these
Fine Part.
Cone.(ug/m )
from XRF
Selected
u
Element
IEMPa
Lewis
Al
1.07
0.4
Si
5.05
0.3
K
2.03
0. 06
CI
0.9
0.05
Ca
0.7
0. 04
Mn
0.04
0. 008
Fe
1.24
0. 08
a. Data
collected
Nov.-Feb
1987/88 from 9am-4pm.
b. Data collected Jan. 11-30,
1982 from 6am-6pm.
25
-------
differences, the comparison of the IEMP data with the Lewis study
resulted in further review of the particulate data. A possible
explanation for the high elemental concentrations was formulated
after a review of additional IEMP particulate data that were
collected at the same time as the fine size particulates. A
complete description of the review is highlighted in Section 4.4
which describes the data validation procedure of reviewing parallel
data sets.
Finally, plots of Dichloromethane (Methylene Chloride)
concentrations collected during the summer indicated unusually high
concentrations (> lOOppb) occurred periodically throughout the
sampling period. These concentrations were higher than any previous
study had shown for the Denver area. These high concentrations
appear random throughout the data for all of the monitoring
stations. Although there is insufficient evidence to conclusively
indicate that the data is erroneous, the high values suggest that
a contamination problem may have resulted in these values being
report. Methylene Chloride data for the summer have been removed
from the final data base until sufficient information is brought
forth to suggest that the data is representative of metropolitan
Denver area.
4.3.3 Volatile Organics Data Comparison.
Volatile Organic Compounds (VOC) were reviewed by comparing
the 2 6 compounds collected during the summer and winter IEMP air
toxics monitoring program with data that had been collected in the
Denver metropolitan area in the past. The comparison was possible
by using an assessment of available VOC data for the Denver area
prepared by the U.S. Environmental Protection Agency (EPA, 1983).
This assessment was designed to summarizes data that had been
collected from many studies across the country in an attempt to
develop a useful and coherent data base that typified VOC
concentrations. An emphasis was place on the quality and
representativeness of the data before it was placed in the
document. It is from this data base that the historical Denver data
was extracted. Some of the VOC compounds sampled by the IEMP
program had not been sampled by the previous studies. In these
cases data collected across the country for the particular compound
was used to develop and upper and lower limit for the
concentration. In all cases where VOC data was presented for the
Denver area in the EPA assessment, the data was extracted from two
documents. Several VOC concentrations for Denver were extracted
from the Denver Air Pollution Studv-1973 (Russell, ed. 1976) but
the majority of the data was taken from Atmospheric Measurements
of Selected Hazardous Organic Chemicals (Singh, et. al. 1979). The
methods by which the VOC samples were collected in these studies
varied and consisted of one or several of the following collection
methods.
26
-------
* canisters
* absorbing resin
* bag samples
* absorbing cartridges
VOC data for the IEMP program was collected exclusively
through the use of steel canisters. The use of a variety of
sampling methods between the studies summarized in the assessment
and the IEMP study makes a direct comparison of the two data bases
not practical. Therefore data collected in the previous studies was
use only to provide a coarse review of the IEMP data.
Table 4-4 depicts the range of VOC concentrations measured
during the summer and winter IEMP air toxics monitoring program
with the range of concentrations presented in the EPA literature.
The IEMP VOC data ranges from non-detected (ND) to significant
levels in parts per billion (ppb/volume). Ten of the 26 compounds
measured during the summer period were significantly above the
range of concentrations presented in the literature. During the
winter sampling period, six compounds were significantly above the
range of concentrations presented in Table 4-4. Compounds that were
above the range developed from past studies were given further
review.
4.4 Consistency of Parallel Data Sets
The examination of the data for outliers as defined in Section
4.2 assumes that the majority of the data are correct and can be
used as an accurate indication of the range of values found in the
ambient air during the sampling period. However, if the sampler
collecting the data has a bias in its collection method, a review
for outliers would not identify the bias. A method of identifying
a systematic bias within the data is to compare the data set with
similar data collected at the same time and location. The IEMP air
toxics monitoring program collected these data sets through the use
of collocated samplers for selected parameters at all of the
monitoring locations.
The collocated eguipment located at each site are presented
below:
* Arvada - 2.5 um particulates, PUF, DNPH Sampler, and
Canisters.
* Auraria - Annular Denuder
* NJH - DNPH Sampler
Examination of the parallel data sets consisted of a review of data
collected by the above sampling eguipment.
27
-------
4.4.1 Percent Relative standard Deviation
The comparison of collocated data to define any bias inherent
in the data is essentially a measure of the precision of the
sampling method. The procedure employed by the IEMP air toxics
monitoring program to assess the precision and hence the
consistency of parallel data sets was through the use of the
percent relative standard deviation (%RSD) of the data. As outlined
in Section 12 of the IEMP Quality Assurance Project Plan (QAPP)
(PEI, 1987), the %RSD is defined as the standard deviation divided
by the sum of the primary and collocated values times two hundred
to achieve the percent. Equation 4-1 represents the calculation
used. This calculation helps eliminate the wide variation that can
often occur when a simple percent difference between two small
values is used to determine precision.
Correlation of the collocated data with the primary data set
will be performed on the two data sets for all parameters. This
will be an extensive effort and will be unavailable for
summarization within this report. It is anticipated that future
reports will address the precision and accuracy of the data.
%RSD = 200 Sj^ 4-1
(Xi+ Yi)
where:
s^= Standard Deviation = (— YjJ/1.414
X^= Primary Value
Y^= Collocated Value
The number of collocated values within the IEMP data base was
extensive and required a considerable amount to prepared. As a
result, the presentation of %RSD is still under preparation and
will be reported in a subsequent report.
4.4.2 Consistency of Parallel Data Sets of Related Parameters
The examination of the %RSD presented a first look at
precision of the equipment monitoring methods and alerted reviewers
of the IEMP data to possible errors in the sample methods used
during the program. The consistency of parallel data sets was also
examined by reviewing two data bases that were sampling different
parameters but where the parameters were related to each other. The
example of this type of review of the IEMP data is the comparison
of 2.5 um particulate data with data collected by PM-10. In Section
4.3 it had been suggested that the 2.5 um particulate data did not
correctly represent that size range of particulates because of the
XRF analyses indicating the presence of elements attributed to
28
-------
Table 4-4. Comparison of IEMP VOC Data with VOC Presented
in the Literature.
Range of
VOC Concentrations3
EPA
Compound
Summer IEMP
Winter IEMP
Literature
(PPb)
(PPb)
(PPb)
n-Octane
ND-3
ND
0.0-13*
n-Nonane
0.2-3
ND
0.06-14*
n-Decane
0.2-17
ND
0.01-14
n-Undecane
0.1-30
ND-2
0.1-10
Chloroform
ND-0.7
ND-3
0.07-0.4
Dichloromethane0
ND-7
0.03-1.9
1,2-Dichloroethane
(2-12)
ND
0.1-0.5
1,1,l-Trichloroethane
0.5-13
ND-2
0.4-1.1
1,1,2,2-Tetrachloroethane
ND-2
ND-1
0.002-0.009
Carbon tetrachloride
0.1-0.4
ND
0.16-0.18
Dichlorodifluoromethane
ND-4
ND-11
0.6-1.5
Trichlorofluoromethane
3-17
ND-3
0.4-0.8
Vinyl chloridee
ND-2 3
ND
0.0-79*
Vinylidene chloride
4-39
ND
0.0-0.14
Trichloroethene
ND-39
ND
NOT AVAIL.
Tetrachloroethene
ND-1
ND-7
NOT AVAIL.
2-Chloro-l,2-butadiene
ND-15
NOT AVAIL.
Benzene^
(2-12)
ND-26
0.3-14
Toluene
3-22
ND-78
0.71-37
Ethylbenzene
o-Xylene.
m-Xylene^"
0.6-5
ND-3
ND-3
0.6-3.8
0.3-30
(2-15)
ND-58
0.6-22.0
p-Xylene^
(2-15)
ND-58
0.6-22.0
Styrene
2-26
ND-24
1.1-1.9*
4-Ethyltoluene
0.5-14
ND-19
0.4-1.5
Chlorobenzene
ND-4 2
ND
NOT AVAIL.
a. Range of data presented is with outliers removed. ND=Not Detected
b. As presented in Volatile Organic Chemicals in the Atmosphere: An
Assessment of Available Data. EPA (1983). * = Not specific to Denver
area. Not Avail.= Range not available from literature for indicated
compound.
c. Data collected during the summer was removed from data base.
d. Parenthesis around range indicates coeluting compound with Benzene
when GC/FID/ECD analytical method used.
e. Historical data range was listed a questionable.
f. Compound not measured during winter.
g. Parenthesis around range indicates coeluting compound with
Dichloromethane when GC/FID/ECD analytical method used.
h. Compound not measured during summer.
i. Parenthesis around range indicates coeluting compound with p-Xylene
29
-------
coarse size particulate. This concern was addressed in this section
of the report by comparing PM-10 data to 2.5 um particulate data
collected at the same time and location. Comparing each individual
concentration of PM-10 data to 2.5 um concentration collected at
the same time would provide an assessment of the magnitude that the
2.5 um sampler allowed larger sized particulates to be collected.
Table 4-5 provides a comparison of PM-10 data with 2.5um data
collected at the two sites Table 4-5. 2.5 vs. PM-10
where both parameters were Comparisons
being collected. Data
collected during the winter
sampling period at Arvada and
Auraria has been averaged for
each month and for the entire
sampling period. Data
collected during the winter
period was chosen since only
one type of particle size
limiting device was used on
the 2.5 um sampler during the
winter period.
The comparison indicates
that on an overall average the
2.5 um sampler collected 84%
of the particulates collected
by the PM-10 sampler. However,
on several individual days the
2.5 um sampler collected the
same or more particulate mass
than the PM-10 sampler. This
indicates that the particle
size cut point of the sampler
may be different than its
design cut point of 2.5 um and may explain why the XRF analyses
indicated a coarse particle size composition on the 2.5 particulate
sampling media. A more complete discussion of the examination of
the 2.5 um particulate data appears in Appendix B. This review of
the 2.5 um data indicated that the mass concentration has bias.
Therefore, it has not been within the context of this report.
However, XRF analyses of the 2.5 um particulate data for selected
elements are summarized in this report. These elements were
believed to be unaffected by large size (>2.5 um) particulates
contaminating the data. This was due to the range of
concentrations being with expected values.
The comparison of related particulate data provided a unique
opportunity to validate the data. Similar comparison with other
IEMP data sets were not possible given the types of parameters
measured. However, the comparison of occurrences of maximums and
minimums within the various data sets was possible. These
comparisons would provide an indication of specific environmental
Month/ 2.5/PM-10
Ave.Period Ratio
Auraria Arvada
NOV. '87
.73
.90
Max.
.96
1.35
Min.
.52
. 68
Dec. '87
.77
. 84
Max.
1. 11
.95
Min.
.63
.57
Jan. '88
.85
.74
Max.
1.04
.88
Min.
.31
. 51
Feb. '88
.95
. 94
Max.
1.44
1.11
Min.
.28
.88
Avg.Auraria
.86
Avg.Arvada
.83
Avg. Al1
CO
30
-------
conditions that affected all ambient sampling. The review of
similarities within the data sets are addressed in Section 4.0 of
this report.
4.5 Future Data Validation
Additional comparisons of related parallel data sets are
planned for the IEMP air toxics data in the future. The 1987/88
Metro Denver Brown Cloud Study also conducted sampling during the
same winter sampling period as the IEMP study. The study placed
sampling equipment at IEMP's Auraria monitoring station. Some of
the sampling conducted by the study involved the measurement of
particulates, and nitrates and sulfates. Comparison of this data
with similar IEMP data would be beneficial.
Validation of the laboratory analyses has not been addressed
in this report. An extensive amount of quality control of the
analyses conducted in the has been performed. Duplicate and spiked
samples were periodically submitted to the laboratory for analysis.
Replicate analysis of the samples was conducted by the laboratory
as part of its own quality control. In addition field blanks and
laboratory blanks were submitted and duplicate analyses of selected
samples by a second laboratory were conducted during the course of
the monitoring program. These quality control measures are being
reviewed and the results will be reported in the future.
At this point in the data validation process, the data which
remained was considered to be of sufficient quality in order for
the interpretation of data to begin. The preliminary interpretation
of the data appears in the remaining sections of this report.
31
-------
5.0 BiimmwHon of Data Results
Interpretation of the results involved the review of the
various parameters collected during each IEMP monitoring period for
selected averaging periods. These averages provided a temporal
representation of air toxic concentrations in the Denver
metropolitan area. Average concentrations were developed for the
time periods of lhr, 12hr, and 24hr. Maximum and minimum
concentrations were also developed for selected time periods. The
concentrations were presented graphical over the summer and winter
monitoring periods. Patterns in the concentrations would then be
observed within these graphs, and relationships between the various
parameters would be established.
It was anticipated that the patterns seen in the data would
relate to the occurrence of specific ambient air conditions and
hence establish the effect these conditions would have on the
quality of metropolitan Denver's ambient air. Depending on the
orientation of sources of air toxics, it was also anticipated that
these patterns in the averages, maximum, and minimum concentrations
would highlight areas in the metropolitan area that would be
subject to high concentrations of a particular air toxic or toxics
on a consistent bases.
The ultimate goal of the interpretation of the data results
was to develop a summary of the data that could be used to develop
exposure and risk assessments of the pollutants measured during the
IEMP air toxics monitoring program. These assessments were
developed and appear in subsequent sections of this report.
5.1 Carbon Monoxide and Particulate Data
It was anticipated that by first reviewing criteria
pollutants, i.e. Carbon Monoxide (CO) and PM-10 um (< 10 um in
size) particulate data, collected by the IEMP air toxics monitoring
program, averages, maximums and minimums developed for these
pollutant concentrations would provide an indication of the general
pattern of concentrations that should be expected for all of the
air toxic pollutants measured during the program. These data sets
would establish a pattern of days or events when concentrations
were high and, therefore, provide the first indication of days in
which the air toxics concentrations measured may also be high.
Subsequent comparison of the days with high toxic concentrations
with meteorological conditions present at the time of the high
concentrations would alert reviewers of the data to atmospheric
conditions under which the pollutants would tend to be higher and
pose more of a health risk.
5.1.1 Carbon Monoxide Data
32
-------
CO was measured at the Auraria and Palmer monitoring stations.
Appendix A presents a tabulation of hourly averaged values of the
CO values measured during the summer and winter at these locations.
During the summer period, CO concentrations were measured only at
the Auraria station. Data collected between July and September
indicated that CO concentrations were small. The monthly maximum
for July through September were 5, 6, and 8 parts per million by
volume (ppm). The small amounts of CO measured during the summer
period made any observation of a pattern to the concentrations
difficult. In general a small increase in hourly values is seen
between the hours of 5am and 8am and again between the hours of 7pm
and 11pm.
A pattern of higher concentrations of CO values is established
during the winter period. During November 1987, a maximum hourly
concentration of 32 ppm, 3 ppm below the Colorado standard of 3 5
ppm, occurred November 20 at 7 pm. This maximum concentration
marked a particular high pattern of CO concentrations that occurred
during the period 19th through the 21th of November. On these
three days, the daily mean was 8, 10, and 9 ppm respectively. For
the remaining days of that month the daily mean average
approximately one half of these values. The exception was the
period from November 10 through the 12 when CO hourly values were
also high during the evening period. The highest value occurred
at 10 pm on the 10th when a value of 29 ppm was measured. High
concentrations were also seen in the morning hours during these
periods. A high concentration of 23 ppm was measured at 2 am on
the 21st of November. Most of the higher ppm values recorded
during these periods occurred from lam through 9am.
The same pattern of CO concentrations also occurred at the
Palmer monitoring station but the magnitude of concentrations was
less than at the Auraria monitoring station. The highest value
measured for the month of November was 15 ppm at 6 pm on the 2 0th
of the month. This is the same day and within an hour (7 pm) of
the occurrence of the monthly maximum at the Auraria station. The
same pattern of high concentrations of CO that occurred at Auraria
was seen at the Palmer monitoring station. Although the individual
hourly values of CO measured at Palmer were less than the same hour
CO value measured at the Auraria monitoring station.
The month of December experienced similar values of CO as
measured by the two sites. The maximum concentration at Auraria
was 21 ppm and occurred at 6 pm on 17 December and a maximum of 16
ppm was recorded at 5 pm on 17 December at Palmer. Note that, like
the month of November, the maximum value again occurred on the same
day and within a hour at each site. It is interesting to note that
Palmer's maximum occurred the hour before the Auraria site measured
its maximum. This suggests not only a relationship between the two
stations but also a lag in CO concentrations measured between the
two stations. December CO values at the two stations were similar
to values recorded during November in the fact that higher CO
33
-------
AURARIA CO DATA C 8hr AVE. }
11/01/ 87 ThRU 02/29/BB
NDV 'B7
DCC,
JAN. '88
FEB '68
8nr AVE CO DMA
MONITORING PERIOD
snr CO 5TAN0.-SFJM
Figure 5-1.
8hr Running Averages of CO Measured at the Auraria
Monitoring Station during the Winter Period.
Palmer 8hr . CO Values
ndv trru Foo e?/ee
den.
wt>riltor|ng p«rioa
8 nr Ave CO vb i ues
00 5tor>a - 9 pptn
Figure 5-2
8hr Running Averages for CO Data Collected at the
Palmer Station during the Winter Monitoring
Period.
34
-------
values occurred during the evening and morning hours. High CO
concentrations occurred less frequently in the morning hours than
occurred in November. Both stations experienced high CO
concentrations during the period of 17 through 18 December.
However the Auraria station indicated higher values.
Despite the December 3 incident when the Auraria and not the
Palmer station monitored high CO concentrations, the pattern of
coincident high values of CO at both stations continued for the
entire winter period. However the Palmer monitoring station
continued to show lower values of CO than Auraria. A comparison
of the Auraria and Palmer data bases can be performed by comparing
the graphical representation of the data in Figures 5-1 and 5-2
respectively. These figures depict the eight hour running average
concentration of CO concentrations in ppm determine from hourly
values measured at the Auraria and Palmer station. It can be seen
from the figure that concentrations of CO are generally higher in
November and December and decrease in January and February.
CO data collected by the State of Colorado at monitoring
stations near the Auraria and Palmer stations indicated a similar
pattern of concentrations in November and December. Data collected
during January and February was not available at the time of
review. The general pattern of high concentrations of CO on
selected days in November and December followed by a general
decrease in CO concentrations in January and February was noted and
sought for in the remaining data bases. If the pattern of
concentrations was similar in other data bases, a relationship
could be developed between CO and the toxic concentrations.
Similar patterns in the air toxic concentrations would suggest that
the occurrence of high concentrations are related to the
meteorological conditions. Any patterns in the data will be noted
in subsequent sections of this report.
5.1.2 PM-10 Particulate Data
PM-10 data was collected during the summer and winter
monitoring periods at the Auraria and Arvada monitoring stations.
During the course of the summer monitoring period no violation of
the PM-10 standard was observed (150 ug/m ). Most concentrations
observed at the two stations were in the 25-35 ug/m ranae. The
average concentration at the Auraria station was 31 ug/m and 2 7
ug/m at the Arvada monitoring location. Figure 5-3 and 5-4 depict
the summer PM-10 concentrations measured at the two monitoring
stations during the period.
Higher concentrations of PM-10 were seen during the winter
months. The Auraria station measured the highest concentration of
117 ug/m on December 16. The second and third highest
concentrations were found on November 20 and December 17.
Concentrations measured on these two days were 115 ug/m3 and 112
ug/m respectively. The highest PM-10 concentrations occurred on
35
-------
Auraria PM-10
June tnru sept. "B7
i i y i—i i i
JULY '07
AUG
M3NIT0RING F^IOO
Figure 5-3. summer PM-10 Concentrations at Auraria.
R
E
Arvada PM-10
June trru Sept. '67
MDNITORING PS^lOD
Figure 5-4. summer PM-10 Concentrations measured at Arvada.
36
-------
the same dates as the occurrence of elevated levels of CO noted
earlier in this section of the report. The average concentrations
for the two sites for the entire monitoring period were 54 ug/m
for the Auraria station and 38 ug/m for the Arvada station.
Concentrations of PM-10 measured at the Arvada station showed
elevated concentrations of particulates on the same dates as the
Auraria station. The highest concentration that occurred at the
Arvada station was 77 ug/m on December 16. Although in general
Arvada PM-10 concentrations were smaller in magnitude than at
Auraria, the Arvada PM-10 concentrations show an increase in
concentrations on the same days as Auraria. Figures 5-5 and 5-6
depict the concentrations at the two stations with time for the
winter period.
5.2 2.5 urn Particulate Data
Fine size particulate data (< 2.5 um in size) were collected
at all of the monitoring stations during both monitoring periods.
In addition to a determination of the concentrations of
particulates in this size range, laboratory analyses were perform
to determine the elementary composition using X-Ray Flourescence
(XRF) and the carbon content of the particulate matter. However,
examination of these laboratory analyses revealed discrepancies in
the analyses when compared to previous studies. Section 5.0 has
describe some of the discrepancies involving this data and a series
of discussions are presented in correspondence in Appendix B
regarding the concerns with the data.
In reviewing these discrepancies, personnel reviewing the data
determine that any reported 2.5 um concentrations would be biased
by the larger than 2.5 um size particulates that were found on the
sampling media. Therefore no discussion of 2.5 um particulate mass
is presented in this report.
The large size particles also biased the organic and elemental
carbon, and XRF analyses that were performed exclusively on the
2.5 um particulate data. Therefore no carbon data is summarized
within this report. However in reviewing the XRF data, some of the
elemental data appears to be unaffected by the contamination of
large particulates on the filters. This opinion was formulated
after the comparison of element concentrations reported by the XRF
analyses associated with soils (i.e. Calcium, Potassium, Iron,
etc.) compared to elements associated with trace metals ( i.e.
Cadium, Lead, Arsenic, etc.). The reported trace metals
concentrations appeared more consistent with concentrations
reported in other studies. This observation may be due to trace
metals found in particulate matter being associated, the majority
of the time, with the finer size particulate fraction. More review
of the data than has taken place for the purposes of this report
will need to be accomplished in order to confirm this observation.
It should also be noted that any conclusions reached by the
37
-------
interpretation of the XRF data regardless of elements under
consideration should take into account the large amount of
uncertainty that has been attributed to the XRF data by the
laboratory performing the XRF analyses. The reader is encouraged
to read the project report for IEMP (PEI, 1987a & 1988) . This
report provides a listing of the uncertainty to the XRF data.
In order to summarize the XRF data effectively, a monthly
tabulation of the data has been prepared and is summarized in
Figures 5-7, 5-8 and 5-9. The figures represent the average monthly
concentrations measured during the winter for Cadmium, Chromium and
Lead. These elements were selected for presentation in this report
because of the health risk associated with these elements and the
magnitude of the concentrations found in the Denver area.
38
-------
Auraria PM-1D Data
H^vwrtw thru February
Q
E
*
u
I
i
hCN I TOR ING PERIOO
Figure 5-5. Winter PM-10 Concentrations Measured at Auraria
Arvacfa PM-10 Data
NovenMr tfiru F«bri*ry
Q
E
S
\J
I
I
hCNI TOPING PB5I00
Figure 5-6. Winter PM-10 Concentrations measured at Arvada.
39
-------
In Figure 5-7, average Cadmium concentrations are depicted for
each month. The number above each bar representing the average is
the number of samples that were used to calculate the average. All
averages were less than or equal to 0.02 ug/m . The monitoring
stations had similar average concentrations in November and
December. However in January and February, Cadmium concentrations
sampled at Auraria during both AM and PM monitoring periods were
high than the average concentrations at the other stations.
EMP X - RAY FLUORESCENCE CADMIUM DATA
D.026
korrrt-T avbuge kw each site
[\T] A UP AM
MONITORING reilOO '87/80
V77X mr. ph *ih
Figure 5-7. Monthly Average cadmium Concentrations for Winter.
Chromium concentrations depicted in Figure 5-8 indicate that
lower average concentrations occurred in November and December.
An increase in concentrations occurred in January and February.
However averages did not exceed 0.02 ug/m . The highest averages
were seen at the Auraria monitoring station.
Lead concentrations were average for all stations and are
depicted in Figure 5-9. The highest concentrations occurred at
Auraria during both AM and PM monitoring periods and at the NJH
monitoring station. The months of November and December had the
highest average concentrations. A small decrease in the averages
are seen in both January and February.
The remaining elements are not summarized within this report.
Concentrations developed from the XRF analyses are contained within
the Project Report for the Denver IEMP Air Toxic Monitoring program
(PEI, 1987a & 1988).
40
-------
IEMP X - RAY FLUORESCENCE CHROMIUM DATA
0.026
o.ow -
D.022 -
0 02 -
D.016 -
2
§ 0.016 -
5 0 0-M -
h
i D.012
^ Q.01 -
0.0D6 -
D 006 -
D.OW -
D 002 -
0
MONTH. T AVBlAGE FOB EACH SITE
8
r^3 AUR. am
MONITORING PCTIOO '07/88
X777X AUR. F* ^33 WH
Figure 5-8. Monthly Average Chromium Concentrations for Winter,
IEMP X - RAY FLUORE5CENCE LEAD DATA
fcCNTHY AVERAGE FOR EACH SITE
11
0
010
*
P\\] WJR AM
WON I TOR IMS TO I CD '07/BB
V7A HX. PM ^ *JH
Figure 5-9. Monthly Average Lead Concentrations for Winter.
41
-------
5.3 Aldehyde Data
Aldehyde data were collected during the summer and winter
monitoring sites at three of the four monitoring locations. Arvada,
Auraria, and NJH monitoring stations each sampled for 14 aldehyde
compounds. Arvada and NJH stations conducted 24 hour sampling and
the Auraria station sampled on a modified sampling schedule to
obtain an AM or PM sample. This was done to determine if aldehyde
concentrations were significantly different during different parts
of the day. During the summer, AM and PM samples were 12 hr
periods and began at 7am and 7pm, respectively. The schedule was
modified during the winter sampling period. AM sampling at Auraria
was conducted between 9am and 4pm (7 hours) and PM sampling was
conducted between 4pm and 7am (17 hours). Laboratory results of
the Aldehyde sampling revealed that only four compounds were
consistently measured throughout the two sampling periods at all
three sites. These compounds were Formaldehyde, Acetone,
Acetaldehyde, and Propionaldehyde. Further examination of the
Acetone concentrations indicated high field blank and laboratory
blank concentrations. High concentrations for these blank samples,
used as a quality control measure, alerted reviewers of the data
to exercise some caution in reviewing the Acetone results. It was
determined that the reported acetone levels from the laboratory
analyses were erroneous and are not presented in this report. A
summary of the contamination is presented in correspondence
contained in Appendix B.
Results from the aldehyde sampling are summarized in a series
of graphs depicted in the following figures. The quantity of data
prevented a convenient method of depicting individual
concentrations of these aldehydes. As an alternative, monthly
averages have been developed for each compound and compared for
each site. Each aldehyde compound has been depicted on a series
of figures. Figure 5-10 depicts Formaldehyde concentrations
measured during the summer monitoring period at all three
monitoring locations where aldehyde data was collected. The graph
in the figure is a bar graph that indicates the monthly average
concentration at all three sites. The number of samples that were
used in the average appear at the top of each concentration. For
the purpose of the data summary, Auraria AM and PM samples have
been combined for a 24hr value.
The graph in Figure 5-10 indicate that the highest
concentrations in Formaldehyde were found at the Auraria monitoring
station. Average monthly concentrations increase each month at
Auraria until August. The remaining stations showed on average
smaller concentrations but a similar pattern to the concentrations
July through September. Acetaldehyde and Propionaldehydes
concentrations are depicted in Figures 5-11 and 5-12, respectively,
for these two compounds very little change in the average monthly
concentration can be seen from month to month.
42
-------
ARVADA, AURARIA, & NJH FORMALDEHYDE
MONTHLY AVEP*GE FOR EACH SITE
3 -
JU^
AUGUST
ARV. 24HR
MONITORING PBPIOD 1967
r\Xl AURA.- AW X777X AURA - PM
SEPT
NJH 24HR
Figure 5-10 Monthly Average Formaldehyde Concentrations in
Summer.
During the winter sampling period a small increase in the
monthly average can be seen in the Formaldehyde levels as depicted
in Figure 5-13. In the summer, averages during the four month
sampling period ranged from 3 to 5 ppb. The range of averages were
4 to 6 ppb during the winter period. Winter Acetone
concentrations, shown in Figure 5-14, averaged approximately 4 to
8 ppb during November and December. In January Acetone averages
increased for the Arvada monitoring station and for Acetone levels
measured in the AM period at Auraria. Arvada average increase from
5 ppb in December to 8 ppb in January. Auraria AM levels increased
from 7 ppb in December to 13 ppb in January. Auraria PM and NJH
Acetone levels showed little change between the months. Very low
average Acetone levels were found in February.
43
-------
ARVADA, AURARIA, & NJH ACETALDEHYDE
MOUTH. Y AVERAGE F0« EACH SITE 3
10 10
ARV
MONITOR INS PER IPC 1907
fV\] AURA.- AM f/y/^ AUKA. - M
Figure 5-11. Monthly Average Acetaldehyde Concentrations in
Summer.
ARVADA, AURARIA, & NJH PROP IONALDEHYDE
M0WTH.Y AVERAGE FOG EACH SITE
0.3 -
0 2
MONITOR I hC PER 100 1957
AUHA - AM V77X AURA - t*A
K353 HJH
Figure 5-12 Monthly Average Propionaldehyde Concentrations in
Summer.
44
-------
ARVADA, AURARIA, AND NJH FORMALDEHYDE
komtt AVWAGE FDR EACH SITE
lil
s!l
VAm
A^|
^11
A||
^11
7 4
I
II
ii
llll
2b:
m
NHAJ> 2-*«
hONITORINC PB^lOO '87/88
IVv] AU^R IA-AW E553 AU»*fllA-F* IUH
Figure 5-13. Monthly Average Formaldehyde Concentrations for
Winter.
In Figures 5-15 and 5-16 Acetaldehyde and Propionaldehyde
concentrations for the winter are summarized. Acetaldehyde monthly
average levels ranged from 4 to 7.5 ppb during November and
December. Monthly averages decreased during January and February.
The range of average concentrations was from 0 to 3 ppb.
5.4 Denuder Data
The Annular Denuder sampling equipment used during the summer
and winter sampling periods provided for the measurement of
inorganic compounds. Compounds of interest involved Nitrous and
Nitric Acid in gaseous form and nitrate and sulfate particulates.
The method in which the denuder sampled for these compounds is
presented in the Quality Assurance Project Plan (QAPP) (PEI,
1987b).
Nitrous and Nitric acid for the summer and winter periods are
depicted in Figures 5-17 through 5-21. During the summer these
compounds were measured only at the Auraria monitoring station.
At selected time periods during the winter monitoring period
Denuder sampling occurred at the Arvada and the Federal Court House
building. However the sampling occurred for a limited time period
and for the purposes of this report is not summarized. The
sampling frequency also increased at the Auraria monitoring
station. A one in three day sampling schedule was increase to an
45
-------
ARVADA, AURARIA, AND NJH ACETONE
kCNTX_Y AVERAGE FOR EACH SITE
15
NOV. '87 DEC. '87 JAN. '80 FEB. "88
MONITORING Pg IQD '87/88
ARVADA fV\l AURARIA-AM Y777X AURARIA-F* NJH 24HR
Figure 5-14. Monthly Average Acetone Concentrations in Winter.
ARVADA, AURARI A, AND NJH ACETALDEHYDE
MOKTKLY AVERAGE FOR EACH SITE
ARVADA
MONITORING PB3IOO '87/88
rO\3 AURAR I A-AM T777X AURARIA-W
E3353 hjh
Figure 5-15. Monthly average Acetaldehyde Concentrations in
Winter.
46
-------
ARVADAJ AURARI A, & NJH PROP IONALDEHYDE
MONTHLY AVtFfeGE FOfi EACH 6ITE
aRvaD*
87 OK. *0?
UDNITOGING PBRIOO g7/BB
fVXl U77X AUPKRlA-F^
10
VS 10 6
£553 njh
Figure 5-16. Monthly Average Proionaldehyde Concentrations in
Winter.
everyday sampling schedule approximately the beginning of December
until the end of January. This was done because of the interest
in developing a larger data base for comparison with other studies
having similar data bases.
It was anticipated that the dispersion conditions associated
with summer time meteorological conditions would allow the Nitrous
and Nitric Acid concentrations to disperse uniformly through the
atmosphere particularly during daytime hours. This would allow for
an assessment of what typical Nitric and Nitrous Acid levels would
be for a large urban area. These levels could then be compared to
wintertime conditions when poor atmospheric dispersion conditions
would allow for concentrations of these compounds to increase. It
should also be noted that Nitrous Acid is considered a nighttime
phenomena since it dissociates in the presence of sunshine. Winter
concentrations of these compounds were compared to CO and
Particulate concentrations sampled at the same time in order to
assess whether similar atmospheric conditions that cause elevated
levels of the more commonly measured pollutants such as CO and
particulates also result in elevated levels of Nitrous and Nitric
Acid.
In Figure 5-17 Nitrous Acid levels are depicted for the summer
sampling period. Both AM and PM levels are shown in the figure.
With the exception of one data point in August, PM Nitrous Acid
levels are higher than AM Nitrous Acid levels due to the compounds
47
-------
AURARIA DENUDER DATA
SUMO GASEOUS NITROUS ACID LEVELS
MONITORING PER ICO
~ I-N02 AM CONC. + I+C2 PM CCNC.
Figure 5-17. Slimmer AM AND PM Nitrous Acid Levels in the Denver
Metropolitan Area.
disassociation in sunlight. The exception in the data may be an
artifact of the laboratory analysis and it is undergoing further
review. The data point will not become part of the final data base
until quality control of the data have been satisfied.
Concentrations of Nitric acid during the entire sampling
period were less than 2 ppb. Summer Nitric acid levels show in
Figure 5-18, as expected, the reverse trend of AM levels being
higher than PM levels when compared to the Nitrous Acid levels.
AM concentrations were 3 ppb or less during the summer period but
PM concentrations were less than 1 ppb.
Winter Nitrous Acid levels were higher than summer
concentrations and are due to the longer nighttime hours and
atmospheric dispersion conditions associated with winter
conditions. PM levels are depicted in Figure 5-19. This figure
shows concentrations in the 4 to 6 ppb range during the period.
It is interesting to note that in November and December and to a
lesser extent in January and February Nitrous levels were elevated
at the same time as CO and Particulate concentrations were
elevated. This would indicate poor dispersion conditions result
in higher concentrations of toxic pollutants as well as the more
typically measured pollutants in the Denver area. During poor
48
-------
AURARIA DENUDER DATA
SU»*ei=! GASEOUS NITPIC ACID LEVB.S
MONITORING PER I CD
~ hN33 AM 03NC + HN03 fV CONC.
Figure 5-18. Winter AM and PM Nitric Acid levels in the
Metropolitan Denver Area.
dispersion conditions in November and particularly in December
Nitrous Acid levels reached a maximum of 8.5 ppb.
AM levels of Nitrous Acid, not depicted in the text, indicated
that elevated levels of the compound occurred during poor
dispersion conditions. AM levels were approximately one half of
the levels measured during the PM sampling period. A maximum of
4 ppb was reached on December 18, 1987. This was the same day that
PM levels reached 8.5 ppb.
Figure 5-20 represents Nitric Acid levels during AM and PM
sampling times in the winter. In general the pattern of
concentrations did not vary above 1 ppb during approximately one
half of the entire monitoring period. Nitric Acid concentrations
for AM measurements were higher than PM measurements. Elevated
concentrations occurred during mid December and early January. The
maximum concentration of approximately 5.5 ppb occurred in mid
December. PM levels of Nitric Acid, not depicted in the text, were
generally less than 1 ppb and did not show the elevated levels as
did AM concentrations. The maximum concentration during PM
sampling conditions was 0.4 ppb.
Concurrent nitrate and sulfate particulate concentrations
were also measured using the same Denuder sampling device at the
49
-------
AURARIA DENUDER DATA
WINTBR FV GASEOUS NITFDU6 ACID COMC.
MONITORING F^RIOO '67/BB
~ I-H02 FM CONC
Figure 5-19. Winter PM Nitrous Acid Levels Measured in the
Metropolitan Denver Area.
same time as the Acid levels described in the previous paragraphs
were measured. Through the use of Teflon and Nylon filters the
particulates contained in the same airstream as the gaseous Acid
concentrations were captured and analyzed. Summer concentrations
of nitrate and sulfate were small and are not depicted in the text.
Nitrate concentrations measured at Auraria were in the range from
0.5 iig/m to 1.5 ug/m . Sulfate concentrations varied from 1.5
ug/m to 3.0 ug/m .
50
-------
AURARIA DENUDER DATA
WlNTER AM GASEOUS NITRIC ACID CONC.
NOV Dec
MONITORING PS?tOO "87/88
O HN03 H4 CONC.
Figure 5-20. Winter AH Nitric Acid Levels Measured in the
Metropolitan Denver Area.
AURARIA DENUDER DATA
WINTER NITRATE FW7T I CU>TE CONC.
M ml I f
NOV. D6C
MONITORING PCTiOO "B?/ 88
~ NITRATE 24hB, CONC,
Figure 5-21. Winter Nitrate Concentrations Measured at the
Auraria Monitoring Station.
51
-------
Figures 5-21 and 5-22 provide concentrations of nitrate and
sulfate concentrations collected during the winter period. In
Figure 5-21 nitrate particulate concentrations collected at the
Auraria monitoring station during a 24 hour period are displayed
in ug/m . Concentrations during the majority of the monitorina
period averaged less than 5 ug/m . In mid December a 40 ug/m
concentration was observed at Auraria. In early January a 34 ug/m
concentration was measured. These concentrations marked the
highest and second highest concentration observed at the site.
Several smaller elevated concentrations were observed in January
and February but concentrations during these events and -for the
remainder of the monitoring period never exceeded 15 ug/m .
AURARIA DENUDER DATA
VINTER SULFATE PARTICULATE COUC.
MONITORING PERIOO '67/88
O SULFATE 24HR. CONC.
Figure 5-22. Winter Sulfate Concentrations Measured at the
Auraria Monitoring station.
Sulfate concentrations depicted in Figure 5-22 for the same
period do not show the same pattern in their concentrations with
time as was depicted for the nitrate concentrations. A maximum
sulfate concentration of 9 ug/nCwas observe the first of February.
A secondary maximum of 8 ug/m occurred in early January. The
range of sulfate concentrations were between 0 and 4 ug/m3.
However, sulfate concentrations during the winter did not follow
similar patterns in concentrations as some of the other IEMP air
monitored pollutants had followed. The nitrate and sulfate data
52
-------
base for the winter period is currently being compared to nitrate
and sulfate data collected during the same period by the
Metropolitan Denver Brown Cloud Study.
5.5 Volatile Organic Compounds
Twenty six unique Volatile Organic Compounds (VOC) were
monitored and analyzed during the summer and winter period.
Section 3.0 of this report has address some of the concerns raised
about the VOC measurements after a preliminary review of the VOC
data. During the summer period the contract laboratory had used
a gas chromatograph (GC) system. During the winter sampling
period, air evacuated from the canisters was analyzed using Mass
Spectrometry. Most of the 26 Organic Compounds were measured in
varying concentrations during the summer period. However, only six
compounds were measured using the Mass Spectrometry system during
the winter. It is unclear at this time whether the lack of
detectable levels on many of the compounds that were attempted to
be measured were the result of atmospheric conditions or the lack
of sensitivity of the MS system. The data is still under review
in the anticipation of resolving these concerns. A summary of the
data is provided in the following paragraphs.
The large data base of summer VOC concentrations prevented a
concise graphical representation of the data. As an alternative,
a summary of the maximum, mean, and minimum concentration for each
compound has been provided in Table 5-1. The reader is encourage
to review the data for his1 or her's own interpretation.
VOC compounds collected during the winter were a smaller data
base due to the nondectability of many of the compounds.
53
-------
Table 5-1.
Maximum, Mean, and Minimum Summer VOC Concentrations in PPB
(Part 1)
COMPOUND
ARVADA
JUNE JULY
AUGUST
SEPT
N-OCTANE
MAXIMUM
MEAN
MINIMUM
CHLOROBENZENE
MAXIMUM
MEAN
MINIMUM
ETHYLBENZENE
MAXIMUM
MEAN
MINIMUM
M-XYLENE/P-XYLENE
MAXIMUM
MEAN
MINIMUM
N-NONANE
MAXIMUM
MEAN
MINIMUM
STYRENE
MAXIMUM
MEAN
MINIMUM
VINYL CHLORIDE
MAXIMUM
MEAN
MINIMUM
TRICHLOROFLUOROMETHANE
MAXIMUM
MEAN
MINIMUM
0.68
0.5
0.34
41.96
12.11
0.59
2 . 04
1.56
1.11
5.88
5. 09
4 . 03
1.54
0.89
0. 52
6.91
5.56
4 .93
8.57
6.36
4 .31
6.89
5.58
4.31
1.31
0.76
0.42
3.04
1.03
0. 17
2 .19
1.34
0. 62
7 . 86
5. 07
2 . 63
0.98
0. 57
0.2
9.52
5.21
1.77
15.93
8.43
3.76
13 . 66
8.44
6.41
1. 89
0.97
0.54
5. 09
1.9
0.25
1. 92
1. 15
0. 61
6.5
3.88
1.98
1. 14
0. 88
0. 59
7. 19
3. 59
1. 54
7.95
6. 59
6.11
14. 31
7 . 47
4. 63
1. 04
0.85
0.67
0.31
0. 28
0.26
1. 84
1. 53
1.23
5. 44
4.89
4.33
1.77
1.51
1.24
6.25
4.5
2.75
8. 19
6.81
5.43
AURARIA
JUNE JULY
0. 91
0.61
0.27
6.3
2.11
0.4
2.03
1.5
1.15
6.71
5.41
4 .33
0.92
0.77
0.65
9.37
5.23
3.22
15.66
10.92
6.97
7
6. 19
4 .76
1.78
0.85
0.33
6. 57
1.87
0. 12
4. 66
1. 64
0.73
14 . 66
6.05
3 .23
2 . 16
0.73
0.27
15.55
4 .93
2.01
37 . 39
13.55
3.27
16.26
8.67
5.27
AUGUST
1. 63
0.8
0. 39
4.48
1.65
0.28
1.74
1. 17
0. 68
6.87
4.27
2.40
1.65
1.02
0. 59
4.89
3.35
I.54
9. 67
6. 39
0.86
II.65
6. 02
2 . 49
SEPT
0.66
0. 57
0.47
0.31
0.24
0.21
1.21
1.1
0.9
5.9
5.46
4.63
2.
2.
1.
3.
3,
2.
9
39
74
48
03
68
7. 61
6.78
5.7
-------
VINYLIDENE CHLORIDE
MAXIMUM 10.60 15.09
MEAN 7.4 3 9.00
MINIMUM 4.87 5.07
BENZENE/1,2-DICHLOROETHANE
MAXIMUM 4.29 8.08
MEAN 3.92 4.61
MINIMUM 3.48 3.16
TOULENE
MAXIMUM 6.29 10.99
MEAN 5.3 7.54
MINIMUM 4.62 4.51
4-ETHYLTOLUENE
MAXIMUM 5.86 3.42
MEAN 3.51 2.17
MINIMUM 2.1 0.8
N-DECANE
MAXIMUM 16.63 1.87
MEAN 6.43 1.3
MINIMUM 1.26 0.58
N-UNDECANE
MAXIMUM 23.33 2.47
MEAN 7.53 1.5
MINIMUM 1.66 0.21
DICHLORODIFLOUROMETHANE
MAXIMUM 3.58
MEAN 2.09
MINIMUM 1.29
CHLOROFORM
MAXIMUM
MEAN
MINIMUM
1.11 TRICHLOROETHANE
MAXIMUM 5.63
MEAN 1.98
MINIMUM 0.77
14.77
7.02
3 . 66
10. 31
4.72
1. 57
12.22
7 . 32
4.49
4 . 34
2 .75
1.75
7. 17
2 . 56
0.25
8. 99
2 . 53
0.18
3.95
1.88
0. 34
4 . 34
4.05
3.76
7 . 3
6.41
5.51
8.18
8.06
7.94
5.68
5. 18
4 . 68
13.88
11.95
10. 03
2.76
1.41
0. 07
1. 67
1.31
0.94
19. 26
12. 31
8 . 56
3.2
2.57
2 .05
7.94
5.84
4 .38
5.7
3 .48
2.32
9 .85
5.39
1.65
29.59
16.47
0.90
39.38
17.32
5.25
9. 65
4.89
2.63
21.67
8.71
4.83
4.78
2.29
0.6
2.24
1. 57
0.64
7.59
2.84
0.26
1.5
1.36
1.11
15.12
10.04
4.27
7.83
4.45
2. 11
11.87
7.01
4.41
13.56
4.06
1.58
14.24
4 . 27
1.37
14.92
3.80
0.28
3.02
1. 89
0. 94
7.04
6.55
6.29
6.95
5.9
4.61
8.95
7.68
5.79
6.49
5. 57
4.82
13.88
11. 66
10.06
8.34
5.97
3.52
1.11
1.06
1.00
1.79
1.27
0.89
8.24
5. 13
2.86
2.88
1.85
0. 87
1. 60
1.39
1. 08
-------
CARBON TETRACHLORIDE
MAXIMUM 0.52 0.
MEAN 0.39 0.
MINIMUM 0.29 0.
TRICHLOROETHENE
MAXIMUM 0.
MEAN 0.
MINIMUM 0.
TETRACHLOROETHENE
MAXIMUM 1.51 2
MEAN 0.9 1
MINIMUM 0.4 0.
1,1,2,2-TETRACHLOROETHANE
MAXIMUM 0.59 0.
MEAN 0.42 0.
MINIMUM 0.17 0.
2-CHLORO-l,3-BUTADIENE
MAXIMUM 7.71 3.
MEAN 3.38 2.
MINIMUM 1.54 1.
51
28
17
4
34
07
. 3
.2
73
96
35
07
59
36
19
0.39 0.15
0.27 0.14
0.16 0.13
0.77 0.59
0.34 0.33
0.05 0.08
1.08 1.62
0.82 1.24
0.51 0.86
I.46
0. 49
0.1
II.8 6.0
7.26 5.82
1.7 5.64
0.4 0.47
0.35 0.29
0.27 0.17
0.17 0.54
0.12 0.24
0.08 0.11
0.87 1.18
0.55 0.71
0.35 0.32
0.6
0.21
0. 08
4.42 4.7
2.63 2.1
0.91 0.61
0.28 0.16
0.18 0.16
0.12 0.16
0.21
0.11
0.06
1.32 0.58
0.71 0.5
0.43 0.34
1. 15
0.4
0. 13
11.3 7.19
6.14 5.02
2.7 1.92
-------
Table 5-1. Maximum, Mean, and Minimum Summer voc concentrations in PPB (Part 2)
NJH
COMPOUND¦ JUNE JULY AUGUST SEPT
N-OCTANE
MAXIMUM 0.98 3.04 1.34 0.97
MEAN 0.65 1.41 0.97 0.92
MINIMUM 0.48 0.61 0.75 0.87
CHLOROBENZENE
MAXIMUM 17.94 6.52 2.78 0.46
MEAN 6.81 2.61 1.2 0.42
MINIMUM 0.58 0.18 0.38 0.39
ETHYLBENZENE
MAXIMUM 2.37 3.27 2.49 1.69
MEAN 2.03 2.07 1.57 1.54
MINIMUM 1.77 0.75 1.0 1.38
M-XYLENE/P-XYLENE
MAXIMUM 8.07 11.73 8.53 7.58
MEAN 7.45 7.61 5.77 6.88
MINIMUM 6.59 3.13 2.9 6.18
N-NONANE
MAXIMUM 1.06 1.11 2.29 2.79
MEAN 0.91 0.74 1.36 2.16
MINIMUM 0.67 0.29 0.73 1.53
STYRENE
MAXIMUM 8.43 7.75 11.02 3.69
MEAN 6.71 5.03 6.06 3.41
MINIMUM 3.71 1.81 3.34 3.12
VINYL CHLORIDE
MAXIMUM 10.73 22.83 9.79
MEAN 8.77 13.05 8.24
MINIMUM 7.36 4.19 7.13
TRICHLOROFLUOROMETHANE
MAXIMUM 9.6 17.33 13.59 11.29
MEAN 7.78 11.61 8.94 9.90
MINIMUM 6.34 6.40 4.24 8.50
-------
VINYLIDENE CHLORIDE
MAXIMUM 8.48 14.01 11.81
MEAN 6.25 8.85 7.60
MINIMUM 4.54 5.70 5.05
BENZENE/1,2-DICHLOROETHANE
MAXIMUM 8.05 8.55 11.72
MEAN 5.3 5.8 7.81
MINIMUM 3.88 3.95 2.69
TOULENE
MAXIMUM 7.72 17.24 12.46
MEAN 7.25 10.41 9.36
MINIMUM 6.43 5.79 5.66
4-ETHYLTOLUENE
MAXIMUM 6.47 4.95 8.0
MEAN 4.74 3.01 4.05
MINIMUM 3.62 0.73 1.08
N-DECANE
MAXIMUM 6.76 1.75 7.76
MEAN 4.13 1.21 2.94
MINIMUM 1.86 0.61 1.27
N-UNDECANE
MAXIMUM 23.95 3.37 8.59
MEAN 10.65 2.18 2.97
MINIMUM 3.32 0.47 0.57
DICHLORODIFLOUROMETHANE
MAXIMUM 1.33 3.56
MEAN 1.30 2.1
MINIMUM 1.28 1.1
CHLOROFORM
MAXIMUM
MEAN
MINIMUM
1.11 TRICHLOROETHANE
MAXIMUM 1.78 3.30
MEAN 1.11 1.57
MINIMUM 0.80 0.68
6. 69
5. 67
4 . 64
11.06
10.04
9.02
10.73
10. 03
9.32
5.48
4 .98
4 . 48
9.98
5.90
1.82
10.11
6.90
3 . 69
1.04
1. 01
0.99
0.86
0. 77
0. 69
-------
CARBON TETRACHLORIDE
MAXIMUM 1.12 1.78 3.30
MEAN 0.94 1.11 1.57
MINIMUM 0.83 0.80 0.68
TRICHLOROETHENE
MAXIMUM 0.47 0.24
MEAN 0.32 0.14
MINIMUM 0.18 0.06
TETRACHLOROETHENE
MAXIMUM 0.72 1.47 1.29
MEAN 0.57 0.90 0.88
MINIMUM 0.31 0.52 0.71
1,1,2,2-TETRACHLOROETHANE
MAXIMUM 0.49 0.39
MEAN 0.32 0.24
MINIMUM 0.11 0.13
2—CHLORO—1,3-BUTADIENE
MAXIMUM 3.37 4.15 14.95
MEAN 2.59 3.06 9.97
MINIMUM 2.12 2.24 6.80
0.86
0.77
0.69
0.70
0. 58
0.46
12.38
10. 89
9 .40
-------
Only six compounds were measured to have significant concentrations
during the winter monitoring period. The six compounds were
Benzene, Ethylbenzene, o-Xylene, m/p-Xylene, Toulene, and 4-
Ethlytoulene. Concentrations measured during the period for these
six compounds are presented in Figures 5-23 through 5-28. in
reviewing the figures, four of the six compounds Benzene, Toulene,
4-Ethyltoluene, and m/p-Xylene were analyzed to have high
concentrations in November. The magnitude of these concentrations
gradually decrease with each sample taken until concentrations
stabilized at low levels. Concentrations remained at low levels
from approximately early December until the end of the monitoring
period. Conversely, EthylBenzene and o-Xylene showed the reverse
trend of low concentrations in November and December followed by
higher concentrations in January. The elevated January
concentrations were followed by decreasing concentrations with each
subsequent sample in February.
It is unclear as to the cause of these concentrations and the
reasoning for the reverse trend in concentrations when the
compounds are compared. Several areas are being explored for a
possible solution. These areas include:
o Emission inventories for these compounds, if
available, may reveal sources which release these
compounds at selected times of the year.
o The oxygenated fuels program was in effect during
part of the winter monitoring program. What effect
did the use of these fuels effect the measured VOCs?
o Elevated concentrations were measured at all three
VOC monitoring stations. Are these concentrations
representative of the entire area as the data
suggests or an artifact of the laboratory analyses?
It is this last area that will receive first priority. In
Section 4.0, quality control measures revealed incorrect data and
was attributed to an artifact of the sampling methods or laboratory
analyses. The elevated concentrations found to exist in November
and then decrease with each subsequent sample are similar to the
trends found in the VOCs measured during the early summer sampling
season. A subsequent review of laboratory methods and
determination as to whether the canisters were evacuated properly
may provide an answer to the concentration trends found in the
winter VOC data.
60
-------
WINTER VOC CONCENTRATIONS
B&JZENE CONC
MONITORING F^IOO 'B7/0B
~ ARVAC* + AUR^RIA
Figure 5-23. Winter Benzene Concentrations
2 G
Z.4
2 2
2
1.8
1.6 -
14-
12-
1 -
0 8-
0.6 -
0.4 -
0.2
0
WINTER VOC CONCENTRATIONS
ETKTLBBCEKC 24HR CONC.
O ~
•+ 0
I I I
-------
WINTER VOC CONCENTRATIONS
TOLUENE CONC
MONITORING PERlOO '07/88
D ARVAE* + AUHARlA
O NJN
Figure 5-25. Winter Toulene Concentrations.
WINTER VOC CONCENTRATIONS
5 ¦
4- ETHYl_T"OLUBvE 24MR CONC .
MON ITORING FB^lOO 67/60
~ ARVAO* + MJRAR1A
Figure 5-2 6. Winter 4-Ethyltoulene Concentrations.
62
-------
WINTER VOC CONCENTRATIONS
3.2
3 -
2.0 -
2,6-
2 .4 -
2.2 -
2 -
1 .8 -
1 6 -•
1 .4 -
1 2 -
1 -
0.8 -
II •
0.6 -
0 4-
0 2-
O-XYLENE 24hF CONC,
D O
~ +¦
MDN I TOR I MG 100 B?/ 80
~ AF7VAI* + MJRAR I *
Figure 5-27. Winter o-Xylene Concentrations.
60
WINTER VOC CONCENTRATIONS
M/P-XYLE« 24HR CONC.
50
0 T t f l I
l + B»+0^ + P +
TT i > i T
+ 11
i I I I I i i—l i i T I 1 I I I I—i i i i—l I I I l I i
NOV DEC «JAN
¦tM.T
MDNITORING PERIOD '87/t
~ ARVAO * AURARIA
Figure 5-28. Winter m/p-Xylene Concentrations.
63
-------
5.6 Semi-Volatile Compounds
Semi-Volatile Compounds were collected using polyurethane foam
( PUF ) cartridges and filters. Sampling was conducted during both
the summer and winter sampling periods. However due to the
constraints of the study a selected number of PUF samples were
analyzed. The selection process involved a screening method in
which extracts of each PUF sample were exposed to ultraviolet
light. Samples containing detectable quantities of semi- volatile
organic compounds would fluoresce when subject to the ultraviolet
light. Those samples that did fluoresce were set aside for
analyses. A qualitative ranking of strong, medium, or light
fluorescence was assigned to these samples.
Of the 144 PUF samples collected during the program, only 39
samples showed any fluorescence. These samples were set aside for
individual analysis. In addition to the individual samples five
composite samples were constructed. These composites consisted of
portions of all samples from Auraria, Arvada, and NJH sample sets.
A portion of the individual samples was include in the composite
for each respective site. The composite sample, therefore,
represented an average concentration of the 6emi-volatiles
collected for the summer and winter periods. Table 5-2 contains
the values found in the composite extracts. More compounds were
found in the winter composites compared to the summer and in
general the winter puf samples were higher in concentration.
Naphthalene had the highest concentration for both the winter and
summer sampling. The remaining compounds were an order of
magnitude lower or less in concentration compared to Naphthalene.
Further analyses of the individual samples is needed before
the results can be presented in this report. It is anticipated
that the individual PUF sample concentrations may be associated
with ambient air quality conditions. This assumption is based on
the pattern other air toxic concentrations have followed and has
been reported on in previous sections of this report. A discussion
of the individuals will be presented in the final report.
For the purpose of this report, the average concentration for
the semi-volatiles., i. e. the composite sample, is summarized in
a series of figures within this section of the report. Only those
compounds in which a concentration could be detected in the
composite sample are reported in the figures. Several compounds
were not detected despite the fact that the sample was a composite
sample. The reviewer of this report is encouraged to read the
project report ( PEI, 1988 ) for a complete description of all of
the compounds that were analyzed for by the laboratory.
64
-------
Table 5-2 PUF Composite Concentrations
MONITOR BITE COMPOSITE CONC. (UG/M3)
COMPOUND ARVADA AURARIA-AM AURARIA-PM NJH
SWaS W S W SW
NAPTHALENE
9-FLUORENONE
PHENANTHRENE
ANTHRACENE
FLUORANTHENE
PYRENE
ACENAPHTHENE
ACENADTHYLENE
FLUORENE
BENZO(g,h,i)PERYLENE
INDENO(1,2,3,-cd)PYRENE
CHRYSENE
a S = SUMMER W = WINTER
0.4200.700 0.380 0.800 0.50 1.400 0.72 0.9
0.0060.010 0.006 0.012 0.009 0.012 0.007 0.012
0.0380.049 0.058 0.060 0.048 0.059 0.035 0.041
NDb ND ND 0.011 ND 0.005 ND 0.002
0.0045 0.0120.006 0.010 0.010 0.013 0.006 0.012
0.0070.015 0.007 0.01 0.009 0.010 0.006 0.025
0.0180.018 0.015 0.028 0.016 0.030 0.017 0.025
ND 0.035 ND 0.022 ND 0.055 ND ND
0.0150.022 0.015 0.025 0.016 0.035 0.017 0.025
ND 0.003 ND ND ND ND ND ND
ND 0.002 ND ND ND ND ND ND
ND 0.004 ND 0.005 ND ND ND ND
65
-------
6.0 Interpretation of the Measured Air Toxics Data Set
Introduction
The primary goal of the Denver air toxics study was to
estimate risks from ambient exposure to selected air toxics.
There are two main approaches that are commonly used in applied
studies to estimate exposure. The most common approach is to use
Gaussian dispersion modeling in conjunction with point and area
source emissions inventories. An alternative to this approach is
to interpret and extrapolate measured data at selected monitoring
sites in lieu of comprehensive modeling analyses.
There are two important elements of data interpretation that
are addressed in this chapter:
1. Does the measured air toxics data set represent air quality
that is typical of Denver conditions during the summer and
winter seasons?
2. How do Denver's air toxics concentrations compare with other
metropolitan areas typical, relatively high, or relatively
low?
The intermittent control measures in place during the winter
season of 1987-88 substantially complicate the interpretation of
the air toxics data set with respect to these questions. There was
an oxygen fuels program in effect during the second half of the
winter season, and intermittent fuel switching at the power plants
in the Denver metropolitan area throughout most of the winter
season. The influence of these control measures on concentrations
of air toxics, and the variability of meteorological conditions in
Denver during the winter ("brown cloud") season, required that a
detailed data interpretation be done in order to use the measured
data set as a basis for an exposure and risk assessment.
First, concurrent data for a broad range of variables were
collected. To support present and future interpretation of the
measured air toxic data set. Variables included those that affect
concentrations of air toxics, or potentially are correlated with
air toxics. Section 6.1 describes this effort. Section 6.2
describes how well the measured air toxics data set represents
typical Denver air quality. Section 6.3 completes the
interpretation by comparing the measured Denver air toxics data set
with concentrations measured in other metropolitan areas.
The following conclusions can be drawn from the data inter-
pretations described in this Chapter:
66
-------
Question 1: Does the measured air toxics data set represent air
quality that is typical of Denver for the summer and winter
seasons?
The analyses suggest the following:
1. Meteorological conditions during the base year appear to be
typical based on the five years of meteorological data that were
compared with the monitoring period (base year).
2. The influence of control measures appear to be evident in the
measured data set:
O Oxygen Fuels Program - CO was observed to decrease roughly
10-20 percent after normalizing for meteorological factors,
which is consistent with the expected benefits of this
program. On the other hand, the expected increase in
formaldehyde during the high oxygen fuels program was not
observed based on the measured data set? in fact even after
normalizing meteorological conditions (as described in Section
6.2) formaldehyde was observed to decrease by 30-4 0 percent
relative to the use of traditional fuels. With the possible
exception of formaldehyde, however, the oxygen fuels program
would be expected to have a relatively small influence on the
exposure assessment.
o Coal / Gas Fuels at Power Plants - Based on limited testing
during the winter of 1987-1988, the use of natural gas in lieu
of the tradition fuel (coal) at local power plants appears to
have substantially reduced concentrations of selected metals
at the downtown Auraria site, which is most affected by local
power plants, (namely the Arapahoe and Cherokee stations).
If natural gas were to be routinely used it appears that
co-control benefit, in addition to visibility benefits, would
be to reduce the risks from exposure to some toxic metals.
The affect was most pronounced during the AM conditions, when
the plumes from these power plants would be most likely to be
dispersed to ground level within the travel distances from
these plants to the Auraria monitoring site.
3. The more qualitative assessments for spatial representative-
ness suggested that the Auraria and Arvada sites could be used
Refer to Chapter 4 (Data Validation) for a description of
the correction factor applied to the formaldehyde data set. This
correction introduces uncertainty into the formaldehyde data.
2
There is uncertainty introduced to the metals data set,
however, that was caused by the failure of the 2.5um samplers to
restrict particles to 2.5 ug or less.
67
-------
to represent typical concentrations within the central and
western portions of the metropolitan area, respectively. The
Palmer site, through correlation with NJH, was hypothesized
to more accurately represent typical concentrations in the
eastern portion of the metropolitan area. The NJH site was
assumed to best represent MEI concentrations in the vicinity
of heavily traveled roadways - - which can be directly used
in the risk assessment to characterize MEI impacts from
mobile sources.
Question 2: How does Denver's air toxics concentrations compare
other metropolitan areas - - typical, relatively high, or
relatively low?
Limited comparisons suggest that Denver has risks from air
toxics that are comparable to those in other metropolitan areas
throughout the country. Refer to the following subsections for a
description of the technical analyses that were conducted to
provide a basis for these findings.
6.1 Compilation of Data Bases
Criteria pollutant, meteorological and emissions variability
data concurrent with the measured air toxics data set were
processed into a data base to provide a basis to interpret the
measured air toxics data set. Refer to Appendices E and G for
listings of these data bases, which are referred to collectively
as the supplemental data set.
6.1.1 Measured Criteria Pollutant Data
Three pollutants were selected as reference criteria pollu-
tants to aid in the interpretation of the measured air toxics
data set:
Carbon Monoxide - As a general indicator of mobile source impacts
(VOCs, semi-volatiles, aldehydes/ketones, and metals).
3
Supplemental data m this report refers to data gathered to
help support the interpretation of the air toxics data set, in-
cluding concurrent meteorological data, daily regional emissions
characteristics, and concurrent criteria pollutant data.
The term reference criteria pollutant is used throughout
this report. One or more criteria pollutants were selected to help
interpret the measured air toxics data within each pollutant class,
such as VOCs.
68
-------
PM-10 - As an indicator of particle-phase air toxics (e.g. metals
and applicable semi-volatile organics).
Ozone - As an indicator of photochemically-produced pollutants
(aldehydes/ketones).
The reference criteria pollutant data set was processed to
match the AM and PM time periods used in the air toxics monitoring
program, i.e. the hourly CO and ozone data were processed into
AM/PM data blocks for these three pollutants; the PM-10 data were
retained in their original 24-hour averaging period. All sites in
the metropolitan area with data that were concurrent with the
measured air toxics data were included . Appendix D presents the
concurrent criteria pollutant data set, along with summarized air
toxics and supplementary data.
6.1.2 Meteorological Data
The documentation of meteorological conditions is particularly
important for interpreting measured air quality data in Denver
because of the strong meteorological influences in this
metropolitan area. The meteorological influence on peak pollution
days can be substantial, especially during the winter season.
Meteorological data were sought to represent general conditions for
each daily AM/PM data block of the base period.
The main problem with documenting meteorological conditions
was retaining the essential information without developing a data
set that was too large to be effectively interpreted. The
following was done:
1. Stapleton International Airport (Stapleton) was selected
as the meteorological monitoring station to support this
data interpretation, in lieu of alternative
meteorological monitoring sites within the South Platte
valley. This was necessary because other data sets were
found to have inadequate data recovery.
5 ...
Note that CO also was measured at all air toxics sites to
help track the mobile source component.
6 Stapleton is located approximately 10km east of the central
business district. Although Stapleton is not necessarily
representative of meteorological conditions within the core
metropolitan area under all conditions, it was selected as a
general indicator of meteorological variability in lieu of more
specific data.
69
-------
2. The following meteorological parameters were extracted:
wind speed, wind direction, stability, precipitation
intensity, mixing height, temperature, relative humidity,
and visibility.
3. For wind and stability data, there are 576 possible
combinations of wind speed (6 classes), wind direction
(16 classes) and stability class (6 classes) if the data
were to be categorized into standard frequency
distributions. This volume of data would have been
unmanageable for this study. The problem was mitigated
by compressing the data into a 12-cell matrix as shown
in Appendix E. The compressed data set presents averages
of wind speed, and stability within each wind direction
quadrant for each AM/PM data block, thereby retaining
only the level of detail that actually was used during
data interpretation.
Refer to Appendix E for a complete listing of the
meteorological data set, which covers each day of the IEMP air
toxics monitoring program. Appendix F presents a comparable,
compressed meteorological data set for a five-year period, which
was used to evaluate the temporal representativeness of the data
set.
6.1.3 Emissions Variability Data
Emissions variability was documented because this factor can
substantially influence the variability in ambient concentrations.
While it is essential to document this component, it is not
feasibly to comprehensively characterize emissions variability for
this study. This term is difficult to document for air toxics
studies because releases often are from an extensive array of
sources, ranging from small and large industrial sources, to area
souirces such as residential/commercial heating, mobile sources, and
so forth. There were too many sources and source categories to
address. This problem needed to be simplified to strike a balance
between the need for emissions variability data, and the
availability of data to characterize this term.
For the Denver study, measures of emissions variability were
focused on three source categories, which were considered most
essential to data interpretation: (1) mobile sources, (2) wood
burning, and (3) power plants. Available data were coded into a
7
Two factors primarily influence variability in air quality:
(1) meteorology, and (2) emissions variability. Most of the effort
in providing supplementary data was focused on documenting these
two key factors.
70
-------
supplemental data base for these three categories to support the
interpretation of the air toxics data set.
Mobile Sources - The goal for this emissions category was to
establish metropolitan-wide relative indicators of the variability
in CO, VOCs, particulate emissions and formaldehyde released from
mobile sources. These indicators incorporate the relative
differences in emissions (on a daily basis) caused by: (1)
differences in regional vehicle miles traveled (VMTs) (Sullivan,
1988) , (2) differences in emissions per mile traveled as a function
of temperature (Stump, 1988) and (3) emissions reductions
attributed to the use of alternative fuels during the winter of
1987-1988 (EPA, 1988). When subsequently combined with daily
dispersion factors, daily emissions provides an important variable
to support regression analyses.
The equation used to represent area-wide relative mobile
source emissions is as follows:
Ex = VMTs * Et/E20c# Px 6-1
where:
Ex = Daily regional emissions indicator for pollutant
"x" (for CO, VOCs, formaldehyde, or PM-10)
VMTs = Regional vehicle miles traveled figure provided by
Colorado Department of Highways (Sullivan, 1988b).
e20c= Emissions of pollutant "x" per mile traveled at
a reference temperature of 20C.
ET = Emissions of pollutant "x" per mile traveled at the
average temperature (C) for each AM/PM data block
under review.
Px = Relative emissions per mile traveled for pollutant
"x" due to high oxygen fuels program. (Note Px= 100
for all pollutants during periods when regular fuels
were in use.
Emissions of CO and VOCs have been shown to be substantially
affected by ambient temperature (Stump, 1988). Tests conducted and
reported in the Stump reference for 20, 40 and 70°F temperatures
were used to characterize the affect of ambient temperature and the
emissions of selected pollutants. Curves were fit to CO, total
hydrocarbons (THC) and benzene, and a logarithmic curve was found
to best represent the functions. Figure 6-1 shows the curves in
relation to the three temperatures used to define the
relationships. Both predicted and observed values are presented
71
-------
in the figure and show a high correlation. Benzene emissions
represented in Figure 6-1 are in milligrams/kilometer.
XHBIIlfT TEHPERATOEE ( cr)
Figure 6.1. Mobile Source Emissions versus Temperature.
A review of available literature resulted in the following
assumptions for emissions differences when the high oxygen fuels
program was in effect:
CO - Assumed emissions from mobile source were decreased 10
percent.
VOCs - Assumed VOC emissions from mobile sources were
decreased 5 percent.
Aldehydes - Assumed emissions from mobile sources were
increased 10 percent.
Particulates - Assumed no change for mobile source emissions.
An estimate of relative emissions from mobile sources for
reference criteria pollutants CO, Ozone (for VOCs) and aldehydes
72
-------
was computed for each AM/PM data block of the air toxics monitoring
program. Refer to Appendix F for the results of this analysis.
Wood consumption - Three parameters were coded to support the
documentation of wood consumption variability : (1) days with wood
burning bans in effect, (2) Weekday/Weekend/Holiday and AM/PM
Variability, and (3) Ambient Temperature.
1. Wood Ban Days - All days with wood burning bans in
effect were identified in the supplemental data base.
A wood burning ban was shown to be in effect if Denver
(city) issued a ban for a given day.
2. Weekday/Weekend/Holiday - Weekday versus weekend/holidays
and diurnal variability in wood consumption can only be
estimated for typical weekends or weekdays, not for
specific days. A local wood survey (Colorado Department
of Health, 1988) provide a basis to estimate typical
differences. Based on these assumptions, and normalizing
to the heaviest wood consumption rate of PM
weekend/holiday, the following relative rates were
established.
Table 6-1. Relative Wood Consumption Rates
winter summer
WEEKDAY WEEKEND HOLIDAY WEEKDAY WEEKEND HOLIDAY
AM 0.12 0.18 0.18 0.00 0.00 0.00
PM 0.67 1.00 1.00 0.00 0.00 0.00
Q
Wood consumption was assumed to be significant only during
the winter season.
9
Based on a survey performed by Community Response, Inc.
(Colorado Department of Health, 1988), it was assumed that the rate
of consumption was 50 percent higher on a weekend or holiday than
during weekdays. It was also assumed based on the survey that 85
percent of the wood was consumed during the PM data block.
73
-------
Using data for woodburning bans and the relative consumption
rates shown above, a relative wood consumption indicator was
computed as follows:
where:
C = Relative wood consumption indicator
D = Relative wood consumption rate (per above table)
Y = 1 (no ban in effect), or 0.10 (ban in effect) (this
factor assumes 90 percent compliance with wood burning
restrictions (Colorado Department of Health, 1988).
3. Ambient Temperature - The average ambient temperature
shown for each daily AM/PM data block in the monitoring
period provides an indication, albeit a rough indication,
of relative wood consumption rates for specific days.
This parameter could be useful in estimating relative
demand for fuel by wood stoves, although it is not as
likely to be a good indicator of wood consumption in
fireplaces. The primary use of ambient temperature
within the supplemental data base is to provide an
additional independent variable that could be correlated
against the tracer pollutants for wood consumption, i.e.
Benzo(a)Pyrene (BaP) . Refer to Appendix G for a
summary of wood combustion variability data.
Power Plants - Power plants fuel demands are fairly constant
throughout the year in Denver. Daily fuel consumption rates were
not included in the supplemental data bases because this factor,
coupled with the lower priority for this source category in terms
of air toxics impacts. A potentially more significant factor is
alternative fuel use during the winter season of 1987-1988. The
alternation between coal and natural gas during the winter seasons
(1987-1988) provides the potential for affecting some toxic air
Potassium was also considered as a tracer for wood
combustion, but the bias of the 2.5 um samplers toward larger
particles limited the usefulness of these correlations. Refer to
Section 3 of volume one of this report for a complete discussion.
The wood consumption variability data have not been fully evaluated
in this report.
74
-------
pollutants measured in the IEMP data set, such as heavy metals.
The fuel type used at the local power plants for each day of the
winter monitoring program is, therefore, documented in Appendix G.
6.2 Representativeness of Measured Air Toxics Concentrations to
Denver Air Quality
The goal of this section is to evaluate how well the measured
air toxics data set represents seasonal average concentrations for
the metropolitan area. There are two separate issues:
(1) How well do the sampling days represent seasonal average air
quality for the Denver metropolitan area?
(2) Do the three primary air toxics monitoring sites represent
typical, relatively low or relatively high ambient exposures
for the Denver metropolitan area?
The most direct means of achieving this objective would be to
compare long-term air toxics data (such as data collected over 2-5
years from a wide range of monitoring sites) with data from the
base monitoring period for the three IEMP monitoring sites.
There are no such long-term air toxics data sets, however, to
support this level of review. Unfortunately, the only such data
sets in this metropolitan area, like most all areas, are for
criteria pollutants.
It is proposed in this report that criteria pollutants can
serve as surrogates to help infer the representativeness (in time
and space) of limited measured air toxics data sets if the fol-
lowing two conditions are met:
1. The selected criteria pollutants should be relatively well
correlated with air toxics data e.g. R >0.50.
2. It should be possible to estimate concentrations of the
selected criteria pollutants through regression analyses as
a function of meteorological parameters (such as a dispersion
term, ambient temperature, etc.) and local site charac-
teristics (e.g. local traffic density, distance to the central
business district, etc.1. Again, the goal is to establish
regressions that account for 50 percent or more of the
variance (i.e. R is >0.50).
The Auraria monitoring station is located within the
drainage flow of the Zuni and Arapahoe power plants, roughly 1 km
from Zuni and 8 km from Arapahoe station. This monitoring site was
expected to be most likely affected by fuel switching.
75
-------
The logic for these two conditions is as follows. The first
step is to show how well air toxics and selected criteria pollu-
tants track based on analyzing concurrent data collected at
monitoring sites where both are measured. If there is a high
correlation with at least one criteria pollutant, such as carbon
monoxide, then the selected criteria pollutant could be used to
help infer how well air toxics data from the more limited sites
and time periods represent more general conditions.
The benefit of meeting the second condition is that if the
criteria pollutant that will be used as a surrogate can be pre-
dicted effectively beyond the existing criteria pollutant moni-
toring network, additional resolution could be provided to help
assess the representativeness of the limited air toxics data set
in time and space.
Section 6.2.1 describes how well the Denver air toxics data
set meets these conditions, and summarizes the approach used to
infer the representativeness of the air toxics data based on using
criteria pollutants as surrogates for air toxic pollutant classes.
6.2.1 Regression Analyses Used to Support Interpretation of air
Toxics Data.
The basic approach is as follows. First, regressions were
established between a selected toxic air pollutant from each major
pollutant class (such as benzene for VOCs) and at least one
criteria pollutant (CO, PM-10 or ozone). Once these relationships
were established at the three primary monitoring sites where both
toxic and criteria pollutants were measured, they could be used as
a basis to estimate air toxic concentrations more broadly in space
and time.
The second step was to estimate concentrations of the
reference criteria pollutants (CO, PM-10. and ozone) as a function
of meteorological data and local site characteristics. If the
variability of the reference criteria pollutant could be estimated
as a function of time and local characteristics, then the
regressions established in step 1 potentially could be used to
further extrapolate measured air toxics data to the metropolitan
scale.
76
-------
Correlation of Reference Criteria Pollutants and Measured Data
for Reference Toxic Air Pollutants
Correlations were considered for the following sets of
pollutants:
Pollutant Class Reference Toxic Reference Criteria
Pollutant Pollutant
VOCs
Aldehydes/Ketones
Metals
Cadmium
Benzene
Bap*
Formaldehyde
PM-10
CO
CO
CO
* Not evaluated in this report. The number of individual BaP
samples was considered insufficient to support a correlation.
Table 6-2 summarizes the correlation between reference toxic
pollutants and reference criteria pollutants. Gas-phase toxic air
pollutants showed good correlation with criteria pollutants;
particle phase air toxics were found to be poorly correlated based
on the limited review of cadmium.
Correlation of Reference Criteria Pollutants with Meteorological
Data and Site Characteristics
Regression analyses were done to determine how well two key
reference criteria pollutants (CO and ozone) can be represented by
regional emissions data, meteorological factors, and local site
characteristics. Each are subsequently described.
Multiple regression analyses were performed for the full
criteria pollutant monitoring network, in order to establish
equations to estimate criteria pollutant concentrations. To
minimize the scatter in this approach, which could have been
adversely affected by bias as a function of wind flow quadrant and
diurnal (AM/PM) conditions, the data sets were partitioned into
eight subsets (four wind flow quadrants) and two periods (AM and
PM) . Empirically-developed regression equations were then computed
based on multiple regression analyses for each of the eight
subsets. Once regressions were established for the eight
combinations of AM/PM and wind flow quadrants, concentrations for
the reference criteria pollutants were estimated by selecting the
applicable equation for each day and diurnal period under review.
For example, if the AM period of the first day of the monitoring
period was found to have wind flow predominantly from the northern
quadrant, then the AM/North regression equation would be used to
estimate the average concentration for the reference criteria
pollutant (such as CO) as a function of the dispersion term and
77
-------
Table 6-2. summary of Regression Analyses Based on Reference
Criteria Pollutants and Reference Toxic Pollutants
Summer Winter
Pollutants
Site
R2
R
n
Coef.
R2
R n i
Coef.
CO
Form.
Auraria/am
0.91
0.96
15
12.89
0.91
0.96
40
4.10
CO
Form.
Auraria/pm
0.91
0.95
16
14 .66
0.90
0.95
38
2.16
CO
Form.
Arvada
0.88
0. 94
17
15.05
-
-
-
-
CO
Form.
NJH
0.93
0.97
16
7.68
0.88
0.94
16
2.67
PM10
Form.
Arvada
0.92
0.96
13
0.42
0.99
0.99
5
0.22
CO
Benzene
Auraria
0.68
0.82
19
15.65
CO
Benzene
NJH
0.87
0. 93
16
12.91
CO
Benzene
Arvada
0.82
0.90
17
21.24
PM10
Cadmium
Arvada
0.51
0.71
5
1 E"
CO
Cadmium
Auraria/am
0.08
0.28
42
.003
CO
Cadmium
Auraria/pm
0.26
0.51
40
. 001
CO
Cadmium
NJH
0.25
0.50
17
. 001
CO
Cadmium
Palmer
0.27
0.5222
. 009
site characteristics. All periods in the monitoring program were
then processed in a comparable manner.
To achieve this goal, multiple linear regression analyses were
done using the following equation:
CN = A + BX + CY + DZ 6-3
where:
CN = Concentration of reference criteria pollutant
(Mg/m )
78
-------
X = (Q)/ (u hmix), this is referred to as the
dispersion factor.
where:
hmix = mixing height
Q = estimated metropolitan-wide daily emission
rate (function of VMTs, temperature and fuel
type in use)
Y = site characteristic variables such as traffic density,
distance from control business district, etc.
A,B,C = Constants
The supplementary data base provided a wide range of options
to perform multiple regressions to estimate CO and ozone concen-
trations. For this report, the following variables were evaluat-
ed:
CO - Dispersion factors (including relative CO emission term for
each day of the base period), traffic density within 0.25 miles of
monitoring stations, traffic density within 4 miles of monitoring
stations, distance of each monitoring station from the central
business district, and crossvalley/downvalley position of each
monitoring station relative to the South Platte River Valley.
Ozone - Dispersion factors (including relative VOC emission tern
for each day of the base period), ambient temperature, distance
from the central business district, and crossvalley/downvalley
position of monitoring stations relative to the South Platte River
Valley.
Predictive models were developed, based on the regression
equations computed using the above variables, by focusing on the
key season that produces peak concentrations for these pollutants,
i.e. the winter season for CO and the summer season (AM periods)
for Ozone. Testing various combinations of these variables
revealed that many were too highly correlated to be included within
multiple regression analyses.
To summarize, it was found that for CO, the dispersion term
alone was reasonably well correlated with concentration across the
various wind direction quadrants and AM/PM periods that were
evaluated. By also including within the regressions analyses a
12
The mobile source indicator shown in Equation 5-1 was used
to define "Q". Note that for ozone, that the "Q" term was used to
represent relative VOC emissions.
79
-------
variable to represent the traffic density within 0.25 miles of each
monitoring station, it was found that a better fit could be
obtained.
For ozone, high correlations were found with ambient temper-
ature during the AM period, with only slight improvement in the
correlation if other variables were included. Only ambient
temperature, therefore, was used for the ozone regressions. Poor
predictability was found for ozone for the less critical PM
averaging period.
Table 6-3 summarizes for each quadrant and AM/PM data block
the equations developed by the regression analyses. On average,
the CO predictions explained roughly 60 percent of the variance,
whiles the AM ozone estimates explained over 90 percent of the
variance. All regressions were computed by forcing the intercepts
through zero. These equations were subsequently used to assess the
representativeness of the measured air toxics data, as described
in Section 6.2.2 and 6.2.3. Only relative differences in the
predicted concentrations were needed to meet these objectives.
The regression equations shown in Table 6-3, were used to
estimate average daily CO and ozone concentrations across the site
networks for each day of the base season (i.e. winter for CO and
summer (AM) for ozone). Figures 6-2 and 6-3 display the predicted
versus observed concentrations for these two reference criteria
pollutants. The R values for predicted versus observed
concentrations for CO and ozone were found to be 0.60 and 0.96,
respectively. These empirical models that were used to predict
relative CO and ozone concentrations based on meteorological data
and site characteristics (i.e. local traffic density) appear to
perform well enough to support the further use in this study of
-criteria pollutants to help infer the representativeness of
measured air toxics data.
In summary, the results of the Denver analyses showed that the
regressions between the criteria pollutants and toxic air
pollutants produced relatively high R values, i.e. generally above
0.50. Furthermore, it was found that by using the daily data on
relative mobile source emissions, the crude box model dispersion
term, and local traffic density at each CO monitoring site that we
could estimate CO fairly well, again with R values for predicted
versus observed concentration generally above 0.50. The two-way
link between CO/toxics and CO observed/CO estimated provided a
basis to extrapolate the limited air toxics data set to estimate
annual average exposures at the metropolitan scale, as described
in Sections 6.2.2 and 6.2.3.
80
-------
CO CONCENTRAT ION5
TOICISI VERGUG 08SBVB)
" ~
0 ° °
o o __ D
D mo °0 ° lw°
Figure 6-2 Predicted versus Observed CO Concentrations
(averaged across monitoring network)
OZONE CONC.
PREDICTED VWBU6 OOSBVED
O D -
ma a
j z H ooooo o m
°On 0°0 Do D
H to«h.S
1.S-I " DO _o
1 .8 ¦
0 % 0 5 °tf
DO DO,
% CZ) Q)
% ft n D
D „
(D 8 ~
~ ~ 4:
j.e - o " ^ o °
D 0 _
3.6- JD DO O ~ °
3.4 •
3.3
3
cP „
cP°Q D
Figure 6-3. Predicted versus Observed Ozone Concentrations
(averaged across monitoring network)
81
-------
Table 6-3 Equations for CO and Ozone as a functions of wind
Flow Quadrant and AM/PM Periods
Winter CO Concentrations (ppm)
Regression
Coefficients
A \M / nU Aiia/Qva«4- T5 n
AM/PM
Quadrant
R2
R n
Dispersion Coef.
Traffic Dens
AM
N
0.59
0.77
182
0.0007
2.54X10"6
AM
E
0. 66
0.81
102
0.0020
1.79xl0~6
AM
S
0.71
0.84
208
0.0024
1.92xl0~6
AM
W
0.51
0.72
76
0.0011
2.72X10"6
PM
N
0. 50
0.71
95
0.0011
2.26xl0~6
PM
E
0.56
0.75
65
0.0048
2.51X10"6
PM
S
0.47
0. 69
358
0.0058
3.08X10"6
PM
W
0.65
0.81
51
0.0113
2.4 7xl0-6
Summer (AM) Ozone Concentrations (PPMxlO-2)
Quadrant Regression
Coefficient
R2 R n
1 0.93 0.96 136 0.0573
2 0.94 0.97 140 0.0606
3 0.94 0.97 180 0.0605
4 not evaluated, only 7 days of data
6.2.2 Representativeness of Measured Air Toxics Data to
Estimate Long-term Averages at Monitoring Sites
Concentrations vary at specific locations as a function of
meteorology and emissions variability. It is desirable to
normalize the meteorological variability terms in order to more
effectively estimate the temporal representativeness of the
measured air toxics data. This step is especially important for
82
-------
evaluating this particular air toxics data set because of the
year-to-year variability in the severity of the Denver "Brown
Cloud" winter season, and the intermittent control programs in
place during the winter season of the field program. An important
question is what do the concentrations collected during the winter
of 1987-1988 represent?
Based on the empirical relationships developed in Section
6.2.1 for the seasons of the base year, it was possible to evaluate
the representativeness of the monitoring seasons in relation to
long-term averages over the five-year period, 1982-1986.
Five years of meteorological data were processed into
sequential 12-hour AM/PM data blocks to match the summer and winter
monitoring seasons of the diurnal periods used during the base year
(1987-1988). The wind direction quadrants with the highest
frequency of flow were identified for each of the days (AM and PM
periods) of the five year data set (1982-1986) . These
meteorological data were then processed through the regression
equations established for the base year, i.e. the regressions
established for CO winter and ozone summer (AM) as a function of
wind direction quadrant and AM/PM data block. Predicted concen-
trations were computed on this basis for each day of the five- year
data set to compare on a relative basis with concentrations
predicted for the base year.
It should be noted that the winter CO regression equations
require relative CO emission rates and dispersion factors. In
order to estimate the representativeness of meteorological
conditions of the base year compared to a five year meteorological
data set, the average relative CO emission rate for the winter base
period was used as a basis to estimate CO concentrations for the
five-year period. Because of limited data that were available for
the base year, as well as the five year data set, CAMP and NJH were
evaluated to estimate the representativeness of the base year to
the five year data sets. The predicted concentrations for these
two sites were similar, such that CAMP ultimately was used as the
reference site to assess the effects of meteorological variability
on seasonal CO concentrations.
Four monitoring sites (CAMP, Arvada, Carriage, and Welby) were
used to predict the average ozone concentrations across the
metropolitan area for each of the AM summertime periods of the base
year and five-year data sets. It should be noted that daily ozone
concentrations were normalized by using predicted averages for the
AM periods because of the lack of regression equations to suitably
represent ozone concentrations during PM periods.
13
These predicted concentrations should be interpreted as
relative, not absolute, estimates of the variability in concentra-
tions .
83
-------
Table 6-4 presents the predicted relative concentrations
compared to the observed average concentrations for CO and ozone
for each season during the base year and the period of 1982 through
1986. The column labeled "Relative Predicted CO" (or Ozone) can
be used to helps assess the influence of meteorological variability
on the measured data. The ratios (predicted/observed) are
presented to help infer changes in emissions over the period of
1982 through 1988. Differences in these normalized concentrations
can be inferred to be caused by trends in emissions over the six
year period.
The following was observed:
1. The predicted relative concentrations (column 1) for CO and
ozone suggest that the base year was not an anomalous year
based on meteorological factors.
2. An 18 percent drop in normalized CO (Column 2) was observed
during the base year winter season based on this approach.
The seasonal meteorological variability, as shown in the
relative predicted concentrations, does not appear to be
contributing to the drop in CO concentrations. This analysis
also is reasonably consistent with the partitioned CO data
(high oxygen fuels, regular fuels) for the base year, which
suggested a 10-20 percent drop in the normalized CO
concentrations during the high oxygen fuels program. Since
the base year contains a phase-in of high oxygen fuels, i.e.
during the first half of the winter period regular fuels were
in use, the 18 percent drop shown here for the winter of
1987-1988 seems relatively high in comparison with the base
year partitioned results. Both analyses, however, suggest a
substantial drop in CO concentrations, which can be inferred
to be caused by the oxygen fuels program.
3. The ratio of observed to predicted ozone concentrations for
the base year shows roughly a 10 percent drop relative to the
more long-term period. It is unclear if this drop is
reflecting scatter in this technique or controls on VOCs or
NOx have resulted in reduced ozone concentrations. More data
would need to be reviewed to sort out this observation. It
appears plausible, however, that the downward trend in
normalized ozone concentrations (Column 2 of Table 6-3) could
be attributed, in part, to the incorporation of more efficient
mobile source controls as the mobile source fleet is
modernized, and tighter local regulations for ozone precursors
but the differences are too small to be definitive.
84
-------
Table 6-4.
comparison of observed to Predicted Relative
Concentrations for the Base Year and Five-year Data
Set
Observed CO (ppm)
Column 1 Column 2
Predicted
Relative CO(ppm) Ratio (obs/pred)
Year
CAMP
CAMP
CAMP
1982
4.7
1.33
3.53
1983
4.5
1. 13
3.98
1984
4.8
1.11
4.32
1985
-
1.04
-
1986
-
-
-
base year
3.5
1. 09
3 .21
Observed Oz (ppm)
Year (ALL SITES)
Predicted
Relative Oz (ppm)
(ALL SITES)
Ratio (obs/pred)
(ALL SITES)
1982
1983
1984
1985
1986
base year
0.032
0. 033
0.031
0.032
0.030
0. 045
0. 046
0.046
0. 046
0.047
0.70
0.71
0. 67
0. 69
0. 63
\
6.2.3 Review of the Influence on Control Measures on the
Representativeness of Measured Concentrations
Partitioning the measured air toxics data by periods with
common control measures provides a means of evaluating the
influence of control options on air quality.
Three partitions of the measured data set were evaluated: (1)
high oxygen fuel blends versus typical fuels (2) coal versus
natural gas fuels at power plants, and (3) wood ban days versus non
wood ban days. All tracer pollutants were evaluated for each of
the three partitions. These partitions are only applicable to the
winter season, for which the data interpretation is complicated by
the above control measures.
85
-------
The following pollutants were used as tracers to help track
these three source categories:
Category
Mobile source fuel type
Power plant fuel
Tracer
CO and formaldehyde
Cd
Wood ban
Bap*
* This analysis was not included in this report because of
insufficient data.
The results of these analyses are compared in Tables 6-5 and
6-6 based on unadjusted and normalized concentrations for the
selected tracer pollutants.
A broader review for Auraria of pollutants within the aldehyde
class was done for the oxygen fuel partitions and metals within the
coal/gas partitions (unnormalized concentration data only). The
following was observed:
Aldehydes (oxygen fuels partitions)
Ratio concentration (oxygen/regular fuel)
Pollutant
AM
PM
Formaldehyde
Acetaldehvde
Acrolein
Acetone
Propionaldehyde
Crotonaldehyde
Benzaldehyde
Isovaleraldehyde
Valeraldehyde
O-Toluenaldehyde
M-Toluenaldehyde
P-Toluenaldehyde
Hexane
Dimethylbenzene
0.78 0.48
0.52 0.38
14.02 15.56
1.02 0.30
0.63 0.43
0.67 0.46
0.82 0.46
0.46 0.23
0.77 0.48
0.71 0.37
0.12 0.08
0.56 0.27
0 0.02
0 0.13
14A large increase in acrolein was observed. This comparison
did not consider normalized meteorological conditions.
86
-------
Metals (Coal/eras Partitions)
The Auraria site was expected to be most affected by the
alternative fuels at the power plants because of its location along
the drainage flow of the Arapahoe and Zuni stations. Limiting the
interpretation to the following pollutants, which are expected to
be least affected by the sampler bias for large particles (Komp,
1988), the following ratios can be compared:
Ratio
Concentration
Pollutant
AM
PM
Ni
0.63
1.82
Cd
3.18
0.85
As
4.17
0.94
V
0.85
0.82
Cr
0.63
0.75
Hg
0.70
1.20
Ag
1.21
1.24
The preliminary analyses suggest that cadmium and arsenic may
be substantially reduced, at least in the downtown area, during
the period when gas fuel was used in place of coal. As expected,
the differences were most pronounced during the AM periods when
greater dispersion rates generally occur. The relatively tall
power plant stack heights, compared to the release heights of most
air toxic substances, are expected to produce impacts on ambient
air quality close to the stack during such periods because the
plumes are more rapidly mixed toward the surface.
87
-------
Table 6-5. Concentrations Data Partitioned by Control Options:
High Oxygen Fuel Partitions
Normalized
Concentrations Units
(CO in PPM, Form.5 (ug/1113)
Regular Oxy % Dif. Regular Oxy % Dif.
Site Pollutant
Fuel
Fuel
Fuel
Fuel
Welby
CO
3.2
1.5
-31.6
0. 15
0.12
-20.6
Camp
CO
3.8
3.3
-13.7
0.29
0.33
+ 15. 1
NJH
CO
3 . 0
2.5
-15.2
0.24
0.17
-28.6
Palmer
CO
1.9
1.3
-29.1
0.15
0.17
+7.9
Auraria
CO
3.6
2.2
-39.4
0.28
0.23
-19. 1
Average
Difference
CO
i
-25.8
-9.1
Arvada
Form.
9.4
3.0
-68.6
0.32
0.16
-47.7
NJH
Form.
11.8
5.2
-56.2
0.47
0.26
-44.2
Auraria
Form.
10.9
6.4
-41.5
0.62
0.40
-35.7
Average
Difference
Form.
1
-55.4
-42 . 5
a Formaldehyde
b In the data partitions shown for the control options, the CO formaldehyde
concentrations were normalized by dividing the observed concentrations for
each site by the dispersion factor times regional VMTs applicable for the
sampling day.
The following is suggested based on the partitioned data sets:
1. CO controls in the form of high oxygen fuels appear to be
reducing concentration of CO in the range of 10-20 percent
based on the normalized data. This finding is tentative,
however, because of limitations in the accuracy of the
measured data set and the scatter in the normalization
procedure. The Palmer and CAMP sites appear to be producing
88
-------
Table 6-6. Concentrations of CO and Heavy Metals by Control
options Coal/Gas Burn Days
Normalized
Concentrations Units3
(CO in PPM, Cd & PM-10 in ug/m3)
Site Pollutant
Coal
Fuel
Gas
Fuel
% Dif.
Coal
Fuel
Gas
Fuel
% Dif.
Welby
CO
1.9
2.4
+22 . 0
0.13
0.15
+ 10.2
Camp
CO
3.4
4.2
+ 18.7
0.27
0.34
+24 .9
NJH
CO
2.6
3.1
+ 18.3
0.22
0.21
-4.4
Palmer
CO
1.3
2.3
+44.4
0.13
0.17
+ 25.9
Auraria
CO
3.2
3.8
+ 15. 6
0.27
0.29
+7.7
Average
Difference
CO
+23 . 8
+12 . 9
Arvada
Cd
0.0016
0.0029
+77.5
NJH
Cd
0.0069
0.0038
-45.3
Auraria
Cd
0.0095
0.0071
-48.4
Auraria(AM)
Cd
-64 .4
Auraria(PM)
Cd
-8.4
Palmer
Cd
0.0223
0.0188
-15. 3
Average
Difference
Cd
-17.4
Arvada PM-
•10
26.00
56. 67
+117.9
aThe Cadmium (Cd) and PM-10 (particulate concentrations) could not
be effectively normalized by the procedures used in this analysis
because of complications caused by fugitive dust and reentrainment.
89
-------
potentially anomalous data in the partitions for high oxygen
fuels and coal/gas burning at the power plants. Excluding
these two sites would result in estimating approximately a 20
percent reduction in CO emissions, compared to a 10 percent
reduction based on all sites. The unadjusted concentrations
show roughly a 30 percent decrease during the high oxygen
fuels program, which is likely due in part to the relatively
favorable dispersion conditions during the latter half of the
winter season when the oxygen fuel program was in effect.
2. The partitions revealed the unexpected finding that formalde-
hyde concentrations were substantially lower during the period
when the high oxygen fuels program was in effect. Unadjusted
concentrations were reduced by roughly 50-60 percent, while
normalized concentrations were reduced roughly 40-50 percent.
This observation should be considered tentative at this time,
pending: (1) confirmation that systematic bias was not
introduced into the aldehyde data set by revised
sampling/analysis procedures during this period, and (2)
reevaluation of emissions data to consider the feasibility for
a substantial reduction in formaldehyde concentrations for the
specific fuel types and other specific factors applicable to
the Denver study area. Such a large decrease in formaldehyde
concentrations does not appear to be consistent with expected
differences in emission rates, or the much smaller
meteorological bias suggested by the normalization procedure.
3. Difficulties in normalizing meteorological conditions during
peak pollution days resulted in the woodban control options
not being considered at this time. Future analyses based on
alternative normalization procedures may yield effective
comparisons.
6.2.4 Representativeness of Air Toxics Data Collected at
Monitoring Sites to Broader Spatial Coverage within the
Metropolitan Area
Only three primary and one secondary monitoring sites were
available to collect measured air toxics concentrations. Since
the gaol was to use this data set to estimate average metropolitan-
wide concentrations/exposures, the limited measured data needed to
be extrapolated. The following describes first a procedure that
was initially selected as a potentially more refined method to do
this extrapolation, followed by a more simplified approach that was
eventually followed.
The goal was to use a two-step process to assess the spatial
representativeness of the air toxics data: (1) define the
relationships between air toxics and supplementary data (e.g.
reference criteria pollutants, meteorological data, emissions
characteristics, and local site variables) at the three primary air
90
-------
toxics sites, and (2) use the same relationships to extrapolate
estimates of air toxic concentrations to sites where air toxics
were not measured.
Theoretically, the regression equations shown in Tables 6-5
and 6-6 could have been used to estimate CO for a wide range of
sites, which could have been used to further extrapolate air toxic
concentrations. Only marginal differences in seasonal averages
would be predicted based on the differences in local traffic
density, however, such that a qualitative approach was taken to
assess the representativeness of the results at this time. Future
studies may benefit by reexamining this approach. The following,
however, describes the more simplified approach that was followed
because of the limitations.
The Auraria (Central Section) and Arvada (Western Section)
sites provide data that appears to be suitable to represent the
typical exposures within these areas. For the Eastern Section, the
NJH site was used to represent worst-"case" exposures to highly
traveled corridors, while the Palmer site was used with empirical
functions based on the regression analyses to represent typical
concentrations within the Eastern Section. Refer to Section 6.2.2
for a description of these empirical functions.
6.3 Comparison of Concentrations With Other Metropolitan Areas
Comparisons were needed with other metropolitan areas to
provide a perspective on Denver's air toxics concentrations.
Although only limited measured data are available to make these
comparisons, nationally available data can be used to provide an
indication if some of the concentrations measured in Denver are
high in comparison to other metropolitan areas.
As a first step, summaries developed for the EPA Regulatory
Integration Division (RID) Comparative Risk Project were used to
represent distributions of annual average concentrations across the
country. Table 6-7 compares estimated 10, 50, and 90th percentile
annual average concentrations based on national data, with the
average of the summer and winter concentrations measured at each
monitoring site in Denver.
91
-------
Table 6-7. Comparison of Denver Air Toxics Concentrations with
Other Metropolitan Areas
Pollutant
Average
Denver
Across All
Sites (ug/m )
National Data
10% 50%
a
90%
Arsenic
0.0007
0.003
0.004
0. 005
BaP
0.0013
0.0005
0.0008
0.0022
Benzene
13.5
3.4
6.7
13. 3
Cadmium
0.008
0.001
0. 003
0.005
Carbon Tet.
1.7
0.3
1.8
2.8
Formaldehyde
9.8
4.9
10.9
19.1
Chromium (tot.)
0. 005
0. 006
0. 008
0. 02
Perchloroethylene
5.6
1.8
3.5
8.1
Trichloroethylene
1.3
0.5
1.0
5.8
a These concentrations are in ug/m and represent the 10th, 50th and
90th percentile values. Data compiled to support the EPA
comparative Risk Project (Sullivan, 1988).
92
-------
7.0 Health Assessments
Exposure and risk assessments were done for two scales of
analysis in this study: metropolitan and MEI scales. The
evaluation of risks at the metropolitan scale makes use of measured
data to characterize exposure, and is the primary objective of this
study. The estimation of MEI risks, a secondary objective of this
study, is based on dispersion modeling to estimate localized
incremental impacts, and measured data in the "case" of mobile
source impacts.
In this chapter, exposure is estimated using ambient
concentration as a surrogate for exposure. Since subjects pass
through numerous microenvironments throughout a typical day, using
ambient concentrations for exposure is obviously a highly
simplified technique. Exposures to pollutants with high indoor
components may be substantially underestimated on this basis. In
this sense, exposure and risks assessments can be used to isolate
the incremental impacts from ambient sources, which is the central
consideration when establishing control options to mitigate
emissions from ambient sources. The risk assessment shown in this
report, therefore, should not be inferred to represent total air
quality risks, but rather rough estimates of incremental risks from
ambient sources.
This chapter is organized to first describe the exposure
assessment (see section 7.1), followed by the risk assessment (see
section 7.2). The limitations of these assessments are described
in Section 7.3.
7.1 Exposure Assessments
7.1.1 Typical Exposure
The main goal of the exposure assessment is to convert
measured concentrations to metropolitan-scale exposures by
assigning concentrations to the total population distributed
throughout the study area. All techniques to convert measured
concentrations to estimates of metropolitan-scale exposures have
significant uncertainty. Ranges of exposures, therefore, are first
displayed, followed by specific exposure estimates.
93
-------
Ranges of Exposure
High and low ranges of exposure were estimated for the entire
population by reviewing annual average concentrations across the
three primary IEMP monitoring sites and Palmer (metals data only).
It was assumed that the maximum and minimum averages applied to the
entire population for the high and low ,,case,,s, respectively, to
attempt to bracket exposure. Table 7-1 shows the results.
15
Based on averaging summer and winter concentrations.
94
-------
Table 7-1 Range in "Cases" and Ratios of Concentrations/RFDs
POLLUTANT NAME CANCER POTENCY ORAL RFD CONCENTRATION "CASES" RATIO CONC./RFD
(M3/UG)
(UG/M3)
LOW
HIGH
LOW
HIGH
LOW
HIGH
VINYL CHLORIDE *#
0.0000026
4.5
18.5
26.8
6.5E+01
9.4E+01
4.1E+00
6.0E+00
CARBON TETRACHLORIDE
0.000015
2.6
1.6
1.8
3.2E+01
3.7E+01
6.2E-01
6.9E-01
1,1.2,2-TETRACHLOROETHANE 0.000058
2 . 2
2.9
1.7E+02
2.3E+02
BENZENE#
0.000008
2.5
14. 1
22.3
1.5E+02
2.4E+02
5.6E+00
8.9E+00
ETHYLBENZENE
349.6
5.8
7.8
1.7E-02
2.2E-02
XYLENES
6993
20.1
29. 5
2.9E-03
4.2E-03
TRICHLOROETHYLENE* #
0.0000013
26
1
1.8
1.8E+00
3.2E+00
3.8E-02
6.9E-02
TOLUENE
1049
26.3
35.3
2.5E-02
3.4E-02
STYRENE
699. 3
17.8
23.5
2.5E-02
3.4E-02
PERCHLOROETHY LENE *
0.0000005
35
4 . 5
6.8
3.0E+00
4.6E+00
1.3E-01
1.9E-01
NICKEL (SOLUABLE SALTS)
69.9
0.001
0.0039
0.0E+00
0.0E+00
1.4E-05
5.6E-05
ARSENIC#
0.0043
13
0.000415
0.000985
2.4E+00
5.7E+00
3.2E-05
7.6E-05
CADMIUM#
0.0018
0.42
0. 004
0.0189
9.7E+00
4.6E+01
9.5E-03
4.5E-02
CHROMIUM (+6)*
0.012
17.5
0.0024
0.0084
3.9E+01
1.4E+02
1.4E-04
4.8E-04
ANTIMONY
1.4
0.0004
0.0036
0.0E+00
0.0E+00
2.9E-04
2.6E-03
FORMALDEHYDE*#
0.000013
12.3
3.1
3.9
5.5E+01
6.9E+01
2.5E-01
3.2E-01
BAP
0.003
0. 001
0.002
4.1E+00
8.1E+00
TOTALS 532 878 11 16
* = DEFAULT CANCER POTENCY (PIPQUIC)
# = DEFAULT RFD (PIPQUIC)
-------
Specific Exposure Estimates
The low and high concentrations (and risks) shown in Table
7-1 were based directly on the measured data from the four IEMP
monitoring sites. ,fiThe highest and lowest averages across the
monitoring sites were matched with the total metropolitan
population to estimate total "case"s A "best" estimate is now
provided that draws on the data interpretation shown in Chapter 5.
The intent of this estimate is to provide expected values for each
pollutant within the low/high range. The following steps were
taken to estimate the specific exposures:
1. Average concentrations data from Arvada and Auraria were
matched with the total population from the Western and Central
Sections of the metropolitan area, respectively, to estimate
typical exposures in these areas.
2. The regressions established and described in Chapter 5 were
used as a basis to factor measured air toxics data from NJH
to represent typical exposures within the Eastern Section of
the metropolitan area. This step was necessary because there
were two objectives to be met within the Eastern Section: (1)
to estimate typical concentrations within this area, and (2)
to estimate concentrations applicable to the most exposed
individuals (MEI) from mobile source impacts. Concentrations
for benzene and other pollutants that were dominated by mobile
source impacts in the Eastern Section were estimated by using
the ratio of (Palmer CO concentrations / NJH CO
concentrations). The ratio, which was observed to be 0.56,
could be computed for the winter period. A simplifying
assumption was, therefore, made that it also could be applied
to the summer period. This simplification does not consider
seasonal differences in the ratios of CO to benzene, xylene,
etc. High correlations were observed between CO and benzene,
and CO and formaldehyde, which suggest that meteorological
influences generally act to proportionately affect mobile
source dominated pollutants. However, the seasonal influence
of woodburning on the ratios of (CO concentrations / specific
VOC concentrations) were not considered and likely are
significant. The following pollutants were factored on this
basis: benzene, ethylbenzene, xylenes, and toluene.
Formaldehyde was not factored because of the influence of
photochemistry in addition to transport and dispersion
16The highest and lowest average concentrations are based on
averaging across the summer and winter seasons at each site.
17
Note that the summertime VOC values were used exclusively
to estimate VOC exposures because of data quality problems with the
VOC winter data set.
96
-------
influences. The absolute value of formaldehyde likely would
have been underestimated for the Eastern Section formaldehyde
concentrations at NJH were multiplied by the factor 0.56 to
represent the entire Eastern Section.
3. BaP data from NJH were directly used to represent the Eastern
Section.
4. Metals data from Palmer were directly used to represent
average exposures in the Eastern Section.
Table 7-2 presents the average concentrations (and risks)
based on the specific exposure estimates. The average
concentrations for each of the three sections of the metropolitan
area were matched with the applicable population, rather than
simply using the total population as was done in Table 7-1. The
population totals that were used in to estimate "cases" are as
follows:
Section
Eastern
Central
Western
Total
Population
370,013
556,783
425,655
1,352,451
7.1.2 MEI Exposures
Industrial Source Complex Model (Long-Term Mode)—(ISCLT)
(Supplement A) was used in conjunction with emissions data to
estimate localized MEI concentrations for selected key sources.
Industrial sources and area sources were considered.
Industrial Sources
The Colorado Department of Health emissions inventory was
used to select a geographic area for MEI review. A cluster of nine
industrial facilities were identified in the Commerce City area,
which appeared to have the greatest potential for relatively high
MEI impacts. As a first step, the available VOC data for these
facilities was modeled using ISCLT, simplifying the facilities to
1-2 sets of release points. While relatively high VOC totals were
modeled for portions of Commerce City, further resolution was
needed in the emissions data to support further review within the
risk assessment.
97
-------
A follow-up modeling analysis was performed based on review
of more specific data from the Colorado Department of Health
emissions inventory. Benzene emissions in the Commerce City area
were found to produce the highest release of potent carcinogens.
Four industrial facilities that were found to be the highest
emitters of benzene were modeled in one composite run to estimate
maximum annual average ambient concentrations. The maximum
incremental MEI concentrations was modeled to be approximately 49
ug/m near the property boundary of one of the facilities.
If the average benzene concentration from Auraria (which is
used to represent the Central Section, in which Commerce City is
located), is added to the incremental industrial contribution, a
maximum annual average of approximately 65 ug/m is computed. This
estimate appears to be substantially higher than what would be
expected for the ambient air in the vicinity of most industrial
complexes. Further confirmation of the benzene emissions data is
needed. Additionally, more detailed modeling analysis, and
possibly some limited ambient monitoring could be considered to
help clarify the magnitude of these impacts.
Area Sources
Area sources are generally reviewed in most studies to
estimate their contributions to average exposures throughout a
study area. For the Denver study, however, an effort was made to
identify maximum annual average concentrations that could be caused
by two key area source categories - - mobile sources and wood
burning.
The following was done:
Mobile Sources - The NJH monitoring site was identified as a
likely MEI location for high mobile source
impacts. NJH is located near a heavily travelled
intersection (Colorado Boulevard and Colfax
Avenue) . The measured data from this site,
rather than modeling, was considered more
applicable to the MEI review for this source
category. The NJH data were, therefore,
directly used to estimate MEI concentrations due
to traffic exposures. Table 7-3 presents these
results.
Woodburninq - A neighborhood with a high density of woodburning
devices (fireplaces and particularly wood stoves)
was identified as an example of high localized
woodburning impacts. The goal was to assess
neighborhood scale averages concentrations in
such a neighborhood based on dispersion modeling.
The neighborhood in which the Palmer School is
located was found to contain a relatively large
98
-------
number of woodburning devices based on the site
reviews performed during the design stages of the
monitoring program. It also is located in an
area shown to be a relatively high woodburning
area based on the Community Response, Inc.
(Colorado Department of Health, 1988) survey.
This area was, therefore, selected to represent
high woodburning areas, although it is not
necessarily representative of the highest
impacted area in Denver.
99
-------
Table 7-2 Risk Assessment based on Interpreted Air Quality Data (Sheet 1 of 2)
POLLUTANT NAME
CONCENTRATION
"CASES
ii
east
central
west
east central
west
VINYL CHLORIDE*#
2.7E+01
2.7E+01
1.9E+01
2.5E+01
3.9E+01
2.0E+01
CARBON TETRACHLORIDE
1.7E+00
1.6E+00
1.8E+00
9.4E+00
1.3E+01
1.1E+01
1,1,2,2-TETRACHLOROETHANE
2.2E+00
2.2E+00
2.9E+00
4.7E+01
7.1E+01
7.2E+01
BENZENE#
1.2E+01
1.4E+01
1.5E+01
3.7E+01
6.3E+01
5.1E+01
ETHYLBENZENE
4 . 4E+00
6 -0E+00
5.8E+00
XYLENES
1.7E+01
2.3E+01
2.0E+01
TRICHLOROETHYLENE*#
1.0E+00
1.0E+00
1.8E+00
4.8E-01
7.2E-01
1.0E+00
TOLUENE
2.0E+01
2.8E+01
2.6E+01
STYRENE
2.4E+01
1.8E+01
1.9E+01
PERCHLOROETHYLENE*
5.4E+00
4.5E+00
6.8E+00
1. 0E+00
1.3E+00
1.4E+00
NICKEL (SOLUABLE SALTS)
1.0E-03
3.9E-03
1.0E-03
ARSENIC#
4.0E-04
1.0E-03
9.0E-04
6.4E-01
2.4E+00
1.6E+00
CADMIUM#
1.2E-02
6.8E-03
1.9E-02
7.9E+00
6.8E+00
1.4E+01
CHROMIUM
2.4E-03
8.4E-03
3.5E-03
1.1E+01
5.6E+01
1.8E+01
ANTIMONY
9.0E-04
3.6E-03
4.0E-04
FORMALDEHYDE*#
3.9E+00
3.9E+00
3.1E+00
1.9E+01
2.8E+01
1.7E+01
BaP
1.0E-03
2.0E-03
1.0E-03
1.1E+00
3 . 3E+00
1.3E+00
TOTALS
* = DEFAULT CANCER POTENCY (PIPQUIC)
# = DEFAULT RFD (PIPQUIC)
Totals 160 285
grand total =
208
653
-------
Table 7-2 Risk Assessment based on Interpreted Air Quality Data (Sheet 2 of 2)
POLLUTANT NAME
VINYL CHLORIDE*#
TETRACHLORIDE
1,1,2,2-TETRACHL0R0ETHANE
BENZENE#
ETHYLBENZENE
XYLENES
TRICHLOROETHYLENE* #
TOLUENE
STYRENE
PERCH LOROETHYLENE *
NICKEL (SOLUABLE SALTS)
ARSENIC#
CADMIUM#
CHROMIUM (+6)*
ANTIMONY
FORMALDEHYDE*#
BaP
east
2.7E+01 2
1.7E+00 1
2.2E+00 2
1.2E+01 1
4.4E+00 6
1.7E+01 2
1.0E+00 1
2.0E+01 2
2.4E+01 1
5.4E+00 4
1.0E-03 3
4.0E-04 1
1.2E-02 6
2.4E-03 8
9.0E-04 3
3.9E+00 3
1.0E-03 2
CONCENTRATION
central west
.7E+01 1.9E+01
.6E+00 1.8E+00
.2E+00 2.9E+00
.4E+01 1.5E+01
.0E+00 5.8E+00
.3E+01 2.0E+01
.0E+00 1.8E+00
.8E+01 2.6E+01
.8E+01 1.9E+01
.5E+00 6.8E+00
.9E—03 1.0E-03
.0E-03 9.0E-04
.8E-03 1.9E-02
.4E-03 3.5E-03
.6E-03 4.0E-04
.9E+00 3.1E+00
.0E-03 1.0E-03
RATIO CONC./RF
east central west
5.9E+00 6.0E+00 4.1E+00
6.5E-01 6.2E-01 6.9E-01
5.0E+00
1.2E-02
4E-03
8E-02
9E-02
4E-02
5E-01
4E-05
1E-05
8E-02
4E-04
4E-04
2
3 ,
1.
3
1,
1,
3 ,
2,
1.
6,
3
3
2
2
3.1E-01
5.6E+00
1.7E-02
2E-03
8E-02
7E-02
5E-02
1.3E-01
5.6E-05
7E-05
6E-02
8E-04
6E-03
2E-01
6.0E+00
1.7E-02
2.9E-03
6.9E-01
2.5E-02
2.7E-02
1.9E-01
1.4E-05
6.9E-05
4.5E-02
2.0E-04
2.9E-04
2.5E-01
TOTALS
Totals 12
12
12
* = DEFAULT CANCER POTENCY (PIPQUIC)
# = DEFAULT RFD (PIPQUIC)
-------
In order to represent woodburning emissions as a
category, Polycyclic Organic Matter (POM) were
modeled. An emission factor of 34 lbs/ton was used
for the percentage of woodstoves, and 23 lbs/ton was
used to represent fireplace emissions. The split
between woodstoves and fireplaces was based on the
survey (Colorado Department of Health, 1988). The
emission factors were based on personal
communication (Sullivan, 1988).
The percentages of housing units with combustion
devices and the specific housing density for this
neighborhood (based on site inspections) were used
to estimate the emission rates per home and per
block (for adjacent blocks).
Receptors within one central block were selected for
modeling purposes, with emissions from the eight
adjacent blocks modeled to help assess the impacts
from adjacent blocks on the selected receptors.
A total of 24 volume sources were used to represent
the residences within the selected block. Eight
areas sources were modeled to represent impacts from
adjacent blocks. ISCLT was used for this modeling
analysis.
The maximum POM concentration (annualized) was
estimated to be approximately 3.5 ug/m . Since
regional measured data are not available for POMs,
this estimate likely is biased low on this basis.
When interpreting the MEI results for this category,
a value of 5-10 ug/m could be used as an intuitive
estimate of ME I impacts including a regional POMs
component. There is insufficient data, however, to
definitively address the regional component at this
time.
7.2 Risk Assessment
7.2.1 Cancer "Cases" Over Metropolitan Area
Risk assessments were done for cancer and noncancer health
effects. Health data were extracted from the Integrated Risk
Information System (IRIS), including cancer potency, strength of
evidence, and reference doses (as available).
It is mathematically straightforward to estimate incremental
cancer incidence from 70-year exposures to ambient concentrations
of pollutants considered in this study:
102
-------
n m
C = E(ZPTn(Ps* ms) / [ 7-1 ]
1 1
where:
c = incremental cancer !lcase"s over 70 per year exposure from
pollutants considered in this study
n = number of pollutants
m = number of sections in metropolitan area (m = 3)
Hs = average concentration (fig/m ) for West, Central or East
section of Metro area
Ps = population for West, Central or East section of
metropolitan area
PTn = cancer potency for pollutant "n" (m /nq)
The ranges in concentrations shown in Tables 7-1 and 7-2 were
applied to the above equation to estimate cancer incidence over a
7 0 year exposure period. The results are shown in the table.
The range in l,case"s was approximately 550 - 950; the specific
exposure estimate showed roughly 700 "case,,s. Using the specific
exposure estimates as the basis to estimate the percentage of the
"case"s (as of among the pollutants studied shows the following:
Pollutant Percent of Total "case"s
Benzene/12DCE
23
Tetrachloroethane
29
Formaldehyde
10
Chromium (+6)
13
Vinyl chloride
13
Carbon tetrachloride
5
Cadmium
4
Other pollutants
3
total =
100
It is possible that some of these pollutants are conserva-
tively represented because of bias in the measured data set and/or
conservative assumptions in speciation. Conversely, the risk
assessment only evaluated the incremental risks from a relatively
small subset of the actual pollutant mix actually present in the
Denver metropolitan area.
7.2.2 Nonc&ncer Health Risks
The primary emphasis for noncancer was evaluating whether
estimated annual average concentrations exceeded thresholds for the
reference doses (RFDs) as previously shown in Table 7-3. Table 7-
3 shows the ratio of estimated annual average concentrations
103
-------
Table 7-3
Summary of the Results of the MEI Risk Analyses
Case
Industrial
Commerce City
Pollutants
Benzene
Estimate of
Annual Avg
Concentration
(ug/m3)
65
Individual Risk
over 70 years
5 x 10~4
Cone./RFD
26
Woodburning
(Palmer area)
Mobile Sources
(NJH)
POMs
Benzene
Ethylbenzene
Xylenes
Toluene
Formaldehyde
3.5
22 .3
7.8
29 . 5
35.3
3.9
4 X 10
-5
2 X 10
-4
5 x 10
-5
9
0. 02
0.004
0. 03
0.9
Note: Potency for POMs (wood burning category) was extracted from
PIPQUIC. The value is 1 x 10-5. See Table 6-1 for other potency
values.
divided by RFD threshold concentrations. At this time, oral RFDs
were used as surrogates because IRIS does not contain inhalation
RFDs for these pollutants.
Interpretation of RFD comparisons - There were only four
pollutants that were found to approach or exceed the RFDs shown in
Tables 7-1 and 7-2, i.e. vinyl chloride, 1,1,2,2 tetrachloroethane,
benzene, and formaldehyde. Benzene and vinyl chloride exceeded the
RFDs by a factor of 5-10. The measured concentrations for both of
these pollutants, however, contain a relatively high degree of
uncertainty, as noted previously. Furthermore, the adequacy of
using oral RFDs in lieu of available inhalation RFDs is another
area of uncertainty that also should be considered when
interpreting these results.
7.2.3 MEI Risk
The same health data as shown in Sections 7.2.1 and 7.2.2 were
used for estimating MEI risks. Table 7-3 summarizes the results,
including estimates of MEI cancer risks, and ratios of maximum
annual average concentrations to RFDs.
104
-------
7.3 Limitations of Ambient-Based Exposure/Risk Assessments
The exposure and risk assessments described in this report need
to be placed in context. How reasonable are the ambient-based
estimates of exposure when subjects continuously move through a
wide range of microenvironments, ranging from ambient to
residential, commercial, workplace and in-transit environments?
How well do ambient-based surrogates represent these exposures?
Since mobile source impacts were expected to dominate air
toxics impacts in this metropolitan area, CO would be a reasonable
reference pollutant to evaluate ambient versus actual exposures.
It is fortunate for this study that a detailed study of personal
CO exposures was conducted in Denver in 1984 (Johnson, 1984).
Versar 1986 provides a summary of the applicability of this CO
study to the exposure assessment for the Denver IEMP air toxics
monitoring program. The conclusion was that ambient concentrations
provide a reasonable first approximation of the distribution of CO
exposures. The highest exposures (e.g. >75th percentile), however,
that were hypothesized to be associated with events such as
exposure to stalled traffic and traffic garages probably are not
well represented by ambient concentrations. Considering the
importance of mobile sources to CO concentrations and many air
toxics under review in the Denver IEMP air toxics monitoring
program it appears reasonable, as a first approximation, that
similar conclusions would apply to many of the toxic pollutants
under review in this report.
105
-------
REFERENCES
Colorado Department of Health, 1988; Denver Metro Woodburnina
Survey; Prepared by Community Response, Inc., Denver, Colorado,
June.
Heisler, S.L.; Henry, R.C.; Watson,J.G.; Hidy,G.M., 1980; The 1978
Denver Winter Haze Study; Document No. P-5417-1; Environmental
Research and Technology, Inc.: Westlake Village, Ca. March 1980.
Johnson, T. , 1984; Study of Personal Exposure to Carbon Monoxide
in Denver. Colorado. EPA-600/4-84-014, U.S. EPA Environmental
Monitoring Systems Laboratory, Research Triangle Park, NC, 1984.
Komp,M.J., Svoboda,L.; Frey,S.J. 1988; Air Toxics Monitoring Plan
for The Denver Metro - Area Integrated Environmental Management
Project; EPA Region VIII/Environmental Services Division
Document:Denver, Co., March 1987.
Komp, M.J. 1988; Personal Communication with Gordon Pierce,
Colorado Department of Health, November 1988.
Lewis, C.W.; Baumgardner, R.E.; Stevens, R.K.; Russwurm, G.M. ,
198 6; Receptor Modeling Study of Denver Winter
Haze:Environ.Sci.Technol.,Vol. 20,No.11,1986,1126-1136.
PEI Associates, Inc. 1987a; Phase I Denver IEMP — Project Report.
Vol I & II, EPA Contract No. 68-02-0390 Work Assignment #67 PN
3655-67
PEI Associates, Inc., 1987b; Phase I Denver IEMP Quality Assurance
Project Plan fOAPP) Vol I & II Contract No. 68-02-0390, Work
Assignment #67, PN 3655-67
PEI Associates, Inc., 1988; Phase II Denver IEMP — Project Report
Vol. I & II. EPA Contract No. 68-02-4394 Work Assignment #10 PN
3759-10
Russell, P. A., Ed. 1976; Denver Air Pollution Study: Proceedings
of a Symposium; U.S. Environmental Protection Agency: Research
Triangle Park, NC., Feb. 1977; Vols 1S.2, EPA-600/9-77-001.
Singh, H. B., L. J. Salas, A. Smith, and H. Shigeishi, 1979;
Atmospheric Measurements of Selected Toxic Organic Chemicals.
Interim Report. Grant 805990, SRI Project 774, SRI International,
Menlo Park, Calif. EPA-600/3-8-072
Stump, Fred et.al., 1988; The Influence of Ambient Temperature on
Tailpipe Emissions from Late Model Light-Duty Gasoline Motor
Vehicles. US EPA, Atmospheric Sciences research Laboratory
Technical Paper Presented to U.S. EPA Mobile Source Laboratory Ann
Arbor, Michigan, RTP, NC, 1988
-------
Sullivan, D., 1988a; Memorandum to A. Koines for "Comparative Risk
Project", U.S. EPA Office of Policy and Program Evaluation,
February 3, 1988
Sullivan, D., 1988b; Personal Communication with Craig Koralek,
Versar, May 26, 1988.
Sullivan, 1988c; Personal Communication with Jim DiLeo of Colorado
Department of Health, August 1988.
Sullivan, D., 1987; Memorandum to IEMP Project
U.S. Environmental Protection Agency, 1980; Validation of Air
Monitoring Data; 1980; EPA 600/4-80-030; June 1980, RTP, NC.
U.S. Environmental Protection Agency, 1983; Volatile Organic
Chemicals in the Atmosphere: An Assessment of Available Data. U.S.
EPA/RTP, EPA-600/3/83-027(A), April 1983
U.S. Environmental Protection Agency, 1988; Guidance on Estimating
Motor Vehicle Emission Reductions from the Use of Alternative Fuels
and Fuel Blends. EPA-AA-TSS-PA-87-4, Office of Mobile Sources, Ann
Arbor, MI, January 1988.
Versar Inc., 1987; Appendix A Draft Monitoring Plan for IEMP, U.S.
EPA Region VIII/ESD Document
Zimmer, R; 1988a; Conversation with William Jesse, PEI, Cinn.,
Ohio; March 1988
Zimmer, R; 1988b; Conversation with William Jesse, PEI, Cinn. Ohio;
August 198 8
-------